The search for reactive functional groups and synthons has long been one of the most productive wellsprings of discovery in chemistry, enabling the design and development of novel transformations1 and opening opportunities for drug discovery2. As recent striking examples, Suero and colleagues unveiled the dual radical and carbene character of carbyne equivalents generated in situ from a hypervalent iodine reagent bearing a diazo ester3, and Garg and colleagues demonstrated that 1,2,3-cyclohexatriene is a powerful and versatile reagent in synthetic chemistry4. The search for synthetic equivalents for not-yet-existing synthons thus continues to drive progress in synthetic chemistry by enabling unprecedented bond disconnections.

Cyclopropenes, the smallest cyclic alkenes, possess substantial strain energy (54.6 kcal mol−1)5, which leads to a unique reactivity in ring-opening transformations6,7,8 and C=C bond functionalization9,10,11. The π-type cyclopropenium cations (CPCs, I)12,13, a C3+ system generated by removing one substituent from the aliphatic C1 site of cyclopropenes, possess extraordinary stability (Fig. 1a, (1)) due to the aromatic character of this system14. Neutral cyclopropene precursors are easily ionized, because the aromaticity helps to offset the cost of generating the positive charge15,16. Since their discovery, π-type CPCs I have led to important advances in aromaticity theory14, catalysis17 and material science18. The π-type CPCs I also provide a good platform for the synthesis of functionalized cyclopropenes of type A by the addition of nucleophiles on the C3 position19. In contrast, σ-type CPCs II, formally generated from cyclopropenes by removing one substituent from the C1 or the C2 position, have remained unexplored in synthetic chemistry (Fig. 1a, (2)). In σ-type CPCs II, the empty σ-orbital is perpendicular to the C1–C2 π-orbital, thereby making the positive charge localized on a single carbon atom without aromatic stabilization. As such, free σ-type CPCs II decay rapidly into propargylic cations III and give open-chain products of type C20,21. Therefore, σ-type CPCs II can usually not be used to access C1/C2-substituted cyclopropenes of type B, making this class of products more difficult to access. Chemists have therefore developed synthetic equivalents of σ-type CPCs, but only with limited success. Cyclopropenyl bromides22 or iodides23,24 can act as σ-type CPC precursors in the presence of a palladium catalyst, but this approach has been limited to 3-difluoromethylated or 3,3′-difluoro cyclopropenes in cross-coupling with terminal alkynes, alkenes or aryl boronic acids. Considering the versatile role of cyclopropenes in synthetic chemistry25, chemical biology26, and medicinal27 and material28 chemistry, the availability of broadly applicable synthetic equivalents of σ-type CPCs would facilitate the synthesis of functionalized cyclopropenes and accelerate progress in these areas.

Fig. 1: Structure of π-type and σ-type CPCs and our design.
figure 1

a, Bonding analysis and reactivity of π-type and σ-type CPCs I and II. b, Design of transition metal-based σ-type CPC equivalents. c, Synthesis of CpBXs via a cyclopropenyl lithium intermediate. d, Development of iodine(III)-based electrophilic σ-type CPC precursors by an umpolung strategy and σ-type CPC transfer reactions to terminal alkynes and vinylboronic acids via redox gold catalysis. Nu, nucleophile.

Our strategy was based on the generation of a transient electrophilic cyclopropenyl-metal species IV, which could act as a σ-type CPC equivalent through ligand exchange with a nucleophile to give V, followed by fast reductive elimination (Fig. 1b). Key design elements are good stability of the transient organometallic intermediate IV, fast ligand exchange and reductive elimination, and especially catalytic generation of IV, as the use of stoichiometric transition-metal reagents would be not sustainable. Based on the limitations of the reported approaches using cyclopropenyl bromides and iodides22,23,24, more reactive yet stable precursors will be needed for our strategy. Over the past three decades, hypervalent iodine reagents (HIRs) have been broadly applied for the umpolung of nucleophiles owing to their unique combination of extreme leaving-group ability, stability and availability29,30,31,32,33. Building on our previous studies on hypervalent iodine-based reagents34, we attempted the synthesis of the proposed σ-type CPC precursors by mixing iodine(III) compounds and nucleophilic cyclopropenyl partners (Fig. 1c). The cyclopropenyl organolithium reagents could be generated through deprotonation of cyclopropenes s-1 with n-butyllithium at −78 °C. Addition of hypervalent iodine precursors I1 or I2 then gave various cyclopropenyl benziodoxoles (CpBXs) 1 in good yields (see Supplementary Section 2.3 for details). The structure of CpBX 1k was confirmed by X-ray crystallography. The CpBX compounds are stable and easy to manipulate.

In this Article we describe the application of these CpBX reagents as σ-type CPC synthetic equivalents together with gold catalysis35 (Fig. 1d). We demonstrate that the oxidative addition of CpBXs to the gold catalyst occurs under mild conditions36, thus giving access to the σ-type CPC synthetic equivalent IV with high functional-group tolerance. Using terminal alkynes or vinylboronic acids as coupling partners, alkynyl- or alkenyl-cyclopropenes B1 and B2 were accessed, often in close to quantitative yields. Concerning cyclopropenyl-gold species, Hashmi and colleagues disclosed that stoichiometric Au(I)–cyclopropenyl complexes can be used as aurated carbenoids or quasi-carbene precursors37. In contrast, the transient Au(III)–cyclopropenyl species in our study reacted as σ-type CPC equivalents. Our findings reveal that the reactivity of σ-type CPC equivalents can be harnessed based on a gold redox process38,39,40 that enables the divergent synthesis of functionalized cyclopropenes. In addition, the obtained alkynyl–cyclopropenes can serve as versatile building blocks to access valuable functionalized cyclopropenes, cyclopropanes and conjugated enynes. The power of the transformation is further highlighted in the late-stage modification of complex natural products, bioactive molecules and drugs.

Results and discussion

Reaction development

The success of the proposed σ-type CPC transfer reaction hinged on our hypothesis that transition-metal catalysts would prefer to cleave the C–I(III) bond of CpBXs (oxidative addition) rather than the C–C bond of cyclopropenes (ring-opening reactions). Furthermore, coordination of the nucleophilic reaction partner and subsequent reductive elimination will need to be efficient to overcome the expected limited stability of the formed cyclopropenyl intermediate. Terminal alkynes are well-established partners for cross-coupling reactions and are widely represented in commercially available compounds and pharmaceuticals41. The cross-coupling between CpBXs and terminal alkynes would give access to useful alkynyl–cyclopropene building blocks. The synthesis of such compounds has been reported by Hashmi and colleagues using a reverse polarity approach (cyclopropenes as nucleophiles and ethynylbenziodoxole (EBX) reagents as electrophiles)42. However, terminal cyclopropenes with two electron-withdrawing groups are required for efficient C–H activation, resulting in a narrow scope. Monocyclopropenation of 1,3-diynes by transition-metal-catalysed reaction of diazo compounds or their surrogates is also a viable process affording alkynyl cyclopropenes43,44,45,46, but is limited to symmetrical 1,3-diynes. Therefore, the coupling of CpBX 1a bearing one ester group on the C3 position and phenylacetylene (2a) was selected as our prototypical system. A wide range of transition-metal catalysts and ligands were investigated, and the selected conditions are presented in Table 1 (a complete list of the conditions screened is provided in Supplementary Section 4). We identified the commercial complex (Me2S)AuCl as an effective transition-metal catalyst, together with ligand L1, to deliver the cross-coupled product 3a in 96% NMR yield in CH3CN at room temperature (r.t., entry 1). Without (Me2S)AuCl, no desired product was obtained (entry 2), and without L1, a drop in yield and a longer reaction time were observed (entry 3). Palladium or nickel, commonly used catalysts for C–C bond formation47, did not promote the desired transformation, even when a higher reaction temperature was applied, thus underscoring the specific role of gold in catalysing the σ-type CPC transfer reaction (entry 4). This result is noteworthy, given that gold(I) species are difficult to oxidize (Au(III)/Au(I) = 1.41 V) when compared to palladium(0) (Pd(II)/Pd(0) = 0.91 V) or nickel (0) (Ni(II)/Ni(0) = −0.24 V)48,49. Our findings further highlight the unique properties of hypervalent iodine reagents to promote redox-gold catalysis42,50,51,52. Increasing the amount of L1 to 25 mol% afforded almost the same yield of 3a as with 10 mol% (entry 5). PPh3-ligated gold catalysts, previously shown to be the optimal catalysts for alkynylation using EBXs53,54, were ineffective (entry 6). Other gold catalysts coordinated by strong σ-donating ligands all failed to give 3a (Supplementary Table 1). Electron-deficient L1 was found to be superior in accelerating the reaction than its more electron-rich analogues L2 or L3 (entries 7 and 8)55. The screening of solvents revealed that CH3CN was the optimal solvent compared with less polar ones, such as CH2Cl2 or tetrahydrofuran (THF; entries 9 and 10). A similar catalytic performance was also observed with AuCl as the catalyst, albeit a longer reaction time was necessary for the full conversion of 1a (entry 11). The slightly lower reaction rate observed when using AuCl as the catalyst might be attributed to a necessary dissociative process of the polymeric gold(I) source compared to monomeric (Me2S)AuCl. Replacing (Me2S)AuCl with AuCl3 resulted in a drop in the yield of 3a, which could be improved when increasing the temperature to 40 °C (entries 12 and 13). The sluggish performance for the reaction using AuCl3 as catalyst could be attributed to an additional induction period needed to generate the catalytically active Au(I) species56. As a control experiment, the use of cyclopropenyl iodide 1a-1 as the coupling partner did not yield any desired product under the standard conditions, highlighting the importance of the hypervalent iodine reagent for successful oxidative addition (entry 14). Although alkynyl silane 2a-157 (entry 15) and alkynyl pinacol boronate 2a-2 (entry 16) also gave good yields of 3a, other acetylide surrogates, such as potassium alkynyltrifluoroborate 2a-3 (entry 17) and alkynylgermane 2a-458 (entry 18) were less efficient. It is noteworthy that the iodoarene 4 obtained during the reaction can be recovered and recycled for the synthesis of hypervalent iodine precursors I1 or I2 (Supplementary Sections 5 and 6 provide details).

Table 1 Optimization of the gold-catalysed σ-type CPC transfer reaction

Scope of σ-type CPC transfer to terminal alkynes

With the optimal conditions in hand, we explored the scope and limitations of this σ-type CPC transfer reaction in terms of both functional-group compatibility and structural diversity. We first examined variation of the terminal alkyne component (Table 2). Our process worked well for terminal alkynes with aryl rings substituted with alkyl (3b), alkenyl (3c), phenyl (3d), CF3 (3e3f), nitro (3g), ester (3h), aldehyde (3i) and carboxylic acid (3j3k) groups in the para and meta positions. Electron-donating functionalities on the aryl substituent of the terminal alkyne, such as methoxy (3l3m), trifluoromethoxy (3n) and carbamate (3o), were well tolerated. Functionalities, such as halogens (-Cl, -Br and -I, 3p3r), which can react in the presence of many transition-metal catalysts, remained intact using our protocol, thus highlighting the orthogonal reactivity of gold over palladium or nickel catalysis and allowing the installation of halogen handles for further diversification. Intriguingly, terminal alkynes bearing functionalities such as aryl silane59 (3s), aryl germane60 (3t) and aryl boronate61 (3u), previously reported to be suitable coupling partners in redox gold catalysis, remained untouched, thus underscoring the chemoselectivity of the σ-type CPC transfer reaction for alkynes. Additionally, this gold-catalysed cross-coupling system was shown to tolerate polyaromatic or heteroaryl-substituted alkynes, as exemplified by the synthesis of alkynylcyclopropenes substituted with naphthalene (3v), phenanthrene (3w), thiophene (3x) and pyridine (3y) moieties. A conjugated enyne was also tolerated (3z). The use of aliphatic terminal alkyl alkynes as coupling partners was also successful and further illustrated the compatibility of the process with a broad range of functionalities, including alkyls (3aa–3ab), cyclic alkanes (3ac3ad), a chloride (3ae), a bromide (3af), an iodide (3ag), free alcohols (3ah, 3al), a silyl ether (3ai), a cyclic carbamate (3aj), an imide (3ak), a benzylic ether (3am), a phenyl ether (3an), an ester (3ao, 3ap) and a cyano (3aq) group. In particular, a propargyl benzoate, which is known to easily undergo a 1,2-migration using gold catalysis62, could also be accommodated, furnishing 3ao in 93% yield. Ethyl propiolate was also suitable for this reaction, giving 3ar in 84% yield. Acetylene gas itself, which is of particular interest owing to its availability in bulk quantity63, led to 3as in moderate yield, although a higher ligand loading and higher reaction temperature were required. Notably, 1,n-diynes tethered by an alkyl chain or an aromatic framework underwent smooth double cross-coupling in excellent yields (3at3av). The structure of 3au with two alkynylcyclopropene units attached to the para positions of benzene was confirmed by X-ray crystallography. Next, we examined the scope of CpBXs using silyl-substituted terminal alkynes as the representative coupling partners. CpBXs bearing different alkyl substituents (R1) attached to the cyclopropene were suitable reaction partners and furnished products 3aw and 3ax in >90% yield. CpBXs featuring various ester substituents provided products 3ay to 3bc in excellent yields. CpBX 1i bearing an additional alkyl substituent at the C3 position of the cyclopropenyl moiety underwent coupling with ethynyltriisopropylsilane to give tetrasubstituted alkynylcyclopropene 3bd in 89% yield. Remarkably, trifluoromethyl-substituted CpBX 1j could also be used in our gold-catalysed σ-type CPC transfer protocol to give 3be in 78% yield. Furthermore, CpBXs derived from cyclopropenes bearing two methyl ester substituents42,64 at the C3 position were also excellent substrates, as showcased by the formation of alkyl- (3bf), phenyl- (3bg), fluoroaryl- (3bh3bi) and bromoaryl- (3bj) substituted alkynylcyclopropenes. Other esters at the C3 position were also tolerated (3bk). For some substrates, the reaction temperature was increased to 40 °C to increase the reaction rate. To demonstrate the practical utility of our method, the synthesis of representative alkynylcyclopropenes 3bl, 3bm, 3bn and 3bo was performed with decreased catalyst loading (2 mol% (Me2S)AuCl and 4 mol% L1) on 2.0, 1.2, 0.6 and 1.8 mmol scales, respectively. The products were obtained in comparable yields, albeit with extended reaction times.

Table 2 Scope of the gold-catalysed σ-type CPC transfer to terminal alkynes

Late-stage functionalization65 has emerged as an appealing strategy for the identification of bioactive compounds and requires further extension of the boundaries of modern synthesis in its ability to build and tolerate molecular complexity. Pleasingly, late-stage modification of complex natural products modified with a propargylic alkyne handle, such as (−)-camphanic acid (3bp) and (−)-borneol (3bq), as well as biologically relevant molecules such as (l)-propargylglycine (3br), (l)-phenylalanine (3bs), α-(d)-allofuranose (3bt) and (d)-biotin (3bu) could be achieved efficiently, thus confirming the generality of our method (Table 3). We next evaluated a selection of drug derivatives, including sulbactam (3bv), fenofibric acid (3bw), ciprofibrate (3bx), naproxen (3by), oxaprozin (3bz), isoxepac (3ca), febuxostat (3cb), indomethacin (3cc), mestranol (3cd) and norethindrone (3ce), resulting in the formation of the corresponding modified drug molecules in 83–99% yield. Remarkably, the sensitive core heterocyclic fragments in ezetimibe (3cf), artesunate (3cg) and gibberellic acid (3ch) were also well tolerated in the cross-coupling.

Table 3 Late-stage functionalization of bioactive natural and synthetic molecules

Scope of σ-type CPC transfer to vinylboronic acids

To expand the generality of our method, other potential acceptors for σ-type CPCs were also examined. Following extensive screening of various sp2-hybridized coupling partners, we were pleased to find that vinylboronic acids (5) also participate in the σ-type CPCs transfer reaction. With only minor modification of the standard conditions (optimization details are provided in Supplementary Table 6), the scope of this σ-type CPC transfer reaction to vinylboronic acids was explored. As shown in Table 4, vinylboronic acids bearing aromatic rings with different electronic properties (6a6c) or a halogen substituent (6d) gave the desired products in 55–86% yield. Vinylboronic acids with an alkyl group (6e) or a benzyl group (6f) on the alkene were tolerated. Cyclic disubstituted-vinylboronic acids were also suitable substrates, furnishing the corresponding coupled products 6g6j in 53–62% yields. In addition, the scope of CpBXs was investigated with (E)-(4-methylstyryl)boronic acid 5b. To our delight, CpBXs 1 bearing various alkyl substituents on the ester underwent coupling smoothly with vinylboronic acid 5b to give vinyl–cyclopropenes 6k6o in 63–88% yield. The use of other nucleophilic partners such as allenamides or indoles was not successful (details are provided in Supplementary Section 6.3).

Table 4 Scope of the gold-catalysed σ-type CPC transfer to vinylboronic acids

Synthetic transformations

The alkynyl–cyclopropene products are versatile building blocks for the synthesis of substituted cyclopropenes, functionalized cyclopropanes or ring-opening products (Fig. 2). Selective reduction of the ester functionality in 3bm with diisobutylaluminium hydride (DIBAL-H)66 afforded hydroxymethylcyclopropene 7 in 94% yield. Alternatively, the alkene unit and the ester functionality in 3bn were both reduced when treated with LiAlH4 (ref. 67) to provide cyclopropane 10 in 32% yield with excellent diastereoselectivity. Moreover, 7 could serve as starting material for a copper-catalysed carbomagnesiation reaction proceeding in a regio- and diastereoselective manner68. The in situ-formed cyclopropyl metal species could be quenched by methanol and allyl bromide, affording polysubstituted cyclopropanes 8 and 9, respectively. Saponification of 3bm using sodium hydroxide furnished cyclopropene carboxylic acid 11 in 82% yield. Interestingly, 3bo could be readily converted into conjugated enyne 12 with excellent stereoselectivity in the presence of a cationic gold(I)–carbene complex69. Gold carbene 18, presumably generated via 1,2-benzoyloxy migration of 3bo, can be proposed as the key reactive intermediate, which then underwent ring-opening of the cyclopropene (a complete speculative mechanism is provided in Supplementary Fig. 1). A Diels–Alder reaction of 3bo with 2,3-dimethylbutadiene gave fused bicycle 13 in 89% yield and >20:1 diastereoselectivity (d.r.). Additionally, desilylation of 3bl using tetrabutylammonium fluoride (TBAF)70 allowed access to cyclopropene 14 bearing a terminal alkyne, which can itself serve as a suitable partner in the gold-catalysed σ-type CPC transfer reaction, affording non-symmetrical 1,2-bis-cyclopropenyl substituted alkyne 15. A gold-catalysed cross-coupling of 14 with hypervalent iodine reagent 19 (ref. 53) gave cyclopropenyl 1,3-diyne 16 in 86% yield. Finally, copper(I)-catalysed alkyne–azide cycloaddition71 of 14 and benzyl azide provided cyclopropenyl triazole 17 in 72% yield.

Fig. 2: Synthetic modifications of the alkynylcyclopropenes.
figure 2

The obtained alkynylcyclopropenes can be used as precursors for accessing different types of functionalized cyclopropenes, cyclopropanes or ring-opening products. Reduction of alkynyl–cyclopropene 3bm with DIBAL-H gave cyclopropene 7, which was converted into alkynyl cyclopropane carbinol 8 and 9 by copper-catalysed carbomagnesiation quenched by methanol and allyl bromide, respectively. Reduction of 3bn by LiAlH4 afforded cyclopropane 10. Saponification of 3bm using sodium hydroxide afforded cyclopropene 11. Gold(I)-catalysed ring-opening of 3bo via gold carbene 18 furnished 12. Diels–Alder reaction of 3bo with 2,3-dimethylbutadiene produced fused bicycle 13. Desilylation of 3bl using TBAF led to the formation of 14, which could be converted into cyclopropenes 15 and 16 by gold-catalysed cross-coupling with 1k and 19, respectively. Copper(I)-catalysed alkyne–azide cycloaddition of 14 and benzyl azide gave triazole 17. NaAsc, (+)-sodium l-ascorbate; IPr, 2,6-bis(diisopropylphenyl)imidazole-2-ylidene; DCE, 1,2-dichloroethane; Tf, trifluoromethanesulfonyl. Supplementary Section 7 provides all the experimental details.

Mechanistic investigations

To gain some insights into the reaction mechanism72,73, we first attempted to identify the active gold species at the start of the catalytic cycle. We prepared the ligand-free polymeric gold(I)–phenylacetylide 2074 and the cationic gold(I)–ethylene complex 2175 as potential gold sources (Fig. 3a). We first investigated the use of 5 mol% 20 in the cross-coupling of CpBX 1a and terminal alkyne 2a under the standard conditions. No cross-coupled product was observed (Fig. 3b, entry 2). Running the reaction at 40 °C for 24 h, only 7% yield of product 3a was observed with most 1a (87%) recovered (Fig. 3b, entry 3). These results do not support a catalytic cycle involving the direct oxidation of a gold(I)–acetylide complex by CpBX 1. When the cationic gold(I)–ethylene complex 21 was used, a 10% yield of 3a was observed at room temperature after 2 h (Fig. 3b, entry 4). The yield could be improved to 57% by running the reaction at 40 °C for 10 h with full conversion of 1a (Fig. 3b, entry 5), indicating that the cationic gold(I) species can be oxidized by CpBX 1. The poor catalytic performance of cationic gold(I) complex 21 under the standard conditions could be attributed to the formation of a catalytically inert gold(I)–acetylide in the presence of an excess of terminal alkyne (Supplementary Section 8.3 presents more details). In contrast, when 5 mol% of 21 and 5 mol% of NBu4Cl were used, 3a was obtained in 94% yield (Fig. 3b, entry 6), with an efficiency similar to the one observed under standard conditions (Fig. 3b, entry 1). A control experiment using NaHCO3 (3.0 equiv.) as an additive did not show any improvement (Fig. 3b, entry 7), supporting the important role of chloride. To support this hypothesis, other halogenide additives were examined. Bromide exhibited a similar effect on the reaction outcome (Fig. 3b, entry 8). In contrast, the use of fluoride and iodide nearly completely suppressed the formation of 3a (Fig. 3b, entries 9 and 10). A catalytically active tricoordinated gold(I) chloride species55, 22, undergoing oxidative addition onto CpBX 1 would be in accordance with these observations (Fig. 3c). The NMR spectra of 22 prepared independently by mixing an equimolar amount of 21 and NBu4Cl in CD2Cl2 showed symmetric ligand backbone signals, which were distinct from those of 21 or L1 (spectra details are provided in Supplementary Fig. 3). Species 22 is therefore proposed to be a fluxional species with fast exchange of the coordination sites of gold(I) to the two nitrogen atoms of L176. A weak coordination of the gold centre to CpBX 1 would afford transient π-type Au(I) species VI. Subsequently, oxidation of gold(I) to gold(III) with the concomitant transfer of the cyclopropenyl moiety from iodine to gold via a concerted four-membered ring transition state VII would generate the square-planar Au(III)–cyclopropenyl species VIII77. The highly electrophilic and reactive species VIII would capture terminal alkyne 2 via ligand exchange or undergo transmetallation with vinylboronic acid 5 to generate gold(III) species IX or X, which, upon reductive elimination, would yield cross-coupled products 3 or 6, respectively, and regenerate the gold(I) catalyst 22.

Fig. 3: Mechanistic studies on the gold-catalysed σ-type CPC transfer reaction.
figure 3

a, Preparation of gold(I)–phenylacetylide 20 and cationic gold(I)–ethylene complex 21. b, Control experiments for determining the catalytically active species. Gold(I)–acetylide complex 20 was not a competent catalyst. Chloride plays a crucial role in accelerating the coupling reaction. c, Proposed reaction mechanism. d, Reaction profile of the stoichiometric reaction of CpBX 1l and (Me2S)AuCl, with L1 as the ligand. e, ESI-MS/MS analysis of the stoichiometric reaction of CpBX 1l, (Me2S)AuCl and L1 supporting the formation of intermediate VIII. f, ESI-MS/MS analysis of the stoichiometric reaction of CpBX 1l, (Me2S)AuCl, L1 and alkyne 2n supporting the formation of intermediate IX. aSolid-state structure of 21, with thermal displacement ellipsoids given at 50% probability. The counterion is omitted for clarity. bDetermined by 1H NMR. cStandard condition (Table 1, entry 1). dReaction carried out at 40 °C for 24 h. eReaction carried out at 40 °C for 10 h. fNaHCO3 (3.0 equiv.) was used. g22 is proposed to be a fluxional species with fast exchange of the coordination sites of gold(I) to the two nitrogen atoms of L1. ESI, electrospray ionization.

Source data

We next performed further computations and mechanistic experiments to support the proposed catalytic cycle and the putative Au(III)–cyclopropenyl species. First, we used density functional theory (DFT) at the B3PW91-D3(BJ)/def2-TZVP//PBE0-D3(BJ)/def2-SVP level (Supplementary Section 8.8 and Supplementary Fig. 19) to further assess the feasibility of the key oxidative addition and reductive elimination steps. Activation energies of 26.9 kcal mol−1 and 9.4 kcal mol−1 were obtained for the oxidative addition and reductive elimination processes, respectively. This further supported that both steps were feasible, even if the energy for the oxidative addition step remains a little high when considering the reaction rate. We then used 19F NMR spectroscopy to monitor the stoichiometric reaction of an equimolar amount of CpBX 1l, (Me2S)AuCl and L1 in CD3CN at ambient temperature. As shown in Fig. 3d, CpBX 1l readily reacted with gold(I) and was completely consumed within 5 h, thus confirming the reactivity of CpBX 1 with gold(I) species. Intriguingly, the homo-coupled product 23 was formed, which may originate from the labile Au(III)–cyclopropenyl species 24 (vide infra). To further probe the intermediacy of the proposed Au(III)–cyclopropenyl species VIII, we then used electrospray ionization mass spectrometry (ESI-MS) techniques to monitor the reaction mixture. As shown in Fig. 3e, a reaction mixture of equimolar 1l, (Me2S)AuCl and L1 in CH3CN (at room temperature for 5 min) was subjected to high-resolution mass analysis. Although 24 was not observed directly by mass spectrometry due to its electroneutral nature, cationic Au(III) species 25 and 26 derived from 24 by losing one anionic fragment were both observed by ESI-MS and structurally confirmed by tandem mass spectrometry (MS/MS) (Supplementary Section 8.6). Interestingly, the cationic Au(III)-bis(cyclopropenyl) species 28, derived from 27 by losing chloride, was also observed, indicating the mechanism for the formation of 23, that is, ligand scrambling78 of 24 to 27 followed by reduction elimination to furnish 23. Finally, we sought to gain support for the putative Au(III) species IX, a key organogold species in the catalytic cycle to connect the transmetallation and reductive elimination step. As shown in Fig. 3f, a reaction mixture of equimolar 1l, (Me2S)AuCl, L1 and terminal alkyne 2n in CH3CN was subjected to high-resolution mass analysis. Gratifyingly, the expected cationic Au(III) species 30, derived from 29 by losing chloride, was observed by ESI-MS and further structurally determined by MS/MS analysis, thus providing direct evidence for the participation of Au(III)–cyclopropenyl species IX in the catalytic cycle. Overall, these experiments support the mechanism proposed in Fig. 3c well, even if it is not completely certain which of the chloride or cationic species is on or off the catalytic cycle.


In summary, we have developed broadly applicable synthetic equivalents of the elusive and untapped σ-type CPCs. The required CpBXs, newly designed iodine(III)-based precursors of σ-type CPCs, can be prepared from readily available reagents. Gold(I) complexes were used as catalysts for the intermolecular σ-type CPC transfer reaction of CpBXs 1 to terminal alkynes or vinylboronic acids under mild conditions, providing straightforward access to alkynyl–cyclopropenes and vinyl–cyclopropenes. The gold-catalysed protocol exhibited a broad substrate scope and tolerated numerous functional groups. This protocol can be further applied to the late-stage elaboration of complex organic compounds and drug molecules containing an alkyne handle. The alkynyl–cyclopropene products have been shown to be versatile synthetic intermediates for downstream diversification. Mechanistic studies support the intermediacy of a highly electrophilic cyclopropenyl–Au(III) species as a σ-type CPC equivalent and provide evidence for the crucial role of chloride as a supporting ligand for efficient coupling. Our work therefore substantially extends the chemical diversity of easily accessible cyclopropene building blocks, with applications in synthetic and medicinal chemistry, and will inspire other researchers in the design of new synthons based on the merger of hypervalent iodine reagents and redox gold catalysis.


General procedure for the synthesis of cyclopropenyl benziodoxoles 1 (CpBXs)

An oven-dried Schlenk tube was charged with a magnetic stirring bar and the terminal cyclopropene s-1 (4.00 mmol, 1.00 equiv.). The Schlenk tube was then evacuated and backfilled with nitrogen three times. THF (40 ml) was added by syringe and the Schlenk tube was placed at −78 °C in a dry ice/acetone bath. We then added n-butyllithium (2.5 M in hexane; typically, 4.2 mmol, 1.7 ml, 1.05 equiv.) dropwise via a syringe pump over 5 min, and the reaction mixture was stirred at −78 °C for an additional 10 min. Hypervalent iodine precursor I1 (typically, 4.40 mmol, 1.78 g, 1.10 equiv.) was added in one portion under nitrogen. The reaction mixture was stirred at −78 °C for 15 min, then the cooling bath was removed. The reaction mixture was allowed to warm to room temperature gradually (typically for ~15 min) while stirring. The reaction mixture was then quenched by adding saturated aqueous NaHCO3 (40 ml). The organic layer was removed, and the remaining aqueous portion was extracted with EtOAc (3 × 10 ml). The combined organic portions were dried over Na2SO4, filtered, and the volatiles removed under reduced pressure. The crude product was purified by flash chromatography on silica gel, and the fractions that contained the product were collected and concentrated by rotary evaporation to afford the purified compound.

General procedure for the synthesis of alkynyl–cyclopropenes 3

An oven-dried 10-ml Schlenk tube with a magnetic stirring bar was sequentially charged with L1 (4.20 mg, 20.0 μmol, 10.0 mol%), (Me2S)AuCl (2.95 mg, 10.0 μmol, 5.0 mol%), terminal alkyne 2 (200 μmol, 1.00 equiv.) and CpBX 1 (200 μmol, 1.00 equiv.). The Schlenk tube was then evacuated and backfilled with nitrogen three times. Subsequently, CH3CN (2.0 ml) was added by syringe. If 2 was a liquid, it was added last. The reaction mixture was stirred at room temperature (~21 °C) for the specified time. The reaction mixture was then filtered through a silica gel pad and washed with CH2Cl2 (3 × 5 ml). Excess solvent was removed under reduced pressure and the desired product 3 was obtained by column chromatography on silica gel.

General procedure for the synthesis of vinyl–cyclopropenes 6

An oven-dried 10-ml Schlenk tube with a magnetic stirring bar was sequentially charged with L1 (2.10 mg, 10.0 μmol, 10.0 mol%), (Me2S)AuCl (1.47 mg, 5.00 μmol, 5.0 mol%), vinylboronic acid 5 (100 μmol, 1.00 equiv.) and CpBX 1 (typically, 130 μmol, 1.30 equiv.). The Schlenk tube was then evacuated and backfilled with nitrogen three times. Subsequently, CH3CN (2.0 ml) was added by syringe. The reaction mixture was stirred at 40 °C for the specified time. The reaction mixture was then filtered through a silica gel pad and eluted with CH2Cl2 (3 × 5 ml). The solvent was removed under reduced pressure, and the resulting crude residue was subjected to a short column chromatography stage (silica). The fractions that contained the products were collected and analysed by 1H NMR spectroscopy. The recovered sample was purified by flash column chromatography (C18 reverse phase) to give cross-coupled product 6.