Introduction

Selective cleavage and subsequent functionalization of C−C single bonds present a fundamental challenge in catalysis and synthesis1,2,3,4,5,6,7,8,9,10,11. This is mainly due to the relatively high bonding energy (BE, about 355 kJ/mol) and directional σ orbitals of C−C bonds. In addition, the competitive C−H bond activation (about 400 kJ/mol, but statistically abundant) also causes chemoselectivity problems12,13,14,15,16,17,18,19,20,21,22. Nevertheless, cleavage and functionalization of C−C single bonds are attracting increasing attention in synthetic organic chemistry because it offers a unique and straight route to target molecules/structures. Synthetic chemists have developed various strategies for the activation of C−C single bonds. They are mainly classified into two mechanistic categories: oxidative addition and β-carbon elimination, associated with metal centers (Fig. 1a). Besides the use of stoichiometric transition-metal complexes23,24,25,26, examples of catalytic processes have been reported in recent years—most involve the use of precious metals27,28,29,30,31,32,33,34,35,36,37,38,39,40,41,42,43,44,45.

Fig. 1: Strategies for C−C single bond activation.
figure 1

a Transition metal mediated C−C single bond activation; b, c stoichiometric amount of zirconium-mediated C−C single bond activation; d zirconium and hafnium catalyzed C−C single bond activation.

Early-transition metals have different electron configurations from late ones. Thus, their complexes often show other or orthogonal reactivities with late-transition metal complexes46,47,48,49,50,51,52,53,54,55,56,57,58,59,60,61. In addition, they are also Earth-abundant (e.g., zirconium is almost as abundant as carbon in the Earth’s upper continental crust). However, zirconium usually exists in the ZrIV oxidation state, which is not viable for direct oxidative addition activation of a single chemical bond. Thus, in situ generated or isolable low-valent zirconocene complexes (ZrII), such as Negishi’s (Cp2ZrBu2)62 and Rosenthal’s reagents (Cp2Zr(py)Me3SiC≡CSiMe3)63, for the activation of B−H64,65, Si−H66,67,68,69 bonds were reported. C−C single bond cleavage by a zirconium species has also been reported intermittently since the 1990s70,71,72,73. In 1994, Rosenthal described the activation of conjugated C−C single bonds of a 1,3-butadiyne moiety (C≡C−C≡C) using Rosenthal’s reagent, resulting in a dimeric complex (Fig. 1b)74. Dimmock and Whitby also found that zirconocene η2-alkene and η2-imine complexes with adjacent cyclopropane rings could undergo cyclopropane ring cleavage75. Then, in 2014, Marek reported an expedient approach, including allylic C−H activations followed by C−C single bond activation (Fig. 1c)76. It is also worth mentioning that since the 1990s, Negishi, Takahashi, and Xi have studied the chemistry of zirconacycles, the transformation of which with other unsaturated molecules usually involved a β, β’-C−C bond cleavage77,78,79,80,81. All the former instances used (over) stoichiometric amounts of zirconium, and no precedents of catalytic methods using homogeneous zirconium catalysis had been developed—to the best of our knowledge (The use of heterogeneous zirconium catalysis for C−C bond cleavage was reported by Basset82,83,84). Consequently, activating C−C single bonds with zirconium catalysis for chemical transformation remains a significant challenge. It is of considerable scientific and practical interest to synthetic organic chemistry to address this. Herein, we report the development of an unprecedented catalytic system that resulted in the realization of the zirconium- and hafnium-catalyzed C−C single-bond hydroboration (Fig. 1d). Mechanistic studies support the formation of N−ZrIV species and then a β-carbon elimination route for C−C single bond activation. Our work provides an alternative catalytic method for C−C single bonds hydroboration, and establishes the bond activation models and catalytic application of group IV transition metals.

Results

Catalytic reaction investigations

As a synthetically significant transformation in organic chemistry, hydroboration of C=C bonds is well-studied. However, catalytic hydroboration of C−C single bonds remains underdeveloped. Only two systems using Ir and Rh are known for the hydroboration of cyclopropanes, as developed by Yamaguchi85 and Shi86,87. Besides making use of precious metals and N- or P-ligands, it is noticed that for the Ir system, the careful choice of a chiral tBu-Quinox ligand is crucial for the C−C bond hydroboration over the C−H boration88,89,90. For the Rh system, the PPh3 ligand is essential in inhibiting side reactions such as the formation of alkenes. Furthermore, in the former case, cleavage of Cβ−Cβ‘ bond is observed, while the latter cleavages Cα−Cβ bond. Thus, it is still highly desirable to develop a facile and inexpensive catalytic system for the hydroboration of C−C single bonds.

We commenced our investigation using 0.2 mmol of N-Piv-cyclopropylamines (1a) with 1.5 equiv. pinacolborane (HBpin) in 1 mL toluene at 120 °C as the model reaction, using 5 mol% Cp2ZrCl2 as catalyst (Table 1). Our preliminary investigations unveiled that the addition of 1 equiv. of base is the key for the zirconium-catalyzed C−C bond hydroboration (Supplementary Table 1); K2CO3 was the optimal choice (Table 1, entry 1, 81% yield of 2a). Various other zirconium complexes were then tested but were unsuccessful. There was no reaction with the sterically bulkier Cp*2ZrCl2 (Table 1, entry 2). With Cp2ZrHCl as catalyst, a 50% yield of 2a was obtained, whereas with Cp2ZrMe2 only 10% 2a was produced (Table 1, entries 3 and 4). It was evident that without Cp2ZrCl2 or K2CO3 no reaction proceeded (Table 1, entries 5 and 6). Interestingly, using Cp2TiCl2 instead of Cp2ZrCl2 also gave no reaction (Table 1, entry 7). An attempt was then made to reduce the amount of K2CO3. Pleasingly, even with 0.1 equiv. K2CO3 already had a 68% yield of 2a (Table 1, entry 8). Finally, a much higher yield of 2a (91%) was obtained with just 0.3 equiv. K2CO3 (Table 1, entry 9). In addition, 95% yield of 2a was obtained using 5 mol% Cp2ZrH2 as catalyst without K2CO3 (Table 1, entry 10). However, considering the simplicity of using readily available and inexpensive Cp2ZrCl2 as the catalyst, we conducted the following studies using Cp2ZrCl2/K2CO3 system (Table 1, entry 9). The results with Cp2ZrH2 gave us some clues for the subsequent mechanism studies (vide infra).

Table 1 Zr-catalyzed hydroboration of cyclopropylamines—condition optimizationa

Substrates scope studies

Having the reaction conditions for the Zr-catalyzed hydroboration of cyclopropylamines in hand (Table 1, entry 9), various cyclopropane rings were investigated to establish the generality of our methodology (Fig. 2).

Fig. 2: Zirconium-catalyzed hydroboration of cyclopropylamines—generality studies.
figure 2

a,baReaction conditions: 0.2 mmol 1, HBpin (1.5 equiv.), K2CO3 (30 mol%), Cp2ZrCl2 (5 mol%) and 1 mL toluene in a 15 mL pressure tube at 120 °C for 24 h; bIsolated yields are given; c130 °C for 24 h; d150 °C for 24 h; eHBpin (2.0 equiv.), K2CO3 (60 mol%), Cp2ZrCl2 (10 mol%) at 150 °C for 24 h; fHBpin (2.0 equiv.), Cp2ZrH2 (5 mol%) at 150 °C for 24 h.

First, the tolerance of substituents on the para-position of the phenyl ring was studied. We found that electron-neutral, electron-donating, and electron-withdrawing groups are tolerated; moderate to good yields were obtained (2a-2k, 40-82% yields). In general, electron-donating groups (−Me, −OMe, −tBu, −SMe, 2c2e, 2k) gave better results than electron-withdrawing groups such as −CF3 (2f). Halide substituents −F, −Cl, −Br, which likely undergo competitive hydrodehalogenation or boration reactions, are untouched in our system (2g2i). We found that steric effects have some influence on the results, as changing the substituents from the para-position to the meta- and ortho-position, led to slightly decreased yields or the need for higher reaction temperatures (2l2o). Significantly, naphthyl, benzodioxole, alkyne, and heteroaromatic rings such as furyl and thiophene substituents are all compatible in our system, with yields in the range of 71 − 78% (2p2t). Similarly, good results were obtained when the phenyl group is at the β’ position (2u). Changing R1 from an aryl to an alkyl group was also successful, with both acyclic and cyclic alkyl substituents (up to 77% yield, 2v2ac). Pleasingly, products 2q, 2ab and 2ac were obtained in 69%, 77%, and 67% yields, respectively, with no double or triple bond interference. Moreover, cyclopropane rings with two substituents on the R1 and R2 positions were also suitable (2ad2ae).

The effect of the substituent on the amide groups was then studied. When R is 1-methylcyclohexyl, the hydroboration product 2af was obtained in 76% yield. Changing R to a sterically bulkier adamantly group resulted in a slightly lower yield (2ag, 56%). Substrates with 2,2-dimethylbutyl and 1-methylcyclopropyl, and substituents containing chloride are all converted to their corresponding hydroboration products 2ah2aj in yields of up to 64%. Sulfonamide is also tolerated in our system, which get the hydroboration product 2ak in 66% yield. The reaction also proceeded with thioamide (2al). Finally, we found that cyclopropylamines derived from Oleanolic Acid and Gemfibrozil also reacted well in our system; the corresponding products, 2am and 2an, were obtained in yields of 49% and 59%, respectively.

Hafnium-catalyzed reaction

Compared with zirconium, hafnium has received less attention as a homogeneous catalyst in organic reactions. To our knowledge, reactivity toward C−C single bonds activation is also unknown. After successfully establishing zirconium-catalyzed C−C single bond activation of cyclopropylamines and their subsequent hydroboration, we further explored the reactivity of a hafnium complex towards C−C single bonds. We established that the base plays an essential role in tuning the reactivity. Eventually, Cs2CO3 was found to be the optimal base (Supplementary Table 5). Then, under the optimal reaction conditions, we conducted substrate scope generality studies (Fig. 3). We found that the hafnium system is not only suitable for substrates that work in the zirconium system but also for substrates that do not work there. For example, substrates with a −CN group do not react with the zirconium catalyst, but a 50% yield of product 2ao was obtained with hafnium. Product 2ap, with two fluorides on the phenyl ring, was also obtained in 51% yield. Additionally, cyclohexyl- (2aq) and phenyl-substituted substrates (2ar) were also applicable in the hafnium system.

Fig. 3: Hafnium-catalyzed hydroboration of cyclopropylamines—generality studies.
figure 3

a,b,c aReaction conditions: 0.2 mmol 1, HBpin (2.0 equiv.), Cs2CO3 (30 mol%), Cp2HfCl2 (5 mol%) and 1 mL toluene in a 15 mL pressure tube at 150 °C for 24 h; bHBpin (3.0 equiv.), Cs2CO3 (60 mol%), Cp2HfCl2 (10 mol%) and 1 mL toluene in a 15 mL pressure tube at 150 °C for 72 h; cIsolated yields are given.

Synthetic derivation

The practical utilization of our system was then demonstrated on a gram scale (Fig. 4). When we subjected 10 mmol of 1a to our standard reaction conditions, we obtained 2a in 73% yield (1.97 g). Furthermore, the synthetic derivatization of 2a was demonstrated. Using the aminoazanium of DABCO as an amination reagent91, and then protecting the amine with TFAA, the corresponding TFA-amide 3a was obtained in 66% yield. γ-Boronated amine 3b, which is otherwise difficult to obtain92, was obtained in 72% yield by reducing the amide functional group to an amine and then protecting it with TsCl. Product 2a can be transformed into potassium trifluoroborate salt 3c using KHF2 (82% yield). Treating 2a with furan-2-yllithium followed by NBS afforded the arylated product 3d in 74% yield. Finally, Pd-catalyzed Suzuki−Miyaura coupling of 2a with Estrone-derived triflate gave 3e in 45% yield.

Fig. 4: Synthetic Applications.
figure 4

Reaction in gram scale and further derivatization of 2a. Isolated yields are given; for detailed reaction conditions, please refer to the supplementary information.

Mechanistic studies

To shed light on the reaction mechanism, several control experiments were performed (Fig. 5). The possible formation of an alkene intermediate via ring-opening of cyclopropanes followed by hydroboration was studied. However, no alkenes were detected after 3 or 12 h under standard reaction conditions with or without HBpin (Fig. 5Aa, Supplementary Fig. 1). Utilization of alkenes 1a’ and 1a” afforded less than 6% 2a (Fig. 5Ab). When enantioenriched substrate (1S, 2R)-1as was applied, the desired product (R)-2as was obtained without erosion of the enantioselectivities (Fig. 5Ac, Supplementary Figs. 2, 3). Those results excluded a consecutive cyclopropane ring opening-hydroboration process. Then, the possibility of a reaction pathway that involved a radical species was investigated. TEMPO (2,2,6,6-tetramethylpiperidinyloxyl) (1−2 equiv) had almost no effect on the results. However, upon increasing the amount thereof (4 equiv.) the yields of 2a decreased to 44% (Fig. 5B). At this point, it should be borne in mind that TEMPO inhibition experiments can sometimes provide ambiguous results93. Thus, additional experiments with the addition of 9,10-dihydroanthracene (DHA) were conducted; no effect on the yield of 2a was detected (Fig. 5B). The results with TEMPO and DHA excluded a radical mechanism.

Fig. 5: Experimental mechanism studies.
figure 5

A possible alkene intermediate formation; B possible radical pathway; C 1H NMR spectra show the formation of Zr−H species; D 1H NMR spectra show the release of H2 or HD by reacting Cp2ZrHCl with 1a; E Control experiments to show the importance of N−H bond; F Detection of imide intermediate 2u’ by crude 13C NMR spectroscopy; G Deuterium labeling experiment by introducing D2 in the standard reaction.

Then, the active zirconium catalytic species was studied. Upon the combination of Cp2ZrCl2 and K2CO3 in d8-Tol heated at 120 °C for 12 h, a new species appeared around 6.0 ppm in the 1H NMR spectrum (Fig. 5Ca). With 2 equiv. K2CO3 and heating for a longer reaction time, the Cp2ZrCl2 was fully converted to this new species (Fig. 5Cb). Then, the isolated new species was characterized by IR spectroscopy and was currently assigned to Cp2ZrCO3 by comparation with literature data (Supplementary Fig. 4)94,95. Nevertheless, upon further adding HBpin to the above solution, we could detect the formation of Zr–H species in the 1H NMR spectrum by trapping with acetone (Fig. 5Cc, Supplementary Fig. 5). This finding, together with the fact that Cp2ZrHCl or Cp2ZrH2 can catalyze the C−C bond hydroboration process without K2CO3 (65% and 95% yields, Supplementary Table 2), we concluded that Zr−H species are essentially the active catalysts via the consecutive reactions of Cp2ZrCl2, K2CO3, and HBpin (Fig. 5Cd). According to the work from Ganem96, Rosenthal97, and Cantat98, the active Zr−H species can interact with 1a to form N−Zr species via metathesis with N−H bond. This is further proved in our case that Cp2ZrHCl reacts with the N−H group of 1a with the release of H2 or HD when 1a-D was used (Fig. 5D, Supplementary Fig. 6). To add further proof of the importance of the N−H, N-methylated analog substrate 1a-Me, and replace the N−H with CH2 or O substrates 1a-C, 1a-O were subjected to our reaction conditions. As expected, no corresponding C−C bond hydroboration product were observed (Fig. 5E). Keep in mind that β-carbon elimination is one of the main pathway for C−C bond cleavage. It is natural to think that after the formation of N−Zr species, a β-carbon elimination may proceed to cleavage the C−C bond to produce an imino propyl zirconium species. This is consistent with the fact that we can observe the presence of the putative imine intermediate both on GC/MS and 13C NMR when substrate 1u was used (Fig. 5F, Supplementary Figs. 7, 8).

Based on the above mechanistic study and our DFT calculation results (Supplementary Fig. 9), we conclude the following general reaction pathway for our Cp2ZrCl2/K2CO3 system (Fig. 6). First, the in-situ formed Zr−H species reacts with N−H bonds of the substrates to form N−ZrIV species A via H2 release (i). Next, the C−C single bond is cleaved via β-carbon elimination of intermediate A to form the imino propyl zirconium species B (ii). Subsequently, intermediate B reacts with HBpin via C−Zr and H−B bond metathesis to give C and regenerate Zr−H species (iii). In the second catalytic cycle, Zr−H hydride transfer to intermediate C gives intermediate D (iv), which is further reduced by the previously released H2 to 2a with hydrogenolysis or H2 metathesis (v). The last step is supported by the experiment that when we introduced 1 atm of deuterium gas into the standard reaction, 29% deuterium labeling at the α-carbon adjacent to N−H of 2a could be obtained (Fig. 5G, Supplementary Fig. 10), suggesting that hydrogen metathesis occurred99,100,101.

Fig. 6: Proposed reaction pathway for the Zr-catalyzed C−C single bond hydroboration.
figure 6

Key steps for the transformation: i) N−H bond metathesis; ii) C−C activation; iii) B−H bond metathesis, iv) Zr−H hydride transfer; v) H2 metathesis.

In summary, an unprecedented zirconium- and hafnium-catalyzed C−C single bond activation and subsequent hydroboration is realized using a catalytic system based on Cp2ZrCl2 and Cp2HfCl2. Our catalytic approach applies to various cyclopropylamines. Selective cleavage of the proximal Cα−Cβ single bond was achieved, with the tolerance of multiple functional groups as well as bio- and medicine-derived substrates. Mechanistic studies disclose that the in-situ generated Zr−H species and the free N−H group of the substrates play key roles in this transformation via Zr−H and N−H metathesis to form N−ZrIV species, and the subsequent C−C single bond activation is realized via a β-carbon elimination route. Our work presents an unprecedented group IV metal-catalyzed C−C single bond activation and hydroboration reaction. The C−C single bond activation model that was well studied for late-transition metals, were also elaborated to be applicable to the group IV metals.

Methods

General procedure for the Zr-catalyzed hydroboration of cyclopropylamines

In a nitrogen-filled glovebox, to a 15 mL pressure tube with a magnetic stirrer was added catalytic amount of Cp2ZrCl2 (0.01 mmol, 2.9 mg), K2CO3 (0.06 mmol, 8.3 mg), corresponding cyclopropylamine substrates (0.2 mmol), HBpin (0.3 mmol, 43.5 μL), and toluene (1 mL) in a sequence manner. Then, the pressure tube was taken out of the glove box and allowed to stir at 120 °C for 24 h. Upon completion, all the solvent was evaporated, and the crude product was isolated on silica gel using flash chromatography with dichloromethane/ethyl acetate as the eluent to give the corresponding products.