Saturated N-heterocycles are ubiquitous in nature in the form of alkaloid natural products and are present in a multitude of biologically active molecules used in agrochemistry and medicinal chemistry. For example, >80% of all drugs approved by the US Food and Drug Administration (FDA) in the past 10 years incorporate at least one N-heterocycle. Of these, the most common saturated examples are six-membered rings, reflecting the wide availability of piperidines and piperazines. By contrast, azetidines, the four-membered homologues of piperidines, occupy only a small fraction of chemical space, rank 26th on the list and account for only six approved drugs (see Fig. 1a for selected examples). As a result of their intrinsic small size and rigidity, azetidines present an atom-efficient three-dimensional scaffold with defined vectors to present pendant functionality. Incorporating azetidines into bioactive molecules has been shown to impart myriad desirable properties, including improved stability to oxidative metabolism, improved solubility and structural rigidity while simultaneously increasing the fraction of Csp centres. Despite these attributes, azetidines remain underutilized scaffolds in drug design, primarily due to the limitations of established synthetic methods. For example, azetidines commonly used in drug design generally feature substituents predominantly at the 3-position, while other positions -- and di-substituted variants -- are rarely represented; this is due to the scarcity of convenient methods to access more complex substitution patterns. Recently, the application of modern synthetic methods to this task has resulted in improved syntheses, including processes based on strain release of aza-bicyclobutanes, C-H amination, photochemical cyclizations and hydrogen atom transfer catalysis. Although these developments have partly alleviated this challenge, they still suffer from a principal limitation of earlier methods: the stringent requirement for complex, prefunctionalized substrates.

An appealing reaction to prepare azetidines would use electronically diverse feedstock olefins together with modular, readily synthesizable imines and directly couple them together via an intermolecular [2 + 2] cyclization; this general transformation, when achieved using light, is known as the aza Paternò-Büchi reaction (Fig. 1b). The parent transformation -- the Paternò-Büchi reaction -- refers to the [2 + 2] photocycloadditions of excited carbonyl compounds and alkenes and is a well-established transformation. The aza variant represents a direct, atom-economical and modular synthesis of azetidines from two equally complex coupling partners. However, the aza Paternò-Büchi reaction remains underdeveloped. This is due to the stringent requirements placed on the imine reaction component, which upon triplet sensitization must circumvent radiationless decay back to the ground-state (I → II; Fig. 1c), be resistant to photoreduction, and avoid fragmentation (I → III). The advancements in intermolecular aza Paternò-Büchi methodologies so far have relied on engineered substrates to prevent decay to the ground state, typically achieved by tethering the imine within a ring to prevent rotation around the carbon-nitrogen π-bond (Fig. 1d). An alternative strategy to skirt this challenge is to induce a formal aza Paternò-Büchi reaction by essentially reversing the reactivity and using activated alkenes that are excited in preference to the imine. A notable recent example of this strategy from the Schindler laboratory utilizes acyclic oximes as imine equivalents, combined with activated alkenes, enabling the matching of frontier molecular orbital energies between the two reaction components to achieve an efficient process (Fig. 1e). There are a small number of less general solutions, including matching substrates for exciplex formation with singlet excitation, and the use of copper catalysis to activate bicyclic alkenes in combination with ultraviolet light. Collectively, these strategies have enabled efficient reactions with broad scope; however, the need to engineer the imine structure, use imine equivalents or employ activated alkenes compromises structural diversity and thus undermines one of the major advantages of an idealized aza Paternò-Büchi reaction.

As a simple and direct aza Paternò-Büchi reaction remains elusive, we sought to address this by designing acyclic imines that fulfil several key requirements. First, these components must have accessible excited states through photochemical or physical means. Second, the excited imines must be able to engage with the olefin component in a productive regioselective manner, favouring this pathway over the aforementioned deleterious alternatives. Finally, it is imperative that the modular, acyclic imines are readily synthesizable from commercial substrates in no more than one step. To achieve these specifications we targeted sulfonyl imines, primarily due to their known photophysics, and triplet excited states accessible through triplet energy transfer catalysis. In selecting sulfonyl imines as our substrates of choice we were conscious of the rich chemistry that has been developed exploiting fragmentation of the derived triplets (that is, I → III; Fig. 1c), leading to many productive addition processes. Nevertheless, we reasoned that the tunability of the sulfonyl substituent should enable effective partitioning between the competing pathways and establishment of a productive catalytic manifold. To implement this reaction design, we propose that controlling the electronic properties of the triplet imine can influence the fragmentation barrier; by minimizing this deleterious pathway, we aim to increase the likelihood of productive catalysis. A recent report has used sulfonyl imines in combination with excitable alkenes to achieve aza Paternò-Büchi products.

Here, we show that acyclic sulfamoyl fluoride-substituted aryl imines deliver reactive triplet intermediates that engage as substrates in intermolecular aza Paternò-Büchi reactions using energy-transfer catalysis and visible light. A broad range of alkenes are used as reaction partners and provide substituted azetidines in high yields. We demonstrate the utility of this synthetic method by further manipulation of the N-sulfamoyl fluoride group present in the azetidine products, which can either be removed to provide the N-H heterocycles or be readily converted into a sulfamide unit.