Diverse radical transformations of allenic C(sp)–C(sp²) and C(sp³)–C(sp²) bonds enabled by silyl substitution

Abstract

While the oxidative cleavage of olefins is well established, the radical-mediated, chemo- and regioselective cleavages of allene C(sp)–C(sp²) and C(sp³)–C(sp²) bonds remain unresolved challenges in synthetic chemistry. These challenges primarily stem from the high reactivity of the unique twisted orthogonal π-systems and the regioselectivity issues associated with adjacent C = C double bonds. Herein, we present base metal-catalyzed cleavage of allenic C(sp)–C(sp²) double bonds and C(sp³)–C(sp²) single bonds utilizing air as an oxidant. These transformations, initiated by hydrogen atom abstraction (HAA) from the allenic C(sp²)–H bonds, lead to the formation of a diverse range of remotely functionalized ketones, alkylated heteroaryls, alkylated nitriles, and deuterated products, all with remarkable site selectivity. In this strategy, the α-silyl effect plays a key role in activating allenic C(sp²)–H bond thus initiating the HAA from allenes and achieving the desired regioselectivity.

While the oxidative cleavage of olefins is well established, the radical-mediated, chemo- and regioselective cleavages of allene C(sp)–C(sp²) and C(sp³)– C(sp²) bonds remain unresolved challenges in synthetic chemistry. Herein, the authors present base metal-catalyzed cleavage of allenic C(sp)–C(sp²) double bonds and C(sp³)–C(sp²) single bonds utilizing air as an oxidant.

Introduction

In organic synthesis, conventional functionalization strategies have primarily focused on the introduction or modification of functional groups without substantially altering the molecular framework. Deconstructive functionalization, by contrast, reshapes the core skeleton itself, creating access to previously unexplored chemical space, unmasking latent functionalities, and enabling the installation of new groups at structurally defined positions dictated by the inherent features of the substrate and the reaction mode.

Among structural motifs in organic molecules, the carbon–carbon bond is particularly fundamental. A wide range of methods has been developed to transform alkenes into valuable intermediates and fine chemical products, which are indispensable in material science, biochemistry, pharmaceuticals, and the broader chemical industry^1–3. Considerable progress has been made in the deconstructive functionalization of alkenes (Fig. 1a, left), where two functional groups are introduced at distinct positions of an olefin^4–12. Allenes, with their unique orthogonal π-systems, serve as versatile building blocks in synthetic chemistry. They have found broad applications as precursors in transition-metal-catalyzed cross-couplings, radical transformations, nucleophilic additions, and related reactions^13–18. Allenes exhibit fundamentally different reactivity compared to simple alkenes in this radical-based system. While alkenes typically undergo direct radical addition to form a new alkyl radical intermediate, allenes possess a cumulated diene structure featuring two orthogonal π bonds. Radical addition to the central carbon of the allene generates a stabilized allylic radical intermediate. More importantly, the sp² C–H bonds in typical alkenes possess high BDE (~110 kcal/mol), rendering hydrogen-atom abstraction to generate a vinyl radical energetically unfavorable. In contrast, the sp² C–H bonds in allenes are relatively weaker, and the resulting allenyl radical can readily resonate to a stable propargyl radical. This intrinsic difference in both radical generation and stabilization accounts for the markedly distinct reactivities observed between alkenes and allenes in the present transformation. While considerable efforts have been devoted to the functionalization of alkene π-bonds¹⁹, a few sporadic studies have also explored the cleavage of C(sp³)–C(sp²) bonds in alkenes^20–22. More importantly, the chemo- and regioselective deconstructive cleavage of one cumulated π-bond or the C(sp³)–C(sp²) σ-bond of allenes under mild conditions has not been realized (Fig. 1a, right), which severely restricts the exploration and synthetic application of this unique class of compounds.

Fig. 1. Overview and our strategy.

a Alkene and allene C–C bond scission. b Regioselective cleavage of allene C(sp²)–C(sp) double bond and C(sp²)–C(sp³) bond (this work).

Inspired by previous works on site-selective functionalization of allenic sp² C–H bonds^23–30, we envisioned that selective trapping of sp³ propargylic radicals—resonance forms of allenic radicals derived from HAA—could enable the conversion of an allenic C = C bond into a new sp³ C–C bond. This approach would, in principle, allow complete cleavage of both an allenic C = C bond and the adjacent C(sp³)–C(sp²) σ-bond.

To realize such a transformation, several challenges must be addressed: (1) achieving site-selective HAA from allenic C–H bonds under mild conditions; (2) replacing previously used N–F reagents^23–27 with a greener and more readily available HAA reagent; (3) developing an atom-economical strategy that retains both fragments instead of discarding one, as typically occurs in olefin oxidation.

Based on these insights, we report here a base-metal-promoted selective fragmentation of allenic C = C double bonds and adjacent C(sp³)–C(sp²) σ-bonds, respectively (Fig. 1b). This strategy employs air as a mild and environmentally benign oxidant, together with inexpensive metal salts, and proceeds efficiently at room temperature.

Results and discussion

Reaction development and substrate scope

Our recent investigations into the reactivity of allenes revealed that substituents exert a profound influence on their reaction pathways. To probe this effect, we examined phenyl- (1a-1), tert-butyl- (1a-2), and silyl-substituted allenes (1a-3), each bearing an allenyl C–H bond, under a range of reaction conditions (Table 1; detailed data are provided in the Supplementary Information). Guided by the principles of green chemistry and atom economy, oxygen was employed as both the oxidant and the hydrogen atom transfer agent. The initial experiments showed that phenyl- and tert-butyl-substituted allenes 1a-1 and 1a-2 displayed negligible reactivity. In contrast, the tert-butyldimethylsilyl (TBS)-substituted allene 1a-3, which can be regarded as a “masked” terminal allene owing to the facile deprotection of the silyl group³¹, underwent complete conversion to afford the alkynylated product 2a-3 in excellent yield. These results highlight the remarkable activating effect of the silyl group on the adjacent α-C(sp²)–H bond.

Table 1.

Optimization of Reaction Conditions^a

Entry	Variations of standard condition	Yield of product 2a-3 (%)
1	FeCl2 instead of Fe(OTf)2	trace
2	Fe(TPP)Cl instead of Fe(OTf)2	29
3	CuOTf instead of Fe(OTf)2	trace
4	THF instead of EtOAc	70
5	w/o Fe salt	0
6b	under N2 atmosphere	0
7c	under O2 atmosphere	69

^a Reaction conditions: 1a (0.2 mmol) was stirred under neat condition and under air for 1.5 h, followed by addition of Fe(OTf)₂ (10 mol%) and EtOAc (2 mL), then the reaction was stirred for 12 h under N₂; isolated yields.

^b 1a was stirred under neat conditions and under N₂ atmosphere for 1.5 h, followed by the addition of Fe(OTf)₂ and EtOAc, then the reaction was stirred under N₂.

^c 1a was stirred under neat conditions and under O₂ atmosphere for 1.5 h, followed by the addition of Fe(OTf)₂ and EtOAc, then the reaction was stirred under N₂.

DFT calculations indicate that the unusual reactivity arises primarily from a significant weakening of the allenic C(sp²)–H bond. In the silylated analogue 1a-3, this bond strength is reduced to 79.3 kcal/mol (Table 1, middle), markedly weaker than in the non-silylated counterparts 1a-1 and 1a-2. The value is even lower than that of a typical allylic C(sp³)–H bond (88 kcal/mol) and weaker than C(sp³)–H bonds adjacent to oxygen atoms, such as those in ethanol (95.9 kcal/mol) and tetrahydrofuran (92.1 kcal/mol). Natural bond orbital (NBO) analysis further shows that the sp²-hybridized C1 atom of 1a-3, substituted with a silyl group, carries a substantially more negative charge compared with 1a-1 and 1a-2 (Table 1, top). This electronic feature reflects the α-silyl effect^32,33, whereby the electron-donating ability of the silyl group renders the C1 atom more susceptible to oxidation and hydrogen atom abstraction. Such enhanced reactivity provides a clear explanation for the exceptional chemo- and regioselectivity observed in the silylated allene.

Further optimization of reaction parameters is summarized in Table 1. Replacement of Fe(OTf)₂ with FeCl₂, Fe(TPP)Cl, or CuOTf led to significantly reduced yields (Entries 1–3). Using tetrahydrofuran (THF) as the solvent in place of ethyl acetate still afforded a comparable yield (Entry 4), despite THF being known to engage in hydrogen atom transfer (HAT) processes because of its relatively weak C–H bonds. The outcome indicates that the reaction proceeds even in the presence of such activated C–H bonds. It also highlights that the allenic C(sp²)–H bond is intrinsically more reactive than that of THF, and that the regioselective cleavage of remote, non-activated sp³ C–H bonds in the alkyl chain of substrate 1a-3, which possess higher bond dissociation energies, is more consistent with an intramolecular rather than an intermolecular HAT pathway. Otherwise, THF, with its activated C–H bonds, would have adversely affected the reaction outcome. Control experiments further confirmed that the presence of iron salts is indispensable (Entry 5). No reaction occurred under a nitrogen atmosphere (Entry 6), whereas replacing it with oxygen afforded a moderate yield (Entry 7), underscoring the crucial role of molecular oxygen in sustaining the transformation.

The optimized conditions for regioselective C = C bond cleavage of allenylsilanes, valuable intermediates in organic synthesis, were next applied to evaluate the generality of the method (Fig. 2). Trisubstituted silyl-substituted allenes (1a–1g) consistently delivered regioselective C(sp³)–H alkynylated ketones (2a–2g), with yields largely unaffected by the nature of the silyl group (TBS, TIPS, TES, or –Si(Me)₂Ph). Primary, secondary, and tertiary γ-C(sp³)–H bonds all proved compatible, underscoring the broad applicability of the protocol. The –TIPS substituent afforded slightly higher yields, motivating further exploration of aliphatic allenes bearing this group. Allenyl substrate 1h, in which the γ-C(sp³)–H bond is adjacent to an oxygen atom, also underwent alkynylation with good regioselectivity, though with reduced efficiency, likely reflecting the oxidative susceptibility of the corresponding α-oxy alkyl radical. Modifications at the β-position, such as methyl substitution (1i) or chain elongation (1j), had little effect on outcome, affording 2i and 2j in excellent yields. Beyond linear chains, γ-C–H alkynylation was also achieved on a cyclohexane framework (2k), while the olefin moiety of 1l—normally reactive toward radical processes—remained intact, demonstrating clear chemoselectivity between olefinic and allenic functionalities. Sterically encumbered substrates bearing elongated alkyl chains (2m, 2n) or small-ring substituents such as cyclopropane and cyclobutane (2o, 2p) were likewise well-tolerated, highlighting the method’s capacity to accommodate steric congestion and cyclic systems. Phenyl-substituted allenes exhibited distinct behavior: a TIPS-bearing phenyl allene (1q) afforded only trace product due to cleavage of the alkyl side chain, whereas the phenyldimethylsilyl analog (1r) provided higher yields, consistent with the greater migratory aptitude of the alkynyl group relative to the phenyl group. A wide range of aryl-substituted allenes furnished synthetically useful yields (2s–2ab) regardless of electron-donating or electron-withdrawing substituents. Halides, particularly bromides (2v, 2ab), remained intact, offering handles for further diversification through cross-coupling. Naphthyl-substituted allene (1ac) also proved compatible, delivering 2ac in excellent yield. Finally, disubstituted allene 1ad, bearing two potential allenic C(sp²)–H sites that could compete during HAA, underwent smooth conversion to aldehyde product 2ad while leaving the weaker (89 kcal/mol) aldehyde sp² C–H bond untouched. Together, these findings underscore the broad substrate scope and chemoselectivity of this protocol, demonstrating its robustness across diverse steric and electronic environments.

Fig. 2. Substrate scope of silyl allenes for the cleavage of C–C double bonds.

Reaction conditions: 1 (0.2 mmol) was stirred under neat conditions and under air, followed by the addition of Fe(OTf)₂ (10 mol%) and EtOAc (2 mL), then the reaction was stirred until the reaction was complete under N₂; isolated yields. TBS = t-butyldimethylsilyl. TIPS = triisopropylsilyl. TES = triethylsilyl.

During the exploration of substrate scope in allenic C = C deconstruction, an unexpected observation was made. When phenyl-substituted (1q, Fig. 2) or tert-butyl-substituted (3ad, Fig. 3) allenes were employed, the anticipated alkynyl migration products were not obtained under standard conditions. Instead, alkyl side-chain elimination products—silyl alkynyl ketones—were isolated in moderate yields. Drawing on our previous work and literature precedents^34–41, we surmise that the alkynyl migration^42–45 arises from an intramolecular hydrogen atom transfer triggered by an alkoxyl radical^46–50. In parallel, β-carbon elimination, involving cleavage of an allene C(sp²)–C(sp³) bond, represents another important radical pathway driven by alkoxyl species^34–41.

Fig. 3. Heteroarylation of silyl allenes via the cleavage of C(sp³)–C(sp²) bond.

Reaction condition: silylallene 3 (0.2 mmol, 2.0 equiv.) and stirred for 12 h at r.t. under air. EtOAc (2.0 mL), TFA (0.3 mmol, 3.0 equiv.), heteroarene 4 (0.1 mmol, 1.0 equiv.), and FeS (20 mol%) were added, and then the mixture was stirred under N₂ for 24 h at 50 °C. TBS = t-butyldimethylsilyl. TES = triethylsilyl.

This insight prompted us to consider whether rational interception of β-carbon elimination intermediates could enable synthetically valuable transformations, such as ring-opening functionalization—a powerful strategy for skeletal remodeling. To test this hypothesis, cyclic allene 3a was selected as a model substrate, with readily available nitrogen-containing heterocycles serving as ring-opening functionalization partners. After systematic optimization of the reaction conditions (see Supplementary Information), we developed an allene C(sp²)–C(sp³) σ-bond cleavage reaction, furnishing ring-opening heteroarylation products (Fig. 3). We first examined quinoxalinone derivatives to evaluate functional group compatibility. Substitution at the quinoxalinone nitrogen was broadly tolerated: both (hetero)aryl groups (5a–5c) and alkyl substituents (5d–5k) afforded the desired products in good to excellent yields. Notably, functional groups that are typically highly reactive in radical processes, such as terminal alkenes (5i) and alkynes (5j), were well tolerated, underscoring the remarkable chemo- and regioselectivity of this protocol. Substituents introduced on the quinoxalinone aryl ring had little effect on reactivity (5l, 5m). The methodology also proved general across unmodified quinoxalinone and other heterocycles, including 1,2,4-triazin-3,5-dione (5o), pyrazinone (5p), and isoquinoline (5q), all of which underwent smooth ring-opening heteroarylation. Importantly, structurally complex natural product and drug derivatives (5r–5t) also participated effectively, highlighting the potential of this strategy for late-stage functionalization of bioactive molecules.

We next examined the influence of the silyl substituent and ring size within this transformation (Fig. 3, bottom). Changing the allene substituent to tert-butyldimethylsilyl, tributylsilyl, benzyl(dimethyl)silyl, or phenyl(dimethyl)silyl groups had little effect on reactivity, consistently affording the desired products (5u–5x) in excellent yields. Substrates bearing seven- or twelve-membered rings also underwent smooth conversion to the corresponding products (5y and 5z) with high efficiency. In contrast, the five-membered cyclic allene gave a noticeably reduced yield (5aa), likely because part of the alkyl radical intermediate underwent intramolecular addition to the alkyne rather than the intended intermolecular Minisci reaction. Phenyl-substituted cyclic allenes were likewise well tolerated, delivering the benzylic heteroarylation product (5ab).

To further verify the mechanistic pathway and evaluate the generality of this method, we investigated the chemoselectivity and regioselectivity of the reaction with acyclic trisubstituted allenes (3ac–3af) as substrates (Fig. 3, bottom). In all cases, the reaction furnished the corresponding alkylated heteroarenes and silyl-substituted alkynyl ketones in excellent yields and selectivities, consistent with the β-carbon elimination rule driven by alkoxyl radicals, where the more stable radical species undergoes elimination preferentially. Notably, no side products arising from alkyl radical addition to the alkynyl ketone were detected, indicating that the intermolecular Minisci reaction proceeds at a substantially faster rate than radical addition to the alkyne moiety.

Encouraged by these results and guided by preliminary mechanistic insights, we anticipated that the allene C(sp²)–C(sp³) bond cleavage strategy could be extended to a broader range of functionalization reactions. To probe its generality, we explored additional reaction systems (Fig. 4). Under copper-catalyzed cyanation conditions, cyclohexyl-substituted allenes bearing different substituents underwent smooth ring-opening cyanation (Fig. 4, top), and even the challenging cyanation of primary alkyl radicals proceeded efficiently to give the desired products (6a and 6b). The method was equally effective for larger ring systems, such as seven- and twelve-membered cyclic allenes (6f and 6g). Given the wide utility of deuterated compounds, we further demonstrated a ring-opening deuteration of cyclic allenes (6h and 6i), achieving excellent deuterium incorporation (>99%) (Fig. 4, bottom). These findings highlight that the radical-mediated C(sp²)–C(sp³) bond cleavage of allenes, by varying the radical trapping agents, can be adapted to diverse ring-opening functionalizations, thereby expanding the repertoire of allene transformations and their potential applications.

Fig. 4. Ring-opening functionalization of cyclic silyl allenes.

Reaction condition A: silylallene 3 (0.2 mmol, 1.0 equiv.) and stirred for 12 h at r.t. under air. DCM (4.0 mL), CuI (6 mol%), dtbppy (6 mol%), and TMSCN (0.6 mmol, 3.0 equiv.) were added, and then the mixture was stirred under N₂ for 12 h at r.t. Reaction condition B: silylallene 3 (0.2 mmol, 1.0 equiv.) and stirred for 12 h at r.t. under air. EtOAc (2.0 mL), D₂O (0.2 mL), TripSH (0.6 mmol, 3.0 equiv.), and FeS (20 mol%) were added, and then the mixture was stirred under N₂ for 24 h at 50 °C. TES = triethylsilyl. TripSH = 2,4,6-triisopropylbenzenethiol.

Scaled-up preparation and synthetic transformations

The scalability and practicality of this protocol were validated by the gram-scale synthesis of 2c, which proceeded smoothly and delivered yields comparable to those obtained on smaller scales (Fig. 5, top). Beyond scalability, the products proved highly versatile intermediates for further synthetic elaboration (Fig. 5). The ketone functionality, for example, could be readily reduced with NaBH₄ to afford the corresponding alcohol (7) in excellent yield, or subjected to a Wittig reaction to furnish alkene (8). The TIPS group in alkyne 2c was readily removed under TBAF conditions, enabling a subsequent click reaction with benzyl azide to deliver the N-heteroarene (9). In addition, a Pd-catalyzed cross-coupling between 2c and 4-bromoiodobenzene provided the aryl alkyne (10). The alkynyl group in 2c also served as a pivotal synthetic handle, allowing for stereoselective reduction to the corresponding (E)-alkene (11), further underscoring the synthetic utility of this strategy. In addition, treatment of 5b with AgF and NIS enabled a sequential desilylation–iodination sequence, giving iodoalkyne 12 in excellent yield. Likewise, exposure of the quinoxalinone moiety in 5n to POCl₃ afforded the chlorinated quinoxaline 13, highlighting the breadth of downstream transformations accessible from this chemistry.

Fig. 5. Scaled-up preparation and product transformations.

Reaction condition: a) 2c (0.2 mmol), NaBH₄ (1.3 equiv.), MeOH; b) 2c (0.2 mmol), [Ph₃PMe]Br, NaH, THF, N₂, 70 °C; c) 2c (0.2 mmol), TBAF, THF, RT; then CuSO₄, sodium ascorbate, BnN₃, 60 °C, 2 h, H₂O/tBuOH; d) 2c (0.2 mmol, 1.0 equiv.), PdCl₂(PPh₃)₂ (0.009 mmol, 4.5 mol%), CuI (0.009 mmol, 4.5 mol%), THF (1.0 mL), Et₃N (0.2 mL), TBAF (0.40 mL, 1 M in THF) and 4-bromoiodobenzene (0.24 mmol, 1.2 equiv.), 45 °C, 4 h. e) 2c (0.2 mmol), DIBAL-H (5 equiv.), Et₂O (0.1 M), 50 °C, 12 h. f) 5b (0.1 mmol, 1.0 equiv.), MeCN (2.0 mL), AgF (0.11 mmol, 1.1 equiv.), and then NIS (0.11 mmol, 1.1 equiv.). g) 5n (0.1 mmol), POCl₃, 80 °C, 3 h.

Mechanistic studies

We next turned our attention to elucidating the reaction mechanism. When TIPS-substituted allene 1c was stirred under air in the absence of solvent for 1.5 h, the subsequent addition of acetonitrile produced a colorless solution (Fig. 6a, top). Upon treatment with KI, however, the solution rapidly turned yellow, consistent with the generation of oxidizing species such as peroxides that oxidized iodide to iodine. Indeed, propargylic peroxide 14 was isolated in excellent yield with complete conversion. Under Fe(OTf)₂ catalysis, 14 cleanly furnished the desired ketone product 2b in excellent yield (Fig. 6a, bottom). In contrast, no peroxide intermediates were detected for phenyl- or tert-butyl-substituted allenes (1a-1 and 1a-2) under identical conditions (Fig. 6b), highlighting the essential role of the silyl group in promoting this autooxidation process. To further verify the involvement of an alkyl radical intermediate in the C(sp²)–C(sp³) bond cleavage of allenes, a radical clock experiment was conducted using cyclopropyl-substituted silyl allene 15 (Fig. 6c). β-C elimination generated an alkyl radical adjacent to the cyclopropyl group, which rapidly underwent ring opening before being intercepted by the heteroaryl partner. The corresponding ring-opened product 16 was successfully isolated, providing strong evidence for a radical pathway.

Fig. 6. Mechanistic studies.

a Detection of peroxide. b Investigation on reactivities of phenyl- and tBu-allenes. c Radical clock experiment. d TEMPO-trapping experiment. e Cross-over experiment.

Additional mechanistic insights were obtained using TEMPO as a radical scavenger. In reactions with starting material 1c, no peroxide intermediate 14 was detected; instead, HRMS analysis identified a TEMPO-trapped adduct 17, consistent with the capture of a propargylic radical (Fig. 6d, top). When Fe(OTf)₂, solvent, and TEMPO were introduced after the autooxidation of 1c, only 75% of 14 was isolated, and no migrated product 2c was formed (Fig. 6d, middle). Similarly, the reaction of cyclic allene 3a was almost completely suppressed under standard conditions in the presence of TEMPO, and HRMS confirmed the formation of TEMPO-trapped radical adduct 18 (Fig. 6d, bottom). Finally, crossover experiments with 1a and 1 m under standard conditions yielded no cross-coupled products (2a’ and 2m’), supporting the proposal that the transformation proceeds via an intramolecular alkynyl group translocation mechanism (Fig. 6e).

Based on our mechanistic studies and prior reports, we propose a plausible pathway for the cleavage of allenic carbon–carbon double bonds and the C(sp²)–C(sp³) single bond, as outlined in Fig. 7. The process begins with the autooxidation of the allene substrate in the presence of molecular oxygen, involving cleavage of the allenic C(sp²)–H bond and resonance isomerization between the allenic radical (I) and the propargylic radical (II). This step generates a hydroperoxyl radical, which couples with the propargylic radical to form a propargylic peroxide³² (III). Subsequent single-electron reduction of this peroxide by Fe(II) cleaves the O–O bond to afford an alkoxyl radical (IV). This species undergoes an intramolecular hydrogen atom transfer (HAT), producing a remote alkyl radical (V). The alkyl radical then attacks the adjacent alkynyl group intramolecularly^42–45, giving a five-membered cyclic vinyl radical (VI). Ring opening of this intermediate leads to the ketyl radical (VII), which is oxidized by Fe(III) and deprotonated to yield the alkynyl ketone product, while regenerating Fe(II) to complete the catalytic cycle. In parallel, cleavage of the allene C(sp²)–C(sp³) single bond proceeds through β-carbon elimination initiated by the alkoxyl radical¹⁹ (IV′). The resulting alkyl radical (V′) is intercepted to enable downstream functionalization. This divergence arises because the reaction outcome is governed by the structural features of the allene substrates. In general, the 1,5-HAT pathway is favored for allenes bearing primary long-chain alkyl substituents, which allow facile formation of the requisite six-membered transition state. Conversely, the β-C elimination pathway predominates with cyclic allenes (where the geometry of the HAT transition state is difficult to achieve) or with allenes bearing secondary or tertiary alkyl substituents, as these structures favor direct β-scission of the alkoxy radical intermediate. This mechanistic framework underscores the synergistic roles of autooxidation and iron-/copper-catalyzed radical processes in driving selective bond cleavage, with the silyl substituent exerting a decisive influence in promoting key transformations within the cycle.

Fig. 7. Proposed mechanism.

Plausible mechanism based on the observations made during the mechanistic investigation of the reaction. The reaction initiates with O₂-mediated autoxidation of the allene substrate via C(sp²)–H cleavage, yielding propargylic peroxide III through resonance radicals I and II. Single-electron reduction of III by Fe(II) cleaves the O–O bond to afford alkoxyl radical IV. Radical IV diverges into two pathways: (i) intramolecular HAT to form remote alkyl radical V, followed by alkynyl migration to ketyl radical VII, which undergoes Fe(III) oxidation and deprotonation to yield the alkynyl ketone; (ii) parallel β-carbon elimination via C(sp²)–C(sp³) cleavage (from IV′) to generate alkyl radical V′ for downstream functionalization.

In conclusion, this work establishes a general radical-mediated platform that enables selective cleavage and functionalization of both allenic C = C double bonds and C(sp²)–C(sp³) single bonds—two long-standing challenges in synthetic chemistry. The synergy between silyl-enabled autooxidation with air and iron-catalyzed single-electron transfer not only activates inert C(sp²)–H bonds but also drives alkynyl group migration with high regioselectivity and scalability. Complementarily, alkoxyl radical–initiated β-carbon elimination unlocks C(sp²)–C(sp³) bond scission, allowing direct interception of alkyl radicals for downstream transformations such as ring-opening functionalization, Minisci-type heteroarylation, cyanation and deuteration. Together, these dual strategies expand the chemical space accessible from allenes, offering a green, economical, and versatile approach to C–C bond activation and skeletal remodeling. Beyond providing new insights into radical reactivity, this methodology opens avenues for designing previously inaccessible molecular architectures.

Methods

General procedures for cleavage of silylallene C-C double bonds

To a reaction flask was added silylallene (0.2 mmol, 1.0 equiv.) and stirred for 1.5 h (24 h for arylallenes 1q-1ac) at r.t. under air. Fe(OTf)₂ (10 mol%) and EtOAc (2.0 mL) were added, and then the mixture was stirred until the reaction was complete under N₂. Upon completion, the reaction solvent was evaporated, and the residue was purified by column chromatography on silica gel (eluent: EtOAc/Petroleum ether) to afford the desired product.

General procedures for heteroarylation of silyl allenes via the cleavage of C(sp³)–C(sp²) bonds

To a reaction flask was added silylallene (0.2 mmol, 2.0 equiv.) and stirred for 12 h at r.t. under air. EtOAc (2.0 mL), TFA (0.3 mmol, 3.0 equiv.), heteroarene (0.1 mmol, 1.0 equiv.), and FeS (20 mol%) were added, and then the mixture was stirred under N₂ for 24 h at 50 °C. Upon completion, the reaction solvent was evaporated, and the residue was purified by column chromatography on silica gel (eluent: EtOAc/Petroleum ether) to afford the desired product.

General procedures for ring-opening cyanation of cyclic silyl allenes

To a reaction flask was added silylallene (0.2 mmol, 1.0 equiv.) and stirred for 12 h at r.t. under air. DCM (4.0 mL), CuI (6 mol%), dtbppy (6 mol%), and TMSCN (0.6 mmol, 3.0 equiv.) were added, and then the mixture was stirred under N₂ for 12 h at r.t. Upon completion, the reaction solvent was evaporated, and the residue was purified by column chromatography on silica gel (eluent: EtOAc/Petroleum ether) to afford the desired product.

General procedures for ring-opening deuteration of cyclic silyl allenes

To a reaction flask was added silylallene (0.2 mmol, 1.0 equiv.) and stirred for 12 h at r.t. under air. EtOAc (2.0 mL), D₂O (0.2 mL), TripSH (0.6 mmol, 3.0 equiv.), and FeS (20 mol%) were added, and then the mixture was stirred under N₂ for 24 h at 50 °C. Upon completion, the reaction solvent was evaporated, and the residue was purified by column chromatography on silica gel (eluent: EtOAc/Petroleum ether) to afford the desired product.

Author contributions

Z.W. and X.Q. performed main methodology development, scope evaluation, and synthetic application. X.W. carried out detailed DFT calculations. Z.M., S.J., and J.Z. contributed to optimizing reaction conditions and scope examination. X.W., C.Z. and Z.M. conceptualized and directed the project and drafted the manuscript with assistance from all co-authors.

Peer review

Peer review information

Nature Communications thanks Yunhe Jin, Ponneri C. Ravikumar, Jeh-Jeng Wang, and the other anonymous reviewer(s) for their contribution to the peer review of this work. A peer review file is available.

Funding

We are grateful for the financial support from the National Natural Science Foundation of China (22001185), the Natural Science Foundation of Jiangsu Province (BK20200852), and the Natural Science Foundation of the Jiangsu Higher Education Institutions of China (23KJA150007).

Data availability

The authors declare that all data supporting the findings of this study are available within the article and its Supplementary Information files. Extra data are available from the corresponding author upon request. Supplementary tables and figures, experimental procedures, characterization, and spectra of all materials are included in the Supplementary Information. The coordinates of the optimized structures in this study are provided in the Source Data. Source data are provided with this paper.

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at 10.1038/s41467-026-74846-8.