Introduction

Over the past decade, mechanochemically promoted organic transformations have gained major momentum due to beneficial features, including shorter reaction times, lower catalyst loadings, unique reactivity, and most notably the reduction of harmful, often toxic organic solvents1,2,3,4. In this context, pioneering studies by Bolm5,6,7,8,9, Ito10,11, and others12,13,14,15 have demonstrated the potential of mechanochemistry in enabling bond activations with precious transition metal catalysts, thus reflecting its unique potential for sustainable molecular assembly.

C–H alkylation via the direct regioselective addition of indoles onto alkenes is arguably the most efficient and atom-economical approach to obtain functionalized indoles16,17, which are omnipresent structural motifs in drug development due to their strong affinity for biological targets (Fig. 1a)18,19. Despite indisputable advances, these hydroarylations largely relied on the use of toxic, cost-intensive and precious metal catalysts20,21, while transformations involving more sustainable 3d metal catalysts remain limited22,23,24,25,26,27,28,29,30,31. Recently, Yoshikai32 and Ackermann33,34,35 groups developed C–H alkylation of indoles and styrenes using the most abundant, low-cost, and non-toxic iron catalyst36,37,38,39,40,41 with monodentate N-heterocyclic carbene (NHC) ligands. However, several inherent limitations in these approaches still persist and need to be addressed. The use of (super)stoichiometric amounts of Grignard reagents drastically reduces the reaction sustainability as well as their functional group tolerance42,43,44. In addition, the reported methods have only achieved Markovnikov-selective hydroarylation of aryl alkenes32,33,34,35 or anti-Markovnikov-selective hydroarylation of heteroatom-substituted alkenes35,40,41, while transformations involving unactivated alkenes remain challenging.

Fig. 1: Towards new sustainable assembly of functionalized indoles by mechnochemical iron-catalyzed regiodivergent hydroarylation.
Fig. 1: Towards new sustainable assembly of functionalized indoles by mechnochemical iron-catalyzed regiodivergent hydroarylation.The alternative text for this image may have been generated using AI.
Full size image

a Indole-containing drug molecules. b Use of magnesium metal in mechanochemistry. c Key features comparison of representative iron(0)-NHC complexes. d This work: mechanochemical iron-catalyzed regiodivergent hydroarylation.

The Grignard reagent serves a pivotal role in iron/NHC catalysis, and, as a strong reductant, is responsible for the generation of catalytically active low-valent iron species32,33,34. The in situ generation of Grignard reagent from aryl halides and magnesium metal through the mechanochemical technique has been achieved to avoid the sensitive and tedious preparation steps (Fig. 1b)45,46,47,48,49,50,51. This one-pot approach benefits from the capability of mechanochemistry for the direct activation of zero-valent magnesium52,53,54 by an electron transfer process. Besides, magnesium metal has proven to be a useful alternative to Grignard reagents in cobalt-catalyzed hydroarylation, albeit with high reaction temperature and narrower scopes55. However, the direct use of magnesium metal as a reductant in iron-catalyzed hydroarylation has thus far proven elusive. In this context, we inferred that magnesium metal might solely serve as a reductant under mechanochemical conditions, circumventing the involvement of the Grignard reagent, to generate a catalytically active low-valent iron species for the hydroarylation.

NHCs represent an indispensable class of ligands for transition-metal catalyzed C–H activations56,57. Their strong σ-donating ability contributes to the more stable metal–ligand bond, thereby enhancing the robustness of the metal complexes58,59,60. Compared with monodentate NHC ligands, bidentate bis(NHC) ligands exhibit distinctive spatial characters and electronic effect61,62. Taking low-valent iron(0) complex as an example, bis(NHC) complex has a larger cone angle63 and buried volume64, which potentially brings opportunities for precise regioselectivity control (Fig. 1c). Moreover, the location of higher spin density at the iron center in bis(NHC) complex may lead to underdeveloped reactivity65,66, presenting avenues for further exploration.

Herein, aiming at addressing the limitations of iron/NHC-catalyzed C–H functionalization, we develop a mechanochemical iron-catalyzed regiodivergent hydroarylation of alkenes with indoles via C−H activation (Fig. 1d). The salient features of this strategy include: (1) the use of magnesium metal as a convenient reductant in iron-catalyzed hydroarylation with the aid of mechanochemistry, which improves the sustainability and functional group compatibility; (2) anti-Markovnikov-selective hydroarylation of unactivated alkenes using an iron/bis(NHC) catalytic system; and (3) regiodivergent hydroarylation of aryl alkenes by the judicious choice of the NHC ligands. Further detailed experimental and theoretical mechanistic studies allow shedding light on the iron/NHC catalytic system from a broader perspective.

Results and discussion

Optimization of reaction conditions

We initiated our studies into the hydroarylation of 1-octene 2a with indole derivative 1a using the iron/NHC catalytic system under mechanochemical conditions (Fig. 2). We probed the reaction using different types of monodentate NHC preligands (L1L7), which were unable to deliver the desired product. Subsequently, we attempted the reaction using bis(NHC) preligands. Pleasingly, the desired alkylated indole 3 was obtained in 21% yield with anti-Markovnikov selectivity when using Fe(acac)3 as the metal catalyst, L8 as the preligand and magnesium chip as the reductant under mechanochemical conditions. Thereafter, the structure of bis(NHC) preligand was further explored. A drastic increase was observed upon the substitution of the methyl group in L8 by the bulkier isopropyl (L9), where the alkylated indole was obtained in 84% yield. Further improvement in the yield (92%) was achieved by using 2,4,6-trimethylphenyl as the substituent (L10). When applying phenyl or 2,6-diisopropylphenyl as the substituent (L11 and L12), only trace amounts of the alkylated product were obtained, indicative of the relevance of stereoelectronic features. The use of a bis(NHC) preligand with saturated backbones (L13) and a tridentate NHC preligand (L15) resulted in low reaction efficiency, while a bis(NHC) preligand bridged by two methylene groups (L14) completely inhibited reactivity. Finally, the reaction conditions were further optimized (see Supplementary Information Table S2 and S3 for details). The reaction efficiently proceeded even under additive-free or solvent-free conditions, albeit with a slight drop in yield. Control experiments indicated that the iron catalyst, the bis(NHC) preligand, and the magnesium chip were crucial for the desired transformation. Notably, replacing ball milling by thermal reaction conditions completely failed to furnish the desired product, even at an elevated reaction temperature of 100 °C.

Fig. 2: Screening of NHC preligands for mechanochemical iron-catalyzed hydroarylation of unactivated alkenes.
Fig. 2: Screening of NHC preligands for mechanochemical iron-catalyzed hydroarylation of unactivated alkenes.The alternative text for this image may have been generated using AI.
Full size image

Reaction conditions: 1a (0.1 mmol), 2a (0.15 mmol), Fe(acac)3 (10 mol%), NHC preligand (10 mol%), Mg chip (50 mol%), TMEDA (0.1 mmol) and THF (50 μL) were placed in a stainless-steel vessel (5 mL) with stainless-steel ball (dMB = 7 mm), milled in a mixer mill (RETSCH MM 400) at 30 Hz for 180 min under nitrogen atmosphere. Then THF (3 mL) and HCl aq. (3 M, 1 mL) were added, and the mixture was stirred for 2 h. The ratio of l:b is > 99:1 for all cases. The yield was determined by 1H NMR spectroscopy using 1,3,5-trimethoxybenzene as the internal standard. aIn the absence of TMEDA: 72% yield; in the absence of THF: 83% yield; in the absence of Fe(acac)3, NHC preligand, or Mg chip: n.d. TMEDA, N,N,N’,N’-tetramethylethylenediamine; THF, tetrahydrofuran; Mes, 2,4,6-trimethylphenyl; Dipp, 2,6-diisopropylphenyl; Xyl, 2,6-dimethylphenyl; n.d., not detected.

Substrate scope

With the optimal bis(NHC) preligand (L10) in hand, we evaluated the viable substrate scope for hydroarylation of unactivated alkenes under the optimized reaction conditions to delineate the full potential and robustness of this transformation (Fig. 3). Indoles bearing a variety of N-substituents gave the corresponding products 37 in moderate to good yields. The azaindole proved to be a suitable substrate, affording alkylated indole 8 in 86% yield. As expected, the presence of either electron-withdrawing or electron-donating substituents at 4-, 5-, 6- or 7-position of the indole ring was well tolerated, delivering the desired products 919 in high yields. Notably, cyano and ester substituents, which are incompatible under the Grignard conditions, were well tolerated in our mechanochemical approach (12,13). In addition, the reaction conditions turned out to be generally compatible with various unactivated alkenes. An array of linear and branched aliphatic alkenes revealed to be efficient in the reaction, providing the desired products 2024 in excellent yields. When employing ethoxyethene as the substrate, ethyl-substituted indole 25 was obtained in 66% yield. A variety of allylbenzenes containing either electron-withdrawing or electron-donating substituents performed well, providing the corresponding products 2630 in high yields. Silicon-, nitrogen- and oxygen-containing alkenes effectively engaged in the reaction, furnishing the corresponding products 3141 in good yields. Noteworthy, this hydroarylation could be further applied for the late-stage modification of complex molecules. As displayed, (-)-menthol, (-)-borneol, geraniol, eugenol, estrone and δ-tocopherol-conjugated indoles 4247 could be easily accessed in high yields. Remarkably, complete anti-Markovnikov selectivity was observed in all cases.

Fig. 3: Substrate scope for mechanochemical iron-catalyzed hydroarylation of unactivated alkenes.
Fig. 3: Substrate scope for mechanochemical iron-catalyzed hydroarylation of unactivated alkenes.The alternative text for this image may have been generated using AI.
Full size image

Reaction conditions: 1 (0.2 mmol), 2 (0.3 mmol), Fe(acac)3 (10 mol%), L10 (10 mol%), Mg chip (50 mol%), TMEDA (0.2 mmol) and THF (50 μL) were placed in a stainless-steel vessel (5 mL) with stainless-steel ball (dMB = 7 mm), milled in a mixer mill (RETSCH MM 400) at 30 Hz for 180 min under nitrogen atmosphere. Then THF (3 mL) and HCl aq. (3 M, 1 mL) were added and the mixture was stirred for 2 h. The ratio of l:b is >99:1 for all cases. Yields are those of the isolated products. a270 min. bUsing ethoxyethene as a substrate. cUsing tert-butyl(hex-5-en-1-yloxy)dimethylsilane as substrate with the reaction quenched by HCl aq. (3 M, 1 mL).

Next, we explored the effects of NHC ligands in the hydroarylation of aryl alkenes (Fig. 4). Using styrene 48a as the model substrate, the mechanochemical iron-catalyzed hydroarylation was first investigated by monodentate NHC preligands (Fig. 4a). Here, several monodentate NHC preligands (L1 − L6) all produced completely Markovnikov products (b:l > 99:1), among which L1 gave the highest yield (87%), whereas no product was obtained when using L7. A series of bis(NHC) preligands were then screened, revealing that L10 was the most suitable one for anti-Markovnikov selectivity (b:l = 6:94), while others, such as L8, L9, L11 − L15 displayed low efficacy and poor regioselectivity. These diverse selectivity data of styrene with different ligands encouraged us to apply the multivariate linear regression (MVLR) approach67 for capturing the structure-regioselectivity relationship in mechanochemical iron-catalyzed hydroarylation. A preliminary MVLR model was developed using four physical organic parameters of the key triplet iron(0)-NHC complexes (Fig. 4b), including cone angle θ, percent buried volume %VBur (sphere radius at 5.0 and 3.5 Å) and spin density on the iron center (vide supra). These four parameters were able to provide reasonable prediction on the regioselectivity of hydroarylation using aryl alkenes with an R2 value of 0.71 in leave-one-out cross-validation. Among them, the cone angle has the largest coefficient in the normalized regression equation, highlighting the significance of ligand geometry in determining regioselectivity. Nevertheless, the steric environment around the iron center and the electron-donating ability of NHC also contribute to the regioselectivity control.

Fig. 4: Optimization of catalyst systems for mechanochemical iron-catalyzed hydroarylation of aryl alkenes.
Fig. 4: Optimization of catalyst systems for mechanochemical iron-catalyzed hydroarylation of aryl alkenes.The alternative text for this image may have been generated using AI.
Full size image

a Screening of NHC preligands for hydroarylation of styrene. b Preliminary multivariate linear regression model for the structure-regioselectivity relationship in iron-catalyzed hydroarylation of styrene. Four parameters for key iron(0)-NHC complexes were involved, including cone angle θ, percent buried volume %VBur at sphere radius of 5.0 Å and 3.5 Å and spin density on iron center.

Having established the optimized reaction conditions, we turned to evaluate the substrate scope for this ligand-controlled regiodivergent hydroarylation of aryl alkenes (Fig. 5). First, the scope of Markovnikov-selective hydroarylation catalyzed by the monodentate preligand L1 was tested. Indoles bearing cyano, ester, or chloro groups on the aromatic ring were well-tolerated, resulting in the formation of alkylation products 4955 in good yields with excellent regioselectivities (b:l > 99:1 for all). A wide range of styrenes as well as vinylferrocene were competent substrates in the Markovnikov-selective hydroarylation, giving products 5663 in high yields (b:l > 99:1 for all). Subsequently, we examined the scope of anti-Markovnikov-selective hydroarylation catalyzed by the bidentate preligand L10. The substitution pattern and electronic properties of the substituents at 5- or 6- position of the aromatic ring did not affect the reaction efficiency and regioselectivity, giving rise to the corresponding products 64 − 69 in good yields with high levels of regioselectivity (l:b = 92:8 to 97:3). While, the substituent at the 7-position of indole, which is spatially proximal to the 2-position, imposed steric hindrance that affected the bond formation at the 2-position (70). Styrenes bearing electron-withdrawing or electron-donating substituents on the para-, meta- or ortho-position of aromatic ring, reacted well to give the desired products 71 − 77 in 51−86% yields (l:b = 85:15 to 98:2). Remarkably, vinylferrocene was also compatible with the optimal reaction conditions, affording the desired product 78 in good yield (67%) and regioselectivity (l:b = 95:5).

Fig. 5: Substrate scope for ligand-controlled regiodivergent hydroarylation of aryl alkenes.
Fig. 5: Substrate scope for ligand-controlled regiodivergent hydroarylation of aryl alkenes.The alternative text for this image may have been generated using AI.
Full size image

Reaction conditions: 1 (0.2 mmol), 48 (0.3 mmol), Fe(acac)3 (10 mol%), NHC preligand (10 mol%), Mg chip (50 mol%), TMEDA (0.2 mmol) and THF (50 μL) were placed in a stainless-steel vessel (5 mL) with stainless-steel ball (dMB = 7 mm), milled in a mixer mill (RETSCH MM 400) at 30 Hz for 270 min under nitrogen atmosphere. Then THF (3 mL) and HCl aq. (3 M, 1 mL) were added, and the mixture was stirred for 2 h. Yields are those of the isolated products.

Scale-up and late-stage transformations

To further demonstrate the utility of this method, a scale-up reaction and several late-stage transformations of alkylated indole products were carried out (Fig. 6). The current strategy was shown to be highly robust and scalable, as a gram-scale reaction of 1a and 2a proceeded smoothly, generating 3 in 86% yield (Fig. 6a). The alkylated indole-3-carboxaldehyde products act as synthetically useful intermediates that can be readily transformed into complex architectures using standard methods. The presence of the formyl group enables extensive structural diversification through subsequent functionalization. It could be removed under palladium catalysis, yielding indole 79 (Fig. 6b). The treatment with Grignard reagent delivered the corresponding alcohol 80 (Fig. 6c). A Wittig reaction produced the internal olefin 81 (Fig. 6d). Reductive amination using morpholine efficiently furnished the corresponding amine 82 (Fig. 6e). Furthermore, the formyl group was also able to be a linchpin for condensation, enabling the access to barbiturate-decorated indole 83 (Fig. 6f). Subsequently, we employed the formyl group as a weakly coordinating directing group to facilitate C4–H activation. Thereby, C4-functionalized indoles 84 − 86 could be synthesized in good yields (Fig. 6g−i). Ultimately, intramolecular crotonization of 86 led to 3,4-fused tricyclic indole 87 (Fig. 6j).

Fig. 6: Scale-up reaction and diversification of alkylated indole products.
Fig. 6: Scale-up reaction and diversification of alkylated indole products.The alternative text for this image may have been generated using AI.
Full size image

a Gram-scale reaction of 1a and 2a. b Transformation to indole 79 under palladium catalysis. cTransformation to alcohol 80. d Transformation to olefin 81. e Transformation to amine 82. f Transformation to barbiturate decorated indole 83. g C4-alkenylation of alkylated indole product 3 under ruthenium catalysis. h C4-amination of alkylated indole product 3 under iridium catalysis. i C4-alkylation of alkylated indole product 3 under rhodium catalysis. j Intramolecular crotonization of 86 to afford 3,4-fused tricyclic indole 87.

Mechanistic studies

To gain mechanistic insights into the modus operandi of the regioselective hydroarylation of unactivated alkenes, experiments and density functional theory (DFT) studies were carried out (Fig. 7). We synthesized a well-defined iron(II)-bis(NHC) complex65 from preligand L10 and demonstrated its catalytic activity under optimal conditions (Fig. 7a). However, it failed to catalyze the reaction in the absence of magnesium chip, indicating that the zero-valent iron-bis(NHC) complex produced by the mechanochemical magnesium reduction might be the active species. The reaction of the C2-deuterated indole [D]1-1a with 1-octene 2a afforded the product [D]1-3 (Fig. 7b). The transfer of the C2-deuterium from indole to two methylene groups implied the reversibility of the alkene insertion step. A deuterium-scrambling experiment was then performed using one equivalent of a mixture of substrates [D]1-1a and 1n under standard reaction conditions (Fig. 7c). H/D crossover observed in both products suggested that the dissociation and re-coordination of the alkene before the formation of alkyl iron species was possibly a reversible process.

Fig. 7: Mechanistic studies.
Fig. 7: Mechanistic studies.The alternative text for this image may have been generated using AI.
Full size image

a Reaction with iron(II)-bis(NHC) complex. b Deuterium-labeling experiment. c Deuterium-scrambling experiment. d Mechanism for the generation of linear product 3. Computational method at TPSSh-D3(BJ)/def2-TZVPP-SMD(Tetrahydrofuran)//B3LYP-D3(BJ)/def2-SVP level of theory. e DFT-computed regioselectivity-determining reductive elimination transition states forming C‒C bond of linear and branched products.

Computational studies

Based on the experimental results, DFT calculations were conducted to investigate the reaction mechanism and the origins of regioselectivity68, using indole 1a and 1-octane 2a as model substrates. It is worth noticing that the solvent effect was included in the free energy calculation based on gas-phase optimized geometries, given that our reaction is liquid-assisted grinding with THF69. The free energy diagram of the most favorable pathway for iron/NHC-catalyzed hydroarylation starting from iron(0) complex int1 is shown in Fig. 7d and Supplementary Information Supplementary Fig. S11. As the coordination of the substrate 1a requires a barrier of 23.9 kcal/mol via transition state TS4, it is most likely to be the rate-determining step in the catalytic cycle. The subsequent C‒H activation occurs through an oxidative addition of the C‒H bond onto the iron center, generating iron hydride species int7. The calculated alkene insertion is facile and reversible, consistent with deuterium-labeling experiments. Then, the resulting alkyl-iron(II) species int9 undergoes reductive elimination forming C‒C bond, which is probably the regioselectivity-determining step. Two competing reductive elimination transition states, TS10-linear and TS10-branched, generating the respective linear and branched products are presented in Fig. 7e. For unactivated alkene 2a, TS10-linear is 4.7 kcal/mol more favorable than TS10-branched in terms of free energy. In TS10-linear, the primary alkyl is positioned away from the substrate, allowing a favorable π-π stacking interaction between PMP and mesityl groups. In contrast, the secondary alkyl in TS10-branched is more steric-demanding, leading to repulsion between indole and alkyl chain, thereby increasing the reaction barrier for branched product. With the combination of experimental and computational studies, the reaction mechanism of iron-catalyzed hydroarylation with bis(NHC) ligand was clarified, along with the rationale of the regioselectivity of representative unactivated alkene.

We have successfully developed a sustainable and highly effective ligand-controlled regiodivergent hydroarylation of alkenes via mechanochemical iron-catalyzed C−H activation, leading to functionalized indoles with high structural diversity. The anti-Markovnikov-selective hydroarylation of unactivated alkenes could be achieved by the aid of a bis(NHC) ligand, while the regiodivergent hydroarylation of aryl alkenes was accomplished by modifying the structure of NHC ligands. Magnesium metal serves as a convenient reductant to form the catalytically active iron(0) species under mechanochemical conditions, making the reaction sustainable, environmentally friendly, and compatible with a broad range of functional groups. Moreover, detailed mechanistic studies and data science analysis revealed the reaction mechanism and elucidated the key features of NHC ligands in controlling the regioselectivity, setting the stage for the further application of this iron/NHC catalytic system.

Methods

General procedure for mechanochemical iron-catalyzed regioselective hydroarylation of unactivated alkenes

In the glove box, a mixture of indole substrate 1 (0.2 mmol), unactivated alkene 2 (0.3 mmol), Fe(acac)3 (10 mol%, 0.02 mmol, 7.1 mg), L10 (10 mol%, 0.02 mmol, 10.9 mg), magnesium chip (50 mol%, 0.1 mmol, 2.4 mg), TMEDA (0.2 mmol, 30 μL) and tetrahydrofuran (50 μL) were placed in a nitrogen-purged stainless-steel vessel (5 mL) with a stainless-steel ball (dMB = 7 mm). Then, the vessel was sealed and milled in a mixer mill (RETSCH MM 400) at 30 Hz for 180 min under a nitrogen atmosphere. Then, the reaction mixture was diluted with tetrahydrofuran (3 mL) and quenched with HCl aqueous solution (3 M, 1 mL). The resulting mixture was stirred at room temperature for 2 hours. The phases were then separated, the aqueous layer was extracted with ethyl acetate (5 mL × 3). The combined organic layer was washed with saturated NaHCO3 solution and brine, dried over Na2SO4, filtered and concentrated in vacuo. The linear and branched ratio was determined by 1H NMR analysis of the crude reaction mixture. The residue was purified by column chromatography on silica gel (n-hexane: ethyl acetate = 10:1) to afford the desired product.

General procedure for mechanochemical iron-catalyzed regiodivergent hydroarylation of aryl alkenes

In the glove box, a mixture of indole substrate 1 (0.2 mmol), aryl alkene 48 (0.3 mmol), Fe(acac)3 (10 mol%, 0.02 mmol, 7.1 mg), NHC preligand (10 mol%, 0.02 mmol), magnesium chip (50 mol%, 0.1 mmol, 2.4 mg), TMEDA (0.2 mmol, 30 μL) and tetrahydrofuran (50 μL) were placed in a nitrogen-purged stainless-steel vessel (5 mL) with a stainless-steel ball (dMB = 7 mm). Then, the vessel was sealed and milled in a mixer mill (RETSCH MM 400) at 30 Hz for 270 min under a nitrogen atmosphere. Then, the reaction mixture was diluted with tetrahydrofuran (3 mL) and quenched with HCl aqueous solution (3 M, 1 mL). The resulting mixture was stirred at room temperature for 2 hours. The phases were then separated, the aqueous layer was extracted with ethyl acetate (5 mL × 3). The combined organic layer was washed with saturated NaHCO3 solution and brine, dried over Na2SO4, filtered and concentrated in vacuo. The linear and branched ratio was determined by 1H NMR analysis of the crude reaction mixture. The residue was purified by column chromatography on silica gel (n-hexane: ethyl acetate = 5:1) to afford the desired product.