Introduction

Chiral organoboron compounds are acknowledged as exceptionally versatile building blocks in chemical synthesis, owing to their diverse reactivity profile1,2,3,4,5,6, and exhibit extensive applications in pharmaceuticals, materials science, and agrochemistry7,8,9,10,11,12,13. The quest for effective and selective protocols to prepare these valuable molecules has been a longstanding and challenging task in chemical synthesis. In this context, transition metal-catalyzed asymmetric hydroboration of unsaturated compounds, such as alkenes and alkynes, has emerged as a potent tool for constructing chiral organoboron compounds14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,32,33. Despite significant advances achieved in the asymmetric hydroboration of alkenes, the selective hydroboration of allenes with boranes to access chiral organoboron compounds poses a formidable challenge34,35,36,37. This is due to the potential generation of a series of organoboron products, resulting from diverse regio- and stereoselectivity issues in this reaction38,39,40. Consequently, the development of effective and selective protocols for the precise enantioselective synthesis of chiral organoboron compounds from readily accessible allenes and commercially available base metal catalysts remains significant and highly desirable.

The utility of allenes in the preparation of a diverse range of compounds has been widely explored41,42,43,44,45,46,47,48,49,50,51,52,53,54, highlighting exceptional versatility in chemical synthesis accentuated by the heightened reactivity of their two orthogonal cumulative π-systems. However, the intrinsic instability and elevated activity of allenes present challenges in the hydroboration process for constructing organoboron compounds, leading to intricate regional and stereoselectivity. This complexity contrasts with the hydroboration reactions of alkenes and alkynes, introducing the potential for over-borohydride during the reaction38,39. Over the last two decades, transition metal-catalyzed hydroboration of allenes with pinacolborane (HBpin) has been well developed by researchers, including Ma55,56,57,58, Ge59,60,61, Huang58, Tsuji62, Miyaura40, and Zhan63, providing various allylic boronate and alkenyl boronate derivatives with excellent regio- and stereoselectivities (Fig. 1a)40,55,56,57,58,59,60,61,62,63. Despite significant progress in the hydroboration of allenes, the enantioselective hydroboration of allenes is rarely developed. The pioneering work includes the hydroboration of allenes developed by Rush using stoichiometric chiral borane reagents to generate chiral allylic boranes, which then undergo allylboration with aldehydes to furnish homoallylic alcohols in high enantioselectivities (Fig. 1b)34,35,36,37. Moreover, the Hoveyda group disclosed a Cu-catalyzed asymmetric protoboration of allenes to provide chiral alkenylboron products in high stereoselectivity using chiral NHC–Cu complexes, B2(pin)2 as the boron source, and tBuOH as the proton source (Fig. 1c)64. However, the catalytic asymmetric direct hydroboration of allenes with borones is yet to be reported to our knowledge, emphasizing the urgent need for a more atom-economical and efficient method in the development of asymmetric hydroboration of allenes for obtaining chiral organoboron compounds.

Fig. 1: Hydroboration of allenes.
figure 1

a Catalytic hydroboration of allenes. b Hydroboration of allenes with chiral boranes. c Catalytic enantioselective protoboration of allenes. d Catalytic asymmetric dihydroboration of allenes enabled by ligand relay catalysis.

Full size image

The alkenyl boronate and allylic boronate derivatives obtained from the hydroboration of allenes through transition metal catalysis can potentially undergo the second hydroboration due to the presence of an unsaturated bond23,31,65,66,67,68. Intriguingly, the sequential dihydroboration of allenes by transition metal catalysis has not been reported to our knowledge (Fig. 1a), but it is highly anticipated, as the resulting dihydroboration products and their derivatives could serve as versatile precursors to various high-value chemical products23,69,70,71,72,73,74. We envision that the asymmetric dihydroboration of allenes represents an efficient method for producing chiral 1,n-diboronate products with high atomic and step economies. However, the challenge lies in designing a reaction system that can promote all steps of dihydroboration of allenes with simultaneous control of regio-, stereo-, and even enantioselectivity. Inspired by the ligand relay catalysis strategy in asymmetric multi-step reactions75,76,77,78,79,80,81,82,83,84,85,86,87,88, we hypothesized that the asymmetric dihydroboration of allenes could be realized with two simple ligands in a relay catalysis fashion. Starting with 1,1-disubstituted allenes, we expect one ligand to control the first hydroboration to selectively form achiral allylic boronate intermediates, and the other ligand to produce the desired chiral diboronates via the second asymmetric hydroboration (Fig. 1d). However, several challenges need to be contemplated, such as the need for well-controlled regioselectivity in the first hydroboration step and excellent stereoselectivity in the second asymmetric hydroboration step. Additionally, each ligand has to recognize two similar hydroboration processes to avoid excessive reactions and cross-side reactions.

Here we have successfully developed a regio-, stereo-, and enantioselective sequential hydroboration/ isomerization/asymmetric hydroboration of 1,1-disubstituted allenes using a ligand relay catalysis strategy, resulting in diverse chiral 1,4-diboronate products with high yields and enantiomeric excess values (Fig. 1d).

Results

We started our investigation with buta-2,3-dien-2-ylbenzene 1a as the model substrate and HBpin as the boron source to explore the asymmetric dihydroboration through the ligand relay catalysis strategy. Based on our previous research on the cobalt-catalyzed enantioselective hydroboration and isomerization/hydroboration of alkenes89,90,91, we selected cobalt as the catalyst. As shown in Table 1, the desired 1,4-diol product 2a was obtained in 40% yield with 82% ee when using an achiral N,N,N-tridentate ligand terpyridine (tpy, for the first hydroboration) and a chiral oxazoline iminopyridine ligand L*1 (for the second hydroboration) with Co(acac)2 in MTBE at 30 °C for 30 h (entry 1). Encouraged by this promising result, various chiral oxazoline iminopyridine ligands with different substituted oxazolines were initially examined, revealing that the steric hindrance of the substituent R’ significantly influences the yield and the enantiomeric excess. Gratifyingly, the product 2a was isolated in 79% yield with 95% ee using the sterically hindered chiral oxazoline iminopyridine ligand L*4 (entry 4), and extending the reaction time slightly increased the yield to 82% with 94% ee (entry 5). The less bulky oxazoline iminopyridine ligands L*1–L*3, compared to L*4, resulted in both lower yields and enantiomeric excesses (entries 1–3). Unfortunately, the chiral ligand L*5 was ineffective in this catalytic system (entry 6). Evaluation of other achiral tridentate nitrogen ligands showed that the phenyl substituent with an electron-donating group (L1) gave the product 2a in 42% yield with 94% ee, and the one with an electron-withdrawing group (L2) provided 2a in 11% yield with 70% ee (entries 7-8). As expected, no desired product 2a was observed when bipyridine was applied as the ligand (entry 9). The above experimental results demonstrated that both achiral and chiral ligands are necessary to control the enantio-, stereo-, and regioselectivities of the asymmetric dihydroboration of allenes. Other cobalt catalysts such as Co(OAc)2 and Co(acac)3 were also investigated, leading to lower efficiency (entries 10–11). Finally, solvents were optimized. When Et2O and THF were used as the solvent, the diol 2a was obtained in 76% yield with 88% ee and 27% yield with 79% ee, respectively (entries 12–13). The nonpolar solvent hexane was also screened, providing product 2a in 20% yield with 29% ee (entry 14).

Table 1 Reaction condition optimizationaView full size image
Full size table

With the optimal reaction conditions in hand, we next explored the substrate scope of allenes in this regio-, stereo-, and enantioselective dihydroboration reaction. As shown in Fig. 2, a wide range of allenes reacted smoothly to afford the corresponding products in good yields and excellent enantioselectivities. A broad range of substituents at the para- position including electron-donating and electron-withdrawing groups, such as -Me, -iBu, -OMe, -OPh, -OBn, -OBz, -OTBS, -CH2OH, -SMe, morpholinyl, -TMS and -F were well compatible, providing the diol products 2b-m in good yields and excellent enantioselectivities. We were pleased to find that the mild reaction conditions tolerated the free hydroxyl group (2i) and the saturated nitrogen heterocycle morpholine (2k), which afforded the desired products in 76% yield with 93% ee and 68% yield with 96% ee, respectively. Additionally, good reactivities were observed with a variety of substituents at meta-position such as -F, -CF3, -Ph, Me, -OMe, -OBn, and acetal, delivering the corresponding 1,4-diols 2n–t in 58–85% yields with 91–95% ee. Furthermore, the allenes with a di- and tri-substituted phenyl group were competent substrates, affording the products 2u and 2v in 67% yield with 95% ee and 82% yield with 90% ee, respectively. This reaction was also applicable to benzo rings such as benzo-1,4-dioxane, leading to the formation of the diol product 2w in moderate yield and high enantioselectivity. In addition to phenyl rings, polycyclic aromatic and heterocyclic motifs such as fluorene (2x), xanthene (2y), and naphthalene (2z) can be readily incorporated into the products 50–76% yields with 72–95% ee. A single recrystallization and removal of the mother liquor by filtration gave the white crystal products 2x and 2z in 90% ee and 88% ee, respectively. We envisioned that the lower optical purity of products 2x and 2z is attributed to the participation of the achiral ligand tpy in the subsequent isomerization/hydroboration after the first hydroboration step to produce a partial racemization products 2x and 2z. Likewise, a series of heteroaromatic substrates, such as those containing furan (2aa, 2ac), thiophene (2ab), indole (2ad), pyridine (2ae'), and pyrrole (2af) were proved to be compatible, delivering the corresponding products in 25–77% yield with 80–96% ee. Of particular note is that the substrates derived from natural products, such as(−)-menthol (2ag) and estrone (2ah), were shown to be well tolerated up to 63% yield with excellent diastereoselectivities with great potential in organic synthesis. Unfortunately, the ethylated, propylated, silylated, and dialkyl allenes failed to provide the desired products (for more details in Supplementary Table 2).

Fig. 2: Scope of allenes.
figure 2

aUnless otherwise noted the reactions were carried out with 1 (0.3 mmol), Co(acac)2 (10 mol%), tpy (5 mol%), L*4 (5 mol%), HBpin (3.5 equiv), MTBE (1.5 mL), 30 oC, 39 h, then oxidized by H2O2 (30%, 0.6 mL), NaOH (3 M, 0.6 mL) in THF (2 ml); Isolated yields. Ee was determined by HPLC on a chiral stationary phase. bWithout oxidation. cNaBO3•4H2O was used as an oxidant. dHBpin (4.5 equiv). eThe reaction was conducted at 20 °C. fEnantiomeric excess after recrystallization in the parentheses.

Full size image

The synthetic utilities of this method were then investigated. First, a gram-scale reaction of 1a was carried out to give the 1,4-diboronate product 2a’ in 1.38 g and 72% yield with 94% ee, which was used for further transformations (Fig. 3a). Initially, product 2a’ was reacted with vinyl magnesuim bromide in the presence of I2 to afford 4-vinyl borylalkane 3 in 52% yield and 93% ee92, which can go further both carbon–carbon double bond functionalization and C–B bond transformation to give more complex chiral fragments. 1,6-Diene 4 can also be formed in 57% yield with 94% ee by increasing the equivalents of vinyl magnesuim bromide and I292. The coupling of furan with 2a’ led to the formation of 5 in 64% yield with 94% ee93. The chiral diamine 6 was obtained in 52% yield and 94% ee via a Cu-catalyzed cross-coupling reaction94. In addition, Pd-catalyzed Suzuki-Miyaura coupling of 2a’ with 6-bromo-1-benzothiophene allowed the synthesis of product 7 in good yield and excellent enantioselectivity95, which can be used for downstream diversification. The synthetic utility of this method was further highlighted by the synthesis of a bioactive molecule. Starting from allene 1d, the desired chiral 1,4-diol (S)-2d was isolated in 58% yield with 95% ee under the standard reaction conditions with (R)-L*4 as the ligand. The Ru-catalyzed oxidation of (S)-2d provided the chiral lactone 996. Subsequent treatment of 9 with LDA followed by trapping with the electrophile m-OMeC6H4CH2Br gave the artigenin analog 10 in 48% yield and 95% ee (Fig. 3b)97.

Fig. 3: Synthetic transformations and applications.
figure 3

a Gram-scale reaction and synthetic transformations. b Artigenin analog synthesis.

Full size image

To gain more insight into the reaction mechanism, we carried out a series of control experiments (Fig. 4a). Initially, we explored the first hydroboration step of allene 1a in the presence of tpy, which gave E-allylic boronate 11 in 87% yield after 1.5 h (1.5 equiv HBpin, Fig. 4a, i), and extending the reaction time to 39 h resulted in the formation of dihydroboration product 2a in 44% yield (3.5 equiv HBpin, Fig. 4a, ii). Standing in sharp contrast with tpy, when L*4 was alternatively used as the sole ligand, a trace amount of product 11 was detected and 85% of 1a was recovered in 1.5 h (1.5 equiv HBpin, Fig. 4a, iii). The extension of reaction time (39 h) did not improve the efficiency, furnishing the 2a in 17% yield with 94% ee (3.5 equiv HBpin, Fig. 4a, iv). These results indicate that the tpy is an effective ligand for both the first and second hydroboration steps (Fig. 4a, i-ii), and is essential for the high regio- and stereoselectivity in the first hydroboration step. Moreover, compared with the tpy, the L*4 showed much lower reactivity for the first hydroboration step (Fig. 4a, i vs iii). Based on the above results, we hypothesized that E-allylic boronate 11 might be the key intermediate in this asymmetric dihydroboration reaction. The E-allylic boronate 11 was then synthesized and subjected to the standard conditions (2.0 equiv HBpin), which provided the desired diol product 2a in 68% yield with 61% ee (Fig. 4a, v). In contrast, Z-allylic boranate 12 failed to give the 2a’ under the standard conditions (Fig. 4a, vi). These results suggested that the E-allylic boronate 11 was the intermediate and responsible for the second hydroboration. Moreover, we investigated the reactivities of L*4 and tpy in the second isomerization/hydroboration step. Not surprisingly, it was found that the addition of L*4 as a ligand delivered the 2a in 85% yield with 95% ee (Fig. 4a, vii), whereas only 42% yield of 2a’ was obtained with the use of tpy instead (Fig. 4a, viii). Time course studies of the isomerization/hydroboration of compound 11 with ligand L*4 and tpy prove that the reaction rate of L*4 was much faster than that of tpy (for more details in Supplementary Fig. 2). The proposed intermediate E was also synthesized and subjected to the standard conditions with 2.0 equiv HBpin (Fig. 4a, ix), which afforded the product 2a′ in 67% yield and 33% ee along with allylic boronate 11 in 14% yield. When L*4 was alternatively used as the sole ligand, the reaction gave 2a′ in 99% yield and 95% ee. However, the use of terpyridyl ligand delivered the 1,4-diboronate 2a′ and allylic boronate 11 in 51% and 23% yields, respectively. Time course studies showed that substrate 1a was consumed in approximately one minute along with the formation of E-allylic boronate 11 intermediate (Fig. 4b). These results indicated that 1) the ligands tpy and L*4 are responsible for the first and second hydroboration respectively, 2) although the formation of D via the addition of Co to the more hindered position is thermodynamically unfavored, the subsequent β-H elimination and the hydroboration could provide the driving force necessary to compensate for the endergonic insertion of the intermediate 11 into the Co-H (Fig. 5).

Fig. 4: Mechanistic studies.
figure 4

a Control experiments. b Time course studies. c Deuterium labeling experiments. d Crossover experiments.

Full size image
Fig. 5
figure 5

Proposed catalytic cycle.

Full size image

Deuterium-labeling experiments were conducted as well (Fig. 4c). When 1a was treated with DBpin (1.5 equiv) for 1.5 h, the corresponding product 13-d1 was afforded in 92% yield (E/Z = 13:1) with 37% D atom incorporations at the C1 position. Previous studies suggest that the hydroboration reaction of allenes proceeds through a Co–H intermediate58, which is consistent with this deuterium-labeling experiment. In addition, we carried out the reaction of 1a with DBpin (3.5 equiv) under the standard conditions, and the corresponding product 2a-d1 was gained in 86% yield with 38%, 26%, 42% D atom incorporations observed at C1–C3 positions. Similarly, D atom incorporations observed at the positions of C1–C3 were 29%, 13%, and 33%, respectively when deuterated substrate 1a-d1 was treated with the standard conditions. These results are consistent with an alkene isomerization mechanism98,99,100. Furthermore, we performed a crossover experiment using a mixture of deuterated 1a-d1 and 1v and found that the H/D scrambled products could be isolated with 86% (2a-d3) and 81% (2v-d) yields, respectively (Fig. 4d), suggesting that a dissociative mechanism was involved in the isomerization.

Based on the above mechanistic experiments and previous literature40,55,56,57,58,59,60,61,62,63,98,99,100,101,102,103, a possible catalytic cycle is proposed as shown in Fig. 5. The cobalt hydride species A can be obtained from the reaction of Co(acac)2 with tpy and HBpin. The allene 1a coordinates with species A, followed by the insertion of 1a into the cobalt hydride to deliver π-allyl-cobalt species C, which then reacts with HBpin via σ-bond metathesis or oxidative addition/reductive elimination to give E-allylic boronate intermediate 11 and regenerating the cobalt hydride species A. The intermediate 11 next goes through the migratory insertion into the cobalt hydride species A’ to afford intermediate D. The following β-H elimination and subsequent migratory insertion generate the alkyl Co species F, which reacts with HBpin to release the final product 2a’ and regeneratethe cobalt hydride species A’.

Discussion

In summary, we have developed a sequential hydroboration/isomerization/asymmetric hydroboration of allenes by ligand relay catalysis, which provides a robust and reliable method for the synthesis of chiral 1,4-diboronate products in good yields and excellent enantioselectivities. This protocol overcame the challenges of stereoselectivity, regioselectivity, and enantioselectivity. The robustness and practicability of this reaction were demonstrated by gram-scale reactions, diverse product derivatizations, and asymmetric syntheses of artigenin analogs. Preliminary mechanistic studies revealed that the achiral ligand was used for the first hydroboration of allenes to deliver E-allylic boranate intermediates with excellent regio- and stereoselectivities, and the subsequent isomerization/asymmetric hydroboration was catalyzed by the Co/chiral oxazoline iminopyridine ligand complex, providing the valuable diboronate products in high enantioselectivities. Additionally, the reaction involves a dissociative alkene isomerization and a cobalt-catalyzed asymmetric hydroboration. This strategy increased the utilization of unsaturated bonds of allenes and enriched the chemistry of allenes.

Methods

In a glove box, to an oven-dried 10-mL vial were added Co(acac)2 (7.8 mg, 0.03 mmol), tpy (3.5 mg, 0.015 mmol), L*4 (6.1 mg, 0.015 mmol) and anhydrous MTBE (1.5 mL). The resulting suspension was stirred for 30 min at room temperature, at which time HBpin (135 mg, 1.05 mmol) was added. After the resulting mixture was stirred for an additional 5 min, allene 1 (0.3 mmol) was then added. The reaction was stirred for 39 h at 30 °C. The resulting solution was quenched with DCM and concentrated. The residue was dissolved in THF (2.0 mL) and cooled to 0 °C, and NaOH (0.6 mL, 3.0 M) and H2O2 (0.6 mL, 30%) were then added. The resulting mixture was stirred for 30 min, then extracted with EtOAc (three times), dried with Na2SO4, filtered and concentrated. The residue was purified by silica gel chromatography to afford the desired products (R)-diol 2.