My project started from a reaction previously developed in our group. We wondered whether the same concept could be extended to cyclic ketones, using an amine to form an imine/enamine intermediate before promoting C–C bond cleavage and ultimately achieving skeletal editing of the ring. I therefore began by reacting a cyclic ketone with a controlled amount of amine under our proposed conditions. However, the reaction did not give the product predicted by our proposed mechanism. Instead, GC–MS showed the formation of a different major product. From the mass spectrum, I proposed a possible structure and then searched for it on Reaxys.
Interestingly, the Reaxys search led me to two papers reporting the formation of the same phthalimide-type product that I had observed experimentally. One approach used a copper-based catalyst, while another achieved the transformation under metal-free catalytic conditions, both giving the product in high yield. I therefore looked carefully at the mechanisms proposed in these two papers. Although the reaction conditions were different, their proposed pathways were remarkably similar. In both cases, oxidation occurred on the carbon framework of the cyclic ketone, introducing additional carbonyl functionality through successive oxidation steps. Further oxidation and C–C bond cleavage, accompanied by loss of CO₂, ultimately generated a carboxylic anhydride-type intermediate, which could then react with the amine to give the phthalimide-type product I had observed.
This gave me a possible explanation for why our reaction was following the “wrong” pathway. Under our initial conditions, oxygen was present from the beginning, so oxidation of the cyclic ketone could potentially occur before sufficient formation of the imine/enamine intermediate required for our proposed skeletal-editing pathway. Rather than simply changing catalysts or screening more conditions, I decided to test this idea directly by separating these two processes. I first ran the reaction at a lower temperature under nitrogen, without oxygen, and left it overnight to favour condensation between the ketone and amine. Analysis of the reaction mixture confirmed that the imine had formed. I could then increase the temperature and switch the atmosphere from nitrogen to oxygen, allowing the oxidative stage to begin only after formation of the desired amine-derived intermediate.
The first step of our proposed mechanism therefore appeared to be relatively straightforward: formation of the imine/enamine intermediate could be achieved successfully. The next challenge was much more difficult. In our proposed mechanism, oxygen would oxidatively cleave the C=C bond of the enamine, opening the cyclic framework and generating the key ring-opened intermediate. There was good reason for us to consider this feasible, as previous research from our group had demonstrated that molecular oxygen could be used as a green oxidant to cleave related enamine double bonds. However, when I introduced oxygen after forming the imine/enamine, I could not observe the expected ring-opened product. One possibility was that this particular enamine was substantially more resistant to oxidative cleavage because the reactive alkene remained constrained within a cyclic system and was influenced by the neighbouring aromatic framework.
At this point, the direction of my project changed. Rather than focusing on the complete skeletal-editing sequence at once, I began concentrating on a much more specific problem: how could I cleave this particular enamine C=C bond? During my literature search, I found a report describing photocatalytic conditions capable of oxidising and cleaving the type of bond I was interested in. I therefore adapted those conditions and applied them to my system, hoping that photocatalysis could achieve the cleavage that molecular oxygen alone had failed to promote. However, the result was again unexpected. Instead of observing the desired ring-opened intermediate, I obtained the same phthalimide-type product seen in my earlier experiments, while the imine intermediate was no longer present. Although this experiment did not deliver the desired transformation, it provided another important constraint on the reaction design: simply introducing a stronger oxidative pathway was not sufficient to achieve selective cleavage, because competing oxidation could still divert the system towards phthalimide formation.