Deoxygenative Single‐Carbon Atom Transfer: Accessing Strained Carbocycles From Alkenes and Aldehydes With Designer Dual‐Functional Reagent

ABSTRACT

Single‐carbon atom transfer is an attractive approach to increasing molecular complexity through the concurrent formation of four new covalent bonds. Yet, the success of these methods depends largely on atomic carbon reagents that harness carbene reactivity, thereby constraining the repertoire of transformations attainable within this regime. Here, we report a deoxygenative single‐carbon atom transfer strategy that directly couples alkenes and aldehydes to access substituted alkylidenecyclopropanes. The key enabling development is the identification of a novel iodomethylphosphonium reagent that orchestrates photocatalytic atom‐transfer radical addition and cyclizative Wittig olefination at a single carbon center. Mechanistic studies reveal that the electronically tailored phosphonium motif endows the key radical intermediate with the desired polarity for carbon delivery and, at the same time, enhances the accessibility of phosphorus ylides en route to both ring closure and Wittig reaction. This operationally simple method provides a modular entry from two abundant starting materials to highly strained small carbocycles featuring four newly formed bonds.

We reported a deoxygenative single‐carbon atom transfer strategy that directly couples alkenes and aldehydes to access strained small carbocycles. This overall transformation is enabled by a designer iodomethylphosphonium reagent that unifies atom‐transfer radical addition and a cyclizative Wittig reaction at a single carbon center.

1. Introduction

Single‐carbon atom transfer has recently emerged as a valuable strategy for building molecular complexity and diversity from the simplest atomic carbon, owing to its potential to forge four new covalent bonds [1, 2, 3]. While the direct use of naked zero‐valent carbon is limited by its intrinsic instability and the dearth of practical generation methods [4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17], recent progress has been marked by the invention of stable surrogates that orchestrate multiple bond‐forming processes and a single‐carbon atom delivery (Scheme 1A) [18, 19]. Among the seminal examples, Tobisu and coworkers repurposed N‐heterocyclic carbenes as C(sp³)─atom equivalents for the synthesis of γ‐lactams bearing four emergent σ‐linkages [20]. Beyond the realm of C(sp³)─atom transfer, Hansmann reported a diazophosphorus ylide that converts aldehydes into either dimeric allenes or terminal alkynes [21], reminiscent of the Seyferth–Gilbert homologation [22]. In parallel, the Glorius group has shown that a certain diazo‐based reagent can transform indoles into oxime ester‐embedded quinolines, which serve as a branching point for divergent skeletal editing [23]. More recently, this approach has further evolved to explore three‐dimensional product space, and in some cases, for introduction of stereogenic centers as demonstrated in notable contributions by Tobisu [24, 25], Hansmann [26], Suero [27], and others [28].

SCHEME 1.

Deoxygenative single‐carbon atom transfer.

Despite these remarkable advances, the vast majority of current examples still rely on the reactivity profiles of carbene‐ or N₂‐liberating precursors, such as diazo compounds and isocyanides. In addition to safety concerns, this reliance constrains the accessible reaction space in this nascent field. Accordingly, to further expand both the repertoire and practical utility of single‐carbon atom transfer, the development of mechanistically and functionally distinct classes of reagents is highly desirable. In this regard, we envisioned that halomethylphosphonium species (1) could provide a complementary platform for deoxygenative single‐carbon atom transfer (Scheme 1B). The guiding principle lies in the recognition that the halogen substituent can serve as a functional handle for carbon‐atom incorporation into alkenes via atom transfer radical addition (ATRA)‐induced cyclopropanation [29, 30, 31, 32, 33, 34], whereas the phosphonium moiety, upon suitable deprotonation, can participate in Wittig‐type olefination to remove atomic oxygen from aldehydes [35, 36]. By uniting these two distinct reactivities within a single‐carbon site, unprecedented coupling modes between alkenes and aldehydes could be established.

Here, we present a designer dual‐functional reagent that realizes the deoxygenative single‐carbon atom transfer, enabling the facile construction of alkylidenecyclopropanes—a class of highly strained small‐ring systems with significant value as versatile synthons for diverse transformations (Scheme 1C) [37, 38]. Such structural motifs have previously been accessed from alkenes via reductive vinylidene transfer using 1,1‐dichloroalkenes—most of which require independent preparation—using bespoke transition metal catalyst and stoichiometric Zn reductant, as exemplified by the elegant work of Uyeda [39, 40, 41]. Complementarily, we demonstrate that electronic modulation of the iodomethylphosphonium reagent dramatically improves the reactivity toward both ATRA and Wittig processes, thereby enabling the direct use of alkenes and aldehydes. Consequently, our approach offers a streamlined and modular entry to these valuable scaffolds from two simple and readily available building blocks.

2. Results and Discussion

2.1. Dual‐Functional Reagent Design

In implementing the envisioned deoxygenative single‐carbon atom transfer, we conceived a one‐pot protocol comprising two sequential transformations: photocatalytic ATRA and cyclizative Wittig olefination (Figure 1A). In the first phase, iodomethylphosphonium (1) can be incorporated into alkenes if a suitable system capable of selectively cleaving the C─I bond is identified. From a mechanistic perspective, α‐phosphonium methyl radical A is posited as a key intermediate preceding radical addition across the double bond. We recently established that an iodomethylthianthrenium salt engages alkenes through an ATRA‐type pathway via intermediacy of an α‐thianthrenium methyl radical [42, 43]. It should be noted that complementary studies employing this reagent were contemporaneously disclosed by the groups of Ritter [44] and Silvi [45, 46]. On the basis of the prior work, we surmise that the phosphonium moiety confers the requisite electrophilicity to render the radical addition process kinetically accessible [47, 48], giving rise to 3‐iodo‐1‐phosphonium species C.

FIGURE 1.

Reaction design and development. (A) Design plan. (B) Optimization study with list of reagents evaluated. All reactions were performed in 0.05 mmol scale. Yields were based on a ¹H NMR analysis of the crude reaction mixture, using 1,1,2,2‐tetrachloroethane (for 3) and dibromomethane (for 5) as internal standards. The x‐ray crystal structure of 1e is shown; PF₆ ⁻ anion is omitted for clarity. [B] = base, BTPP = tert‐butylimino‐tri(pyrrolidino)phosphorane. ^a E/Z ratio = 1.7:1.

Here, the subsequent addition of the aldehyde and base is expected to gate the second synthetic phase, denoted as cyclizative Wittig olefination [49, 50]. Upon deprotonation, in‐situ generated phosphorus ylide D is proposed to undergo a facile 3‐exo‐tet cyclization that outcompetes direct bimolecular capture by the aldehyde [51]. The ensuing cyclopropyl phosphonium salt E then engages in a Wittig olefination to eventually provide the desired alkylidenecyclopropane product. In the overall sequence, as the release of two functional groups from 1 is accompanied by twofold deprotonation, the atomic carbon from 1 is, in effect, delivered to the final product F.

As a starting point in our design plan, we investigated the ATRA reactivity of iodomethyltriphenylphosphonium salt (1a•PF₆) with unactivated alkene 2 (Figure 1B). Whereas initial trials employing the photocatalyst alone proved ineffective, extensive optimization revealed that a combination of 5 mol% 4CzIPN and 10 mol% AgBF₄ could promote C─I bond cleavage and subsequent fragment transfer to the alkene, providing the ATRA product 3a in 46% yield (entry 1). Despite the promising result in the photocatalytic step, subjecting the crude mixture containing 3a to conditions for Wittig reaction (benzaldehyde 4 1.5 equiv., BTPP 4.0 equiv., THF at 60°C for 18 h) led to only trace amounts of the target product 5. Further evaluation of base, solvent, and temperature failed to improve the chemical yield.

At this stage, our attention turned to modifying the phosphonium motif of the reagent in pursuit of enhancing both ATRA and Wittig reactivity. When reagents adorned with electron‐donating groups (1b and 1c) or a fluorine substituent (1d) at the para position were employed, only marginal improvements or even slightly diminished outcomes were observed in the first photocatalytic process, and the ensuing cyclizative Wittig reaction remained sluggish (entries 2–4). Conversely, the para‐trifluoromethyl‐substituted reagent could not be prepared presumably due to the low nucleophilicity of the corresponding phosphine. Alternatively, incorporating an ortho‐fluorine group on the aromatic ring proved synthetically feasible, permitting the synthesis of reagent 1e•PF₆. To our surprise, the use of 1e•PF₆ led to the nearly quantitative conversion of alkene 2 to intermediate 3e. More notably, direct solvent exchange to THF followed by the addition of benzaldehyde 4 and BTPP in the same vessel afforded the target product 5 in 94% yield with an E/Z ratio of 1.7:1 (entry 5). Further replacement of the fluorine with a methoxy group (1f) or substitution of one aromatic ring with a methyl counterpart (1 g) resulted in a slight diminution in the yield of 3 while completely shutting down the Wittig transformation. Control experiments indicated that the photocatalyst, silver additive, and light are integral for achieving high‐yielding formation of 3. We note that the optimal reagent 1e•PF₆ is straightforward to prepare from tris(2‐fluorophenyl)phosphine and diiodomethane, and was isolated as a bench‐stable beige solid (see Supporting Information for details).

2.2. Substrate Generality

With the optimal reagent and conditions in hand, the generality of the deoxygenative single‐carbon atom transfer protocol was investigated. As shown in Table 1, the transformation accommodates a range of unactivated alkenes, furnishing the corresponding alkylidenecyclopropane products with modest E selectivities (5–7). Nitrogen functional groups such as phthalimide (8), sulfonamide (9), and piperidine (10) could be incorporated into the desired products, despite the potential for catalyst poisoning due to substrate coordination to the silver center. Although the current tactic leverages Wittig‐type reactivity, carbonyl functionalities, including ester (11) and carbonate (12), remained intact on account of the chemoselectivity toward aldehyde coupling partners. In addition, phosphate (13) and thioether (14) were also compatible under the reaction conditions. Aliphatic chloride, which might otherwise be labile under nucleophilic or transition metal‐catalyzed conditions [52], was preserved in this transformation (15). The retained halogen in the product could serve as a strategic synthetic site for downstream functionalization. Finally, alkenes bearing electron‐rich furans (16) or electron‐deficient heteroarenes such as pyridines (17) and quinolines (18) readily underwent deoxygenative ring construction with benzaldehyde 4.

TABLE 1.

Reaction scope for one‐pot synthesis of substituted alkylidenecyclopropane. ^a

^a Conditions unless specified otherwise: alkene (0.2 mmol), 1e•PF₆ (1.2 equiv.), 4CzIPN (5 mol%) and AgBF₄ (10 mol%) in MeCN (0.4 mL) at ‐10°C for 13 h upon blue LED irradiation; then BTPP (0.8 mmol), and aldehyde (0.3 mmol) in THF (1.0 mL) at 60°C for 18 h. Isolated yields are reported. The E/Z ratio was determined by ¹H NMR analysis of the crude mixture.

^b30 mol% of AgBF₄ was used.

We next examined alkenes derived from biologically relevant molecules to evaluate the applicability of the present method to more complex systems. For instance, derivatives of Theobromine (19) were readily converted into the corresponding 3‐membered ring product, despite possessing potentially sensitive heteroaromatic C(sp²)─H bonds. Substrates derived from pharmaceutical agents such as Ataluren (20), Ibuprofen (21), and Febuxostat (22) were also successfully engaged, highlighting its potential utility in the context of medicinal chemistry. In addition, the reaction proved capable of incorporating a Clofibric acid motif (23), a known bioactive metabolite, into the strained carbocycle scaffold.

Additional synthetic examples illustrating the scope of aldehydes are presented in Table 1 below. An assessment of the electronic demand on the aryl aldehyde moiety revealed that substrates bearing various substituents, ranging from electron‐donating (24 and 25) to electron‐withdrawing groups (26–28), all provided the strained carbocyclic products in high yields. The reaction efficiency was not significantly influenced by the steric perturbations on phenyl rings, as both meta‐ and ortho‐substituted aldehydes were readily accommodated (29–31). Common protecting groups such as benzyloxy (32) and oxidation‐sensitive thiomethyl groups (33) were also well‐tolerated. Moreover, heterocyclic aldehydes, including 2‐bromofuran (34) and pyridines (35 and 36) efficiently underwent net three‐component coupling with alkenes and 1e•PF₆ to afford the respective three‐membered carbocyclic products. However, further evaluation with aliphatic aldehydes provided unsuccessful results.

2.3. Synthetic Utility

Leveraging the versatility of the cyclopropane and alkene motifs in alkylidenecyclopropanes [41], we subsequently explored their synthetic utility (Figure 2). First, under Pd catalysis using Pd(t‐Bu₃P)₂, alkylidenecyclopropane 5 was smoothly converted into the conjugated diene 37 via catalytic ring‐opening isomerization. Notably, despite the use of an E/Z mixture of starting materials, the product was obtained with high stereoselectivity as the (E,E)‐isomer, highlighting the stereoconvergent nature of this transformation. The alkenyl moiety was selectively functionalized through reaction of 5 with 5,5‐dimethyl‐1‐pyrroline N‐oxide (DMPO) as a 1,3‐dipole, affording the cycloaddition product 38 in 89% yield. Furthermore, catalytic alkene hydrogenation furnished cis‐disubstituted cyclopropane 39 in 92% yield with a 20:1 diastereomeric ratio. Given the greater steric constraints associated with cis‐configurations relative to trans counterparts, this sequence provides a reliable route to otherwise difficult‐to‐access diastereomers in the synthesis of disubstituted cyclopropanes. Upon treatment with m‐CPBA, ring expansion was also achieved, presumably via the intermediacy of an oxaspiro[2.2]pentane, resulting in 1,2‐ and 1,3‐disubstituted cyclobutanones (40 and 41) in excellent combined yields. Although the regioselectivity was moderate and could not be improved further, the two regioisomers were readily separable, allowing access to two distinct substitution patterns.

FIGURE 2.

Derivatization of alkylidenecyclopropanes. Conditions: (i) Pd(t‐Bu₃P)₂ (5 mol%), toluene (0.5 M), rt, 4 h; (ii) DMPO (1.5 equiv.), toluene (0.4 M), 80°C, 24 h; (iii) Pd/C (2.5 mol%), H₂ (1 atm), EtOAc (0.4 M), rt, 3 h; (iv) m‐CPBA (1.5 equiv.), CH₂Cl₂ (0.1 M), 0°C, 2 h.

2.4. Mechanistic Considerations

After demonstrating the synthetic virtue of the deoxygenative single‐carbon atom transfer strategy, we undertook mechanistic studies. The key mechanistic questions encompassed (1) the necessity of the synergistic combination of the photocatalyst and silver additive in the ATRA step, (2) the unique competence of 1e•PF₆ for the overall two‐part one‐pot transformation, and (3) the fundamental properties of the key radical species involved. Control experiments examining the identity of the silver salt provided insight into its role (Figure 3A). Not surprisingly, replacing AgBF₄ in the optimal conditions with AgPF₆, AgOTs, or AgF still gave near‐quantitative conversion of 2 to 3e•PF₆ (entries 1–3). In contrast, omitting the photocatalyst revealed a striking dependence of reactivity on the nature of the counteranion. Silver salts with weakly coordinating counteranions (i.e., BF₄ ⁻ and PF₆ ⁻) were ineffective alone (see Figure 1B, entry 9 and Figure 3A, entry 1), while replacement with OTs⁻ led to a modest improvement in reactivity (entry 2). Unexpectedly, silver fluoride was capable of mediating the ATRA reaction with high reactivity even in the absence of the photocatalyst (entry 3). However, in the case of AgI, the reaction was suppressed irrespective of the presence or absence of the photocatalyst, suggesting that it does not participate in the catalytic process (entry 4).

FIGURE 3.

Mechanistic insight into the photocatalytic ATRA process. (A) Effect of silver salt identity on ATRA reactivity with and without a photocatalyst. (B) Reaction progress monitoring by ¹H NMR spectroscopy. (C) TEM image of the inorganic solids formed during the reaction. (D) XPS spectra of the inorganic solids formed during the reaction for Ag 3d core levels. Experimental data (blue circles), fitted envelope (blue line), and deconvoluted components corresponding to Ag⁰ (green line) and Ag⁺ (red line) are shown. (E) Reaction with pre‐formed silver nanoparticles (20 wt% on carbon, 10 mol%, top); reaction with benzil (1.0 equiv., middle); and triplet quenching experiments using quenchers (20 mol% each, bottom). (F) Proposed mechanism for the ATRA reaction. E _T = triplet energy (kcal/mol); ΔG°_sol = computed Gibbs free energy difference (kcal/mol).

The foregoing results indicate that silver salts are not merely additives (e.g., iodide scavengers), but instead play the dominant role in reagent activation, rather than the employed photocatalyst. In this regard, the distinct reactivity of AgF may be attributed to its well‐known ability to undergo ligand‐to‐metal charge transfer (LMCT) upon photolysis, generating Ag(0) that subsequently aggregates into silver nanoparticles (AgNPs). Indeed, Wang and co‐workers demonstrated that the combination of AgF and light can effectively mediate related dehalogenation reactions of aryl and alkyl halides via an energy‐transfer pathway gated by in‐situ formed AgNPs [53]. We therefore propose that, in our optimal photocatalyst/AgBF₄ system, the photocatalyst may facilitate the reductive generation of AgNPs, which likely serve as the catalytically active species responsible for initiating the ATRA process.

To probe this hypothesis, the reaction progress was monitored by ¹H NMR spectroscopy using aliquots collected over 8 h. As shown in Figure 3B, a clear induction period of approximately 30 min was observed, during which no product formation occurred. Following this period, the reaction proceeded with a steady increase in product formation. This kinetic behavior can be interpreted as reflecting the initial formation of catalytically relevant species. Regarding the identity of the newly formed species, the reaction mixture became a suspension upon completion, and the corresponding inorganic solid was isolated by centrifugation. Combined analyses using transmission electron microscopy (TEM), high‐angle annular dark field−scanning transmission electron microscopy (HAADF‐STEM), and energy dispersive x‐ray spectroscopy (EDS) elemental mapping confirm that the isolated solids comprised AgNPs and AgI (Figure 3C; see the Supporting Information for additional data). Furthermore, x‐ray photoelectron spectroscopy (XPS) analysis provided quantitative evidence for the coexistence of Ag(0) and Ag⁺ species, with metallic Ag(0) estimated to account for approximately 47.4% of the total Ag signal (Figure 3D). Finally, when preformed AgNPs on graphite (AgNPs/C) were employed, they showed appreciable catalytic activity, corroborating their potential role in heterogeneous photocatalysis (Figure 3E, top).

Concerning the photochemical reactivity of AgNPs, we considered the possibility that an energy‐transfer pathway is operative. As an experimental probe, the known triplet sensitizer benzil was utilized. Indeed, the desired product 3e•PF₆ was obtained in 59% yield (Figure 3E, middle). Furthermore, additional control experiments were designed to assess whether triplet quenchers could impede the productive pathway. The addition of cyclooctatetraene, trans‐stilbene, and anthracene under standard conditions significantly suppressed the photocatalytic ATRA reaction, while it was completely inhibited under an O₂ atmosphere (Figure 3E, bottom).

Based on the preliminary experimental evidence, a plausible pathway for the ATRA process is illustrated in Figure 3F, with additional support from DFT calculations. The reaction commences with the formation of AgNPs via a redox process between the photocatalyst and the silver precursor, although more in‐depth studies are required to fully elucidate this step. The resulting AgNPs can act as energy‐transfer catalysts, presumably due to surface plasmonic effects [53], thereby promoting excitation of reagent 1e. This is further supported by the relatively low triplet energy (E _T) of reagent 1e•PF₆, estimated to be 45.8 kcal/mol. The consequent excited species ³1e is calculated to possess a markedly elongated C─I bond (3.44 Å vs. 2.17 Å in 1e), with one of its SOMOs primarily localized on the corresponding antibonding orbital. Subsequently, ³1e evolves into the α‐phosphonium methyl radical A‐1e via C─I bond homolytic cleavage with coproduction of I₂, which is calculated to be energetically favorable by –12.4 kcal/mol (see Figure S44 in the Supporting Information for details). Considering the half‐wave peak potential of 1e•PF₆ was measured to be –0.65 V vs. SCE in MeCN, the photocatalyst‐mediated single‐electron transfer (E(PC⁺/PC*) = –1.04 V vs. SCE) cannot be completely excluded at this stage. While the excited state of 4CzIPN undergoes bimolecular quenching by reagent 1e•PF₆, with a Stern–Volmer constant of 25.0 M⁻¹, in the presence of AgBF₄, the quenching efficiency is significantly attenuated (see Supporting Information in detail). We infer that AgBF₄ interferes with the direct interaction between the photocatalyst and the substrate. Regardless of the mechanism, both pathways converge on A‐1e, which initiates the ATRA process through a chain‐propagation pathway.

To validate the mechanistic features of the post‐radical generation step, the reaction pathway stemming from A‐1e was evaluated by quantum chemical calculations (Figure 4A). In an effort to assess the impact of fluorine substitution in this step, radical species A‐1a derived from less effective reagent 1a was also examined as a control. As shown in Figure 4A, both radicals traverse the transition state TS‐1 by engaging with alkene 2; however, TS‐1e (ΔG ^ǂ _sol = 7.4 kcal/mol) lies 4.9 kcal/mol lower in energy than TS‐1a (ΔG ^ǂ _sol = 12.3 kcal/mol). These results indicate the observed kinetic preference might be primarily governed by electronic properties of carbon radical center. This trend is further reflected in the thermodynamic profiles, wherein the radical addition of A‐1e (ΔG°_sol = –12.8 kcal/mol) is calculated to be more exergonic than that of A‐1a (ΔG°_sol = –9.9 kcal/mol). The resulting radical adducts (B‐1e and B‐1a) subsequently undergo halogen‐atom transfer with reagents 1e and 1a, respectively. This process is kinetically feasible for both cases, with kinetic barrier (ΔG ^ǂ) of 12.0 kcal/mol for B‐1e and 11.7 kcal/mol for B‐1a, ultimately affording the desired product. During this event, the corresponding α‐phosphonium methyl radicals are concurrently generated, allowing the next propagation to proceed.

FIGURE 4.

Mechanistic studies of the post‐radical generation pathway. (A) DFT energy landscape of radical‐addition pathway. (B) Additional experiments supporting a chain‐propagation mechanism. (C) FMO analysis of the α‐phosphonium methyl radicals. All computations were performed at the SMD(MeCN)‐B3LYP‐D3//6‐311+G**/SDD(I, Ag)//6‐311+G**/ at the level of theory (ΔG°_sol in kcal/mol). Molecular orbitals were visualized with an isovalue of 0.04. The global electrophilicity index (ω) was calculated according to the recent report by Nagib. SOMO = singly occupied molecular orbital. LUMO = lowest unoccupied molecular orbital.

We next sought experimental clues to substantiate the involvement of a radical‐mediated chain process. To this end, the known radical initiator was employed to examine whether the ATRA reaction is operative under conditions devoid of photoirradiation and a photocatalyst (Figure 4B). When a mixture of the alkene and 1e•PF₆ was exposed to the Et₃B/O₂ system [54, 55], the desired transformation proceeded even in the absence of light, albeit with low efficiency (entry 1). In a similar vein, the use of benzoyl peroxide under thermal activation [56] also led to the product formation in 22% yield (entry 2), supporting the feasibility of a chain propagation. On the other hand, light/dark experiments confirmed that under the standard photocatalytic conditions, the reaction ceased in the absence of light and resumed upon re‐irradiation (see Supporting Information for details). A key inference from this result is that the radical species involved may operate with short chain lengths [57], thereby necessitating continuous photoirradiation to sustain radical generation.

2.5. Intermediacy of α‐Phosphonium Methyl Radical

Since α‐phosphonium methyl radicals have rarely been proposed as synthetic intermediates, we sought to elucidate their fundamental properties. As seen in Figure 4C, Mulliken spin‐density plot and natural bond orbital (NBO) analysis of A‐1e show that the radical character resides mostly on C1 (spin density: 1.09), whereas the positive charge is predominantly localized on the phosphonium center (NBO charge on P: +1.53). This charge–spin separation is characteristic of a distonic radical cation [58, 59]. Accordingly, the phosphonium moiety is poised to impart an electrophilicity that is well‐matched toward unactivated alkenes [47]. This notion was further quantified using the global electrophilicity index (ω), which has been demonstrated to be a reliable descriptor of radical polarity in comprehensive studies by Nagib and coworkers [60].

For direct comparison with known neutral carbon‐centered radicals, the PF₆ ⁻ counteranion was explicitly included in this evaluation. The global electrophilicity index of A‐1e is determined to be 2.79 eV, which surpasses that of typical carbon‐centered radicals bearing multiple electron‐withdrawing groups (e.g., •CF₃, 1.67 eV) and is comparable to that of β‐diketone radicals (e.g., 1,3‐dioxo‐1‐phenylbutan‐2‐yl radical, 2.75 eV) [60]. Frontier molecular orbital (FMO) analysis provides a rationale for the beneficial effect arising from electronic tuning. The SOMO of A‐1e arises from the mixing of the p orbital at C1 with one of the three aromatic π orbitals. Substitution at the aromatic ring is thus expected to accentuate the electrophilicity of the radical center. Upon further evaluation of other α‐phosphonium methyl radicals A (Figure 4C, bottom), the SOMO energy was found to vary with the nature of the arene substituents. Notably, that of A‐1e is the lowest among the series (–7.41 eV) and even lies below the π‐bonding HOMO of alkene 2 (–7.31 eV). As a consequence, electron donation from the HOMO of 2 to the SOMO of A is energetically favored only in the case of A‐1e. Taken together, these computational results suggest that electronic perturbation at the aromatic ring tunes the SOMO energy to maximize electronic communication between the α‐phosphonium methyl radical and the alkene substrate. Lowering the addition barrier is crucial to compensate for the fleeting nature of the distonic radical cation, thereby enabling more efficient chain process.

2.6. Reagent‐Dependent Chemoselectivity in Cyclizative Wittig Olefination

We next designed a series of comparative studies to gain insight into the superiority of 1e•PF₆ in the Wittig olefination stage (Scheme 2). To this end, two photochemically generated 3‐iodo‐1‐phosphonium intermediates—one from 1e•PF₆ and the other from its ineffective analogue 1a•PF₆—were independently prepared and subjected into the stoichiometric conditions. For practical reasons, 4‐phenyl‐1‐butene was selected as the model alkene for this mechanistic study. When the pre‐isolated 3‐iodo‐1‐phosphonium salts 42a and 42b were treated with 1.0 equivalent of BTPP, ylide‐mediated cyclizations proceeded in an anti‐selective manner, affording the corresponding cyclopropyl phosphonium salts 43a and 43b in high yields (Scheme 2A).

SCHEME 2.

Reactivity of cyclopropyl phosphonium salts.

In contrast to the similar cyclization efficiency, a dramatic switch in chemoselectivity was observed in the ensuing Wittig olefination step (Scheme 2B). For instance, exposure of 43a to benzaldehyde 4 in the presence of 2.0 equivalents of BTPP resulted predominantly in undesired oxidation at the phosphorus center, as evidenced by the formation of the phosphine oxide–containing cyclopropane 44a. In line with this side pathway, quantitative ³¹P NMR analysis of the crude mixture from the one‐pot reaction with 1a•PF₆ identified the phosphine oxide 44a as the major product (48%, see Supporting Information for details). By contrast, under otherwise identical conditions, 43b reacted cleanly with 4 to generate the strained ring product 6 in 82%, without any detectable amount of the unwanted phosphine oxide 44b.

These comparative studies conform to a mechanistic scenario in which hydrolysis (E → F) and Wittig olefination (E → I) occur in parallel, with their relative extents dictated by the electronic properties at the phosphorus center (Scheme 2C). We attribute this dichotomy to the equilibrium distribution between the phosphonium E and phosphorane ylide H [61, 62]. In the absence of the electron‐withdrawing group, phosphonium salt E dominates the equilibrium. We postulate that hydroxide, generated in‐situ from residual water, attacks the phosphorus center of E, forming hydroxylphosphorane F [63, 64, 65]. The pentavalent intermediate would then decompose to the phosphine oxide G. Relatedly, independent treatment of 43a with ¹⁸O‐labeled water in the presence of BTPP afforded the ¹⁸O‐incorporated phosphine oxide ( ¹⁸O‐G) (see Supporting Information for details), thereby demonstrating the provenance of the oxygen atom source. Conversely, incorporation of a fluorine substituent increases the acidity of α‐C─H bond adjacent to the phosphonium center, thereby making the formation of phosphorane ylide H more thermodynamically favored. This species then reacts with an external aldehyde through a [2+2]/retro‐[2+2] sequence to eventually provide an alkylidenecyclopropane product. Overall, the electronic design of the reagent is crucial for diverting competing hydrolysis toward the productive Wittig pathway.

3. Conclusion

In summary, we have introduced an iodomethylphosphonium reagent that establishes an unprecedented assembly mode between alkenes and aldehydes for the direct synthesis of alkylidenecyclopropanes. The success of this net deoxygenative single‐carbon atom transfer is rooted in the electronic tailoring of the reagent, which not only enables photocatalytic ATRA but also steers the phosphorane ylide reactivity toward cyclizative Wittig olefination over competing oxidation pathways. The method is robust, operates in a one‐pot fashion, and exhibits high functional group compatibility, thereby providing a modular platform to navigate the product space of strained small carbocycles. Whereas current single‐carbon atom transfer reactions are largely reliant on highly reactive and often explosive precursors (e.g., carbene donors and diazo compounds), this study demonstrates that harnessing orthogonal reactivity at a single‐carbon center can unlock new opportunities to design stable yet reactive atomic carbon equivalents for practical synthesis.

Author Contributions

Youngeun Song: conceptualization, methodology, investigation, writing – review and editing, validation, visualization, formal analysis, writing – original draft. Mike Ong: methodology, validation, writing – review and editing, formal analysis, investigation. Suhyeon Kim: formal analysis, investigation. Ahyeon Choi: formal analysis, visualization, investigation, validation. Yongmin Kwon: writing – review and editing, investigation, validation. Seung Youn Hong: conceptualization, funding acquisition, writing – review and editing, supervision, resources, visualization, project administration, writing – original draft.

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

The authors have cited additional references within the Supporting Information.

Supporting file 1: anie72737‐sup‐0001‐SuppMat.pdf.

Supporting file 2: anie72737‐sup‐0002‐DataFile.cif.

Data Availability Statement

The data that support the findings of this study are available in the supplementary material of this article.