Introduction

In the past decade, enantioselective C–H activation reactions via asymmetric metalation have emerged as a promising platform for asymmetric catalysis1,2,3,4. Significant progress has been made in this field with the development of catalytic systems using Pd(II)5,6,7, Pd(0)8,9,10,11,12, Rh(III)13,14,15,16, Rh(I)17, Ir18,19,20,21,22,23,24,25, and other precious metals. In particular, a wide range of chiral Pd(II) catalysts have been developed for inducing point chirality, axial chirality, and planar chirality owing to the discovery of bifunctional chiral mono-N-protected amino acid (MPAA) ligands and next-generation ligands capable of accelerating C–H activation5,6,7. In contrast, there are few reports of asymmetric C–H activation using first-row transition metal catalysts, such as Fe, Co, Ni, or Cu, despite their abundance and low cost26,27,28,29,30,31,32,33. For example, Cu-catalyzed C–H activation reactions have immense potential utility given the superior tolerance of Cu catalysts for heterocycles and other sensitive functionalities compared to precious metal catalysts34,35,36,37,38,39,40,41,42,43,44,45,46,47,48,49,50,51,52,53,54. However, the development of enantioselective transformations has not been successful. The discovery of chiral ligands that accelerate Cu-catalyzed C–H activation is essential, as it enables the chiral ligand–catalyst complex to outcompete the racemic “background” reaction that is prevalent with unligated Cu. In addition, the establishment of a stereo model will provide unique insights into the transition state structures.

Given the aforementioned advantages of Cu-catalyzed C−H activation reactions, our group has devoted extensive effort to the development of Cu catalysis since the first report in 200634. We have developed diverse transformations using a wide range of coupling partners enabled by our class of practical oxazoline-aniline directing groups (DGs)46,47,48,49,50,51,52,53,54. Our oxazoline-aniline DG not only unlocks reactivity with an enormous range of coupling partners for C–H activation, including free amines/amides46 and trifluoromethylating reagents48 but also demonstrates exceptional tolerance for otherwise challenging functional groups, especially a broad range of heterocycles52,53.

In the realm of stereoselective Cu-catalyzed C−H activation, we recently reported the diastereoselective C–H thiolation of ferrocenes using a chiral oxazoline directing group (Fig. 1a)55. Despite its practical limitations, this chiral auxiliary approach represents a promising lead for the ultimate development of Cu-catalyzed enantioselective C−H activation reactions. Here, we report the discovery of BINOL-derived ligands that accelerate the Cu-mediated C–H alkynylation of ferrocenes with an oxazoline-amide directing group. When using (S)−6,6’-dibromo-BINOL as the chiral ligand, the Cu-mediated enantioselective C–H alkynylation of ferrocene carboxylic acid derivatives was achieved with high enantioselectivity (Fig. 1b). Mechanistic studies support the presence of ligand acceleration and indicate the acceleration occurs at the key C–H cleavage step of the catalytic cycle.

Fig. 1: Ligand accelerated Cu-mediated enantioselective C–H alkynylation of ferrocenes.
figure 1

a Stereoselective transition metal-catalyzed C–H activation of ferrocenes. DG directing group, FG functional group. b Ligand-accelerated enantioselective Cu-mediated C–H ferrocene alkynylation.

Results and discussion

Reaction optimization for Cu-mediated C–H alkynylation

We targeted the ortho-C(sp2)–H activation of planar chiral ferrocene carboxylic acid derivatives as the model system to develop ligands to accelerate C–H activation with Cu. We derivatized the carboxylic acid with our oxazoline-amide directing group as it promotes a range of transformations and is readily installed and removed46,47,48,49,50,51,52,53,54. Achieving ligand acceleration could enable the development of asymmetric Cu-catalyzed C–H activations. A number of catalytic asymmetric C–H activations of ferrocenes using other metals have been reported18,56,57,58,59,60,61,62,63,64,65.

Planar chiral ferrocenes are prevalent in synthetic chemistry, materials science, and medicinal chemistry66,67,68,69, as well as privileged scaffolds for chiral ligands and catalysts70,71,72,73,74,75. More specifically, ortho-substituted planar chiral ferrocene carboxylic acids and their derivatives, the products of this method, are highly valuable in the preparation of chiral monodentate (carboxylates)76 and bidentate (oxazolines/phosphines, etc.)77 ligands. Traditional directed ortho lithiation approaches for the enantioselective preparation of ortho-substituted ferrocene carboxylic acids rely on chiral auxiliaries78,79,80,81,82. They suffer from efficiency and atom economy losses associated with the use of chiral auxiliaries, as well as the scope limitations associated with organolithium reagents. A Cu-catalyzed enantioselective C–H activation approach towards ortho-substituted planar ferrocene carboxylic acid derivatives would circumvent these limitations and permit direct installation of diverse functionality using an abundant base metal catalyst. Notably, very few enantioselective catalytic C–H activation reactions to prepare ortho-substituted ferrocene carboxylic acid derivatives are known83.

Preliminary experiments with the racemic Cu-mediated C–H alkynylation of ferrocene carboxylic acid substrate 1 with ethynylbenzene (2a) provided only 14% yield of product 3a under our previously developed conditions for Cu-mediated C–H alkynylation, which did not use an exogenous ligand36. Through a systematic evaluation of reaction parameters, the yield of 3a was improved to 56% (17/1.0 mono/di). Encouraged by our success in developing ligands capable of accelerating Pd-catalyzed C−H activation, we then initiated a search for ligand acceleration effects for this Cu-mediated transformation. A range of ligand scaffolds were systematically investigated (Fig. 2, see Supplementary Information for more details). Common mono- and bi-dentate donating ligands such as triphenylphosphine (L1), BINAP (L2), 1,10-phenanthroline (L3), and bis-oxazoline (L4) either inhibited or showed no beneficial impact on reactivity. We next evaluated mono-protected amino acid (L5) and pyridone (L6) ligands, two scaffolds widely used in Pd-catalyzed C−H activation and capable of directly accelerating the key C−H activation step. Unfortunately, both ligand classes were ineffective in this transformation. Pyridine-type (L7) ligands, which are also widely used to improve the reactivity of Pd C−H activation catalysts, failed to show any improvement. Finally, mono-oxazoline ligands (L8), which are effective ligands in Cu-mediated C−H activations with monodentate directing groups43, were not beneficial in this transformation.

Fig. 2: Ligand effects for Cu-mediated C–H alkynylation of ferrocenes.
figure 2

Reaction conditions: 1 (37.4 mg, 0.1 mmol, 1.0 equiv), 2a (27.5 μL, 0.25 mmol), CuOAc (18.4 mg, 0.15 mmol), Ligand (40 mol%), NaOAc (8.2 mg, 0.1 mmol), DMSO (5.0 mL), 70 °C, air, 12 h. The yield was determined using tetrachloroethane as the internal standard, and the ratio of mono/di was determined by analysis of the crude 1H NMR. The data in the bracket indicates the isolated yield.

In contrast to the results with established ligands for Pd-catalyzed C−H activations, we found that (S)-BINOL (L10) showed a significant improvement in reactivity, increasing the yield of 3a to 69% and indicating the potential for ligand accelerated catalysis84. Other bisphenol scaffolds such as L9 and (S)-spirosilabiindane diol (SPSiOL, L12)85 showed inferior results, while (S)-spirobiindane diol (SPINOL, L11) demonstrated similar reactivity as L10. Given the established modularity and ease of derivatization of the BINOL core, we chose the BINOL platform for further investigation. We first evaluated 3,3’-Me2-BINOL (L15), which to our delight gave an improved yield of 76% (4.8/1 mono/di). Control experiments revealed that both phenolic hydroxyl groups are crucial for the high reactivity (L13, L14 vs. L10). The observation of this potential ligand acceleration effect with Cu not only will promote the discovery of Cu-mediated/catalyzed C−H activations with this abundant and economical metal but also enable the development of enantioselective catalysis (vide infra).

Substrate scope for the Cu-mediated C–H alkynylation

With the optimal conditions in hand, the scope of the ligand-accelerated Cu-mediated C–H alkynylation was investigated. Employing ferrocene monocarboxylic acid-derived 1 as the model substrate, we first evaluated the breadth of terminal alkynes (Fig. 3). In general, this method could tolerate a series of aryl, alkenyl, and alkyl-substituted terminal alkynes, providing the desired products in moderate to good yields (3a–r). Both electron-rich (2b–d, 2g, 2i) and electron-deficient (2e, 2f, 2h, 2j) substituents on the phenyl group of phenylacetylenes are compatible with this protocol. To our delight, terminal alkyne 2l bearing a heterocyclic thiophene moiety was also compatible, providing product 3l in 64% yield. Notably, aliphatic alkynes containing both cyclic (2m, 2n) and acyclic (2o–q) alkyl groups, were all compatible, giving products 3m–q in good yields. Conjugated enyne 2r gave the ferrocene-containing enyne 3r in 62% yield, further demonstrating the generality of this protocol.

Fig. 3: Scope of alkynes for ligand accelerated Cu-mediated C–H alkynylation of ferrocenes.
figure 3

Reaction conditions: 1 (37.4 mg, 0.1 mmol), 2 (0.25 mmol, 2.5 equiv), CuOAc (18.4 mg, 0.15 mmol), L15 (12.6 mg, 40 mol%), NaOAc (8.2 mg, 0.1 mmol), DMSO (5.0 mL), 70 °C, air, 12 h. Isolated yield and the ratio of mono/di were determined by analysis of the crude 1H NMR. For 3f, the reaction was conducted for 16 h.

Next, the scope of ferrocene substrates was evaluated with phenylacetylene (2a) as the alkynylating reagent (5a–i), as presented in Fig. 4. Ferrocenes bearing both electron-deficient ketone and ester groups (4a–e) and electron-rich alkyl (4f–h) substituents on the lower cyclopentadiene ring were all tolerated, affording the corresponding alkynylated ferrocene derivatives in moderate to good yields. Bulky ferrocene 4i bearing a Cp* on the lower ring gave a lower yield (38%) with excellent mono selectivity, potentially due to steric hindrance.

Fig. 4: Scope of ferrocene derivatives.
figure 4

Reaction conditions: 4 (0.1 mmol, 1.0 equiv), 2a (27.5 μL, 0.25 mmol), CuOAc (18.4 mg, 0.15 mmol), L15 (12.6 mg, 40 mol%), NaOAc (8.2 mg, 0.1 mmol), DMSO (5.0 mL), 70 °C, air, 12 h. Isolated yield and the ratio of mono/di was calculated according to the isolated yield.

Reaction optimization for Cu-mediated asymmetric C–H alkynylation

Encouraged by the potential ligand acceleration effect in Cu-mediated C–H activation with BINOL ligands, we hypothesized that an asymmetric Cu-mediated C–H alkynylation of ferrocenes could be achieved with enantioenriched chiral diol ligands. This method would enable the enantioselective preparation of useful planar chiral ferrocene scaffolds. Ligand acceleration of the catalytic cycle is particularly critical for asymmetric Cu C–H activation catalysis, in light of the prominent achiral “background” reaction often observed with Cu. Upon evaluating enantiopure BINOL ligands in our racemic reaction protocol, we obtained a promising initial hit with (S)-BINOL ((S)-L10) providing product 3a in 78.5:21.5 er. Additional optimization of reaction parameters to a mixture of CuOAc (50 mol%) and Cu(OAc)2 (70 mol%) as the Cu sources and 60 mol% of (S)-L10 ligand, as well as the use of Ag2CO3 (0.5 equiv) and NaOAc (1.0 equiv) improved the enantioselectivity to 82:18 er in 63% yield (6.0/1.0 mono/di) (Fig. 5). Next, ligand effects were explored. (S)-H8-BINOL (L16) led to 54:46 er, while (S)-SPINOL (L11) and (R,R)-TADDOL (L17) gave the racemic product. Although (S)−3,3’-Me2-BINOL (L15) showed superior reactivity under the racemic reaction conditions, it only provided the product at 55:45 er. The installation of other substituents at the 3,3’-positions of BINOL (L18–20) generally resulted in inferior selectivity to L10, although results with L19 and L20 (70:30 er, 74:26 er, respectively) suggested that electron-withdrawing groups could benefit asymmetric induction by rendering the phenolic groups more acidic. Following this reasoning, we then evaluated (S)−6,6’-Br2-BINOL [(S)-L21], with distal Br substituents capable of acidifying the phenolic groups without steric interference. Gratifyingly, these changes improved stereocontrol to 87:13 er. Phenyl substituents at the 6,6’-positions provided inferior results (L22, 72:28 er). Control experiments with enantioenriched mono-((S)-L13) and di-((S)-L14) protected BINOL ligands showed that both hydroxyl groups are essential for selectivity and reactivity.

Fig. 5: Evaluation of chiral diol ligands for Cu-mediated asymmetric C–H alkynylation.
figure 5

Reaction conditions: 1 (37.4 mg, 0.1 mmol), 2a (33.0 μL, 0.30 mmol), CuOAc (6.1 mg, 50 mol%), chiral diol ligand (50 mol%), Cu(OAc)2 (12.7 mg, 0.07 mmol), NaOAc (8.2 mg, 0.1 mmol), DMSO (5.0 mL), 60 °C, air, 12 h. Isolated yield and the ratio of mono/di were determined by analysis of the crude 1H NMR. The er value for the mono-alkynylated product was determined by chiral HPLC.

Substrate scope for the asymmetric C–H alkynylation

With (S)−6,6’-dibromo-BINOL [(S)-L21] as the optimal ligand, the enantioselectivity was further improved to 92.5:7.5 by reducing the loading of CuOAc/[(S)-L21] and lowering the reaction temperature. The absolute configuration of the planar chiral alkynylated ferrocene was determined by X-ray crystallography of (Rp)-3a. This novel process features broad substrate scope with respect to the terminal alkynes (3a–r, Fig. 6). Both substituted phenylacetylenes and alkyl-substituted acetylenes are suitable substrates, delivering the enantioenriched mono-alkynylated ferrocenes in synthetic useful yields (35–68% yields) with good enantioselectivities (88:12 er to 93:7 er). The mono/di selectivity is lower than the racemic reaction (Figs. 3 and 4), probably due to the higher reactivity of (S)−6,6’-dibromo-BINOL [(S)-L21] in comparison with 3,3’-Me2-BINOL (L15). Notably, heterocyclic arylalkynes containing thiophene (2l) also proceeded in high enantioselectivity (89.5:10.5 er). A range of ferrocene carboxylic acid derivatives (4f–h) are also compatible with this protocol, providing the desired planar chiral ferrocenes (Rp)-5f-h in high yields and high ers.

Fig. 6: Cu-mediated enantioselective C–H alkynylation of ferrocenes.
figure 6

Reaction conditions: 1 or 4 (0.1 mmol), 2 (0.25 mmol), CuOAc (3.7 mg, 30 mol%), (S)-L21 (13.3 mg, 30 mol%), Cu(OAc)2 (12.7 mg, 0.07 mmol), NaOAc (8.2 mg, 0.1 mmol), DMSO (5.0 mL), 60 °C, air, 12 h. Isolated yield and the ratio of mono/di were determined by analysis of the crude 1H NMR. The er value for the mono-alkynylated product was determined by chiral HPLC.

To demonstrate the synthetic utility of the Cu-mediated asymmetric C–H alkynylation, we conducted the transformation on a gram-scale and elaborated the product to synthetically useful compounds. First, the gram-scale reaction was conducted with 70 mol% of Cu(OAc)2, 20 mol% of CuOAc, and 20 mol% of (S)-L21 (Fig. 7a), providing enantioenriched mono-alkynylated product 3a in 52% yield with slightly lower enantioselectivity (90:10 er), and the di-product (di-3a) in 7% yield. Recrystallization from hexane/dichloroethane provided 3a in very high selectivity (99.5:0.5 er) and 38% yield. To explore potential applications of the product (Fig. 7b), we derivatized 3a through a synthetic sequence, first through hydrogenation to ortho-alkylated 6a, followed by removal of the directing group to reveal ferrocene carboxylic acid 7a, with minimal loss of enantioenrichment (98:2 er). Ortho-substituted planar chiral ferrocene carboxylic acids have recently been shown by Matsunaga to be effective chiral ligands in Co-catalyzed enantioselective C(sp3)–H activation reactions76. As a demonstration of our method’s potential impact, we employed 7a as a chiral ligand in the Co-catalyzed C(sp3)–H amidation of thioamide 8 (Fig. 7b), providing product 10 in high yield (95%) and promising enantioselectivity (70.5:29.5 er). This example highlights the synthetic utility of our Cu-mediated enantioselective C–H alkynylation, which enables access to novel chemical space in valuable ortho-substituted planar chiral ferrocenes.

Fig. 7: Gram-scale reaction, product elaboration, and synthetic application.
figure 7

a Gram-scale reaction. b Product elaboration and synthetic application in catalysis.

Mechanistic insights

The increased reactivity and high enantioselectivity imparted by the BINOL ligands were suggestive of ligand-accelerated catalysis, particularly in light of the strong background reaction. To further probe the possibility of ligand acceleration and collect data on the catalytic cycle, we then conducted some preliminary mechanistic studies on the system. We first performed a one-pot intermolecular competition kinetic isotope effect (KIE) experiment using 1a and ortho-deuterated D2−1a (Fig. 8a) in the presence of L21. A primary KIE was observed (4.9), indicating that C–H cleavage is likely the rate-determining step. This result indicates that the Cu-mediated C–H alkynylation is likely proceeding through CMD C–H activation at Cu, as opposed to SET pathways that do not involve Cu in C–H bond breaking, which would show no KIE (~1)86. This is a critical mechanistic insight as it both establishes that Cu cleaves C–H bonds in a similar manner as Pd and suggests a plausible mechanism for ligand involvement and acceleration. Having identified C–H cleavage at Cu as the rate-determining step, we next conducted initial rate studies to test for ligand acceleration (Fig. 8b). We observed a striking initial rate difference when comparing catalyst reactivity with and without BINOL-derived ligand (S)-L21. A rate acceleration of 11.6 times was observed with (S)-L21, strongly supporting our initial ligand acceleration hypothesis. Taken together, these experiments indicate that the BINOL scaffold accelerates the key rate- and enantio-determining C–H cleavage step at Cu, similar to the behavior of MPAAs and related ligands for Pd5,6,7.

Fig. 8: Mechanistic studies.
figure 8

a Kinetic isotope effect experiments. b Initial reaction rate in comparison with and without (S)-L21.

In conclusion, Cu-mediated enantioselective ortho C–H alkynylation of ferrocene carboxylic acid derivatives has been realized using a chiral BINOL-derived diol ligand. This scaffold demonstrates ligand-accelerated catalysis with the Cu-mediated C–H activation, which both boosts the reactivity and also enables enantioselective catalysis. The development of additional Cu-catalyzed/mediated asymmetric C–H activation reactions with these chiral diol ligands is an ongoing research direction in our laboratory.

Methods

General procedure for Cu-mediated asymmetric C–H alkynylation of ferrocenes

A 15 mL scale tube was charged with substrate 1 or 4 (0.1 mmol, 1.0 equiv.), Cu(OAc)2 (12.7 mg, 0.07 mmol), CuOAc (3.7 mg, 0.03 mmol), (S)-L21 (13.3 mg, 30 mol%), Ag2CO3 (13.8 mg, 0.05 mmol), NaOAc (8.2 mg, 0.1 mmol), 2 (0.25 mmol, 2.5 equiv.), DMSO (5.0 mL) under air atmosphere. The tube was capped tightly, and the reaction mixture was stirred at room temperature for 30 s and then stirred at 60 °C for another 12 h. Upon completion, the reaction was cooled to room temperature, and then EtOAc was added to dilute the reaction mixture. The organic layer was washed with NH3·H2O, saturated brine, and dried over Na2SO4. Volatiles were removed under a vacuum. The crude product was purified by column chromatography to afford the desired product (Rp)-3 or (Rp)-5. The ratio of mono/di was determined by the analysis of crude 1H NMR. Full experimental details and characterization of new compounds can be found in the Supplementary Methods.