Introduction

Catalytic C−H activation has emerged as a powerful strategy for modifying carbon centers in organic molecules, offering atom-efficient and streamlined synthetic pathways1,2,3,4,5,6,7,8,9,10. In complex molecules where multiple C–H bonds often reside in similar environments, the incorporation of a directing group has proven to be the most effective method for achieving selective transformation (Fig. 1a)11,12,13,14,15,16. These directing groups, whether inherently present or artificially introduced, serve to precisely position the metal catalyst at the desired C–H bond. Such process is thermodynamically favored due to the formation of stable five-, six-, and seven-membered metallacyclic intermediates, which ensure regioselectivity through conformational rigidity. Non-metallic elements, particularly those from the chalcogen (O, S) and pnictogen (N, P) groups, are widely utilized as coordinating atoms in directing groups. In contrast, the integration of directing groups containing metalloid elements has seen limited progress, primarily due to their propensity to induce undesirable metathesis reactions in the presence of transition metals, a consequence of their metallic character17. Although some success has been achieved with silicon atoms, current methods are only confined to specific substrate classes that form five-membered metallacyclic intermediates, significantly limiting their broader applicability18,19,20. Consequently, there is a pressing need to develop more versatile C−H activation strategies that can accommodate a wider range of metalloid elements while moving beyond the conventional constraints of rigid metallacycle formation.

Fig. 1: Background and the discovery of catalytic C−H borylation via strained C(sp2)−Si/Ge metallacycles.
figure 1

a General modes for directed C−H activation. b Precedent discovery of strained silametallacycles. c Pd-catalyzed β-selective C–H activation of aliphatic amines. d Metalloid Si/Ge-directed C−H borylation (this work).

Full size image

Arenes ortho-bisfunctionalized with silyl and boryl, or germyl and boryl groups, have been extensively utilized in photoelectronics, semiconductors, catalysis, and organic synthesis due to the unique properties of these distinct metalloids21,22,23,24,25,26,27. However, current synthetic methods for these compounds often require intricate, multi-step processes, particularly when forming two C−metalloid bonds individually28,29,30,31. Seminal studies by Milstein32,33, Tilley, and Bergman34 have demonstrated that iridium complexes derived from triphenylsilane can produce benzo-fused four-membered silametallacycles through oxidative addition of ortho C−H bonds (Fig. 1b). Building on these pioneering works, we are inspired to explore three key questions: whether such strained metallacycles can be generated in a catalytic fashion, whether they can further couple with boron species, and whether the metalloid elements can be extended to germanium atoms. Notably, Gaunt and coworkers have reported palladium-catalyzed β-selective C−H activation of aliphatic amines, predominantly achieved via the formation of four-membered metallacycles (Fig. 1c)35,36. In contrast, catalytic C−H borylation leading to the formation of benzo-fused metallacycles introduces even greater ring strain and a metalloid motif that is more susceptible to uncharted side reactions37,38,39,40,41.

Here, we introduce an ortho-selective C−H borylation directed by silicon and germanium atoms (Fig. 1d). Our system, which utilizes arylhydrosilanes or arylgermanes as substrates, combines an iridium catalyst with a bipyridine-type ligand, an acetate base, and HBpin to efficiently facilitate the formation of a benzo-fused four-membered metallacycle within the catalytic cycle. Importantly, this methodology effectively overcomes the common issue of byproduct formations from C−metalloid cleavage26,42,43,44,45 and metalloid−H bond cleavage46,47,48,49,50. This approach enables the one-step preparation of a diverse array of ortho-boronated arylhydrosilanes and arylgermanes, offering superior efficiency in terms of atom, step and redox economy. The ability to selectively functionalize these aryl compounds with boron atoms while maintaining the integrity of the metalloid atoms represents a significant advancement in the field of C−H activation.

Results

Condition optimization

To establish a platform for this chemical process, we initially opted for the reaction between triphenylsilane (1a) and HBpin (Table 1). The reaction was employed [Ir(cod)OMe]2 as the catalyst along with a tridentate ligand L1 at a temperature of 125 °C in THF. However, merely trace amounts of the desired C–H borylation product 2a were observed via GC-MS analysis (entry 1). To enhance the overall yield of the reaction, we explored the influence of various additives. The incorporation of LiOAc into the reaction system could slightly improve the reactivity (entry 2). Additionally, we examined alternative salts such as NaOAc (19%) and KOAc (26%), which exhibited superior effectiveness in facilitating the formation of 2a (entries 3–4). Conversely, the utilization of K2CO3 resulted in a significant decrease in yield (entry 5). These discoveries underline the significance of both the K+ cation and OAc- anion in influencing the overall outcome of the reaction. Moreover, we investigated the impact of different ligands on the reaction. The inclusion of bipyridine-type ligands L2–L4 proved to be crucial in promoting the reaction (entries 6–8), with the highest achievable yield of 2a (75%) obtained in the presence of L4 (entry 8). Under the optimized reaction conditions, we observed the absence of diborylated products, with Ph-Bpin emerging as the primary byproduct, presumably resulting from C-Si bond cleavage. The structure of product 2a was further confirmed by X-ray crystallography. In contrast, employing 1,10-phenanthroline (L5) as the ligand led to a significantly lower yield (entry 9). Other solvents such as dioxane, toluene, and DMF resulted in much lower reactivity (entries 10–12). We also explored the potential of alternative boron reagents, such as B2pin2, which yielded inferior results (entries 13). Interestingly, alternative iridium sources such as [Ir(cod)Cl]2 and [Ir(cod)acac]2 resulted in a substantial decrease in yield (entries 14–15), while other transition metals like [Rh(cod)Cl]2 completely failed to promote the reaction (entry 16). Reducing the loading of [Ir(cod)OMe]2 to 1.0 mol% still exhibited an acceptable reactivity (entry 17). Finally, a control experiment confirmed that the reaction did not proceed in the absence of the metal catalyst (entry 18).

Table 1 Optimization of reaction conditionsa
Full size table

Substrate scope of silanes

Having the optimized reaction conditions in hand, we proceeded to explore the suitability of various arylhydrosilanes for ortho C–H borylation (Fig. 2). A wide range of triarylsilanes bearing substituents such as methyl (1b), isopropyl (1c), tert-butyl (1d), phenyl (1e), methoxy (1f), and methylthio (1g) were subjected to the system producing products 2b–2g with modest to good yields. Notably, the reaction of tri(naphthalen-2-yl)silane (1h), which possesses two potential sites for C–H borylation, demonstrated remarkable selectivity towards borylation at the less hindered C3 position. The structure of product 2h was also confirmed by X-ray crystallography. However, when identical reaction conditions were applied to dialkylarylhydrosilanes, the reaction efficiency decreased markedly, necessitating further optimization. The inclusion of NaOAc in the conditions with a mixture of B2pin2 and HBpin was found to significantly enhance the reaction efficiency. For example, silane 1i, bearing two isopropyl substituents, was successfully converted to product 2i in a 54% yield. In contrast, without NaOAc, the reaction became very chaotic. Substrate 1j, containing a para-methyl group, displayed favorable reactivity and selectivity at the less sterically hindered position. Conversely, substrate 1k, with a meta-methyl group, exhibited significantly reduced reactivity. Furthermore, other dialkylarylhydrosilanes with various substituents at the para position, including iPr (1m), tBu (1n), Ph (1o), OMe (1p), TMS (1q), F (1r), Cl (1s), CF3 (1t), and OCF3 (1u), were successfully utilized under these conditions. Additionally, the reaction of diisopropyl(naphthalen-2-yl)silane (1v) showed exclusive C3 selectivity. Under these optimized conditions, when diarylalkylhydrosilane 1w, containing two aryl and one methyl substituent, was tested, product 2h was only 17% yield, and a byproduct 2a, resulting from σ-bond metathesis, was observed. Interestingly, substrates with isopropyl (1x) and cyclopentyl (1y) groups exhibited compatible reactivity, indicating that the introduction of bulky substituents on the silicon atom can be advantageous in suppressing σ-bond metathesis. Finally, the reaction of silane 1z, containing two aryl and one benzyl substituent, displayed extraordinary selectivity for C–H bond activation within the benzyl motif, resulting in the formation of the dual C–H borylation product 2z with a yield of 68%. This intriguing outcome suggests a preferential formation of five-membered silametallacycles during the C–H activation process.

Fig. 2: Reaction scope of hydrosilanes.
figure 2

Reaction conditions: [Ir(cod)OMe]2 (2.5 mol%), L4 (5.0 mol%), 1 (0.50 mmol), HBpin (1.25 mmol), KOAc (1.25 mmol) in 1.5 mL of THF at 125 °C for 6 h under argon; For entries 8–25: [Ir(cod)OMe]2 (5.0 mol%), L4 (10.0 mol%), HBpin (0.15 mmol), B2pin2 (0.75 mmol), NaOAc (1.25 mmol) at 150 °C. aIsolated yield.

Full size image

Substrate scope of germanes

Based on the above findings, we extended our investigation to explore the application of hydrogermanes in C–H borylation (Fig. 3). Initially, when triphenylgermane (3a) was substituted with 1a under the standard conditions, a significant presence of byproducts arising from σ-C–Ge bond metathesis was observed. To address this challenge, we performed a comprehensive screening and identified an optimized reaction protocol employing a stoichiometric quantity of the tridentate ligand L1. This ligand effectively inhibited the metathesis pathway, enabling the formation of the desired product 4a with a 72% yield (entry 1). Diminishing the catalyst load significantly reduced the yield of the target product (entry 2). Nonetheless, decreasing the ligand amount to 30 mol% still facilitated the formation of 4a with an acceptable yield (56%). Notably, substituting ligand L1 with L4 resulted in only trace amounts of the desired product (entry 4). Additionally, KOAc was found to be indispensable for the reaction’s progression (entry 5). Leveraging these insights, we investigated the feasibility of C–H borylation using a variety of arylhydrogermanes. Triarylhydrogermanes with different substituents on the benzene ring, such as methyl (3b), isopropyl (3c), methoxy (3d), methylthio (3e), fluoro (3f), and trimethylsilyl (3g), were tested in the reaction system, yielding products 4b–4g in moderate yields. Notably, the reaction involving tri-3-methylphenylgermane (3h) and tri-2-naphthylgermane (3i), which have two potential sites for borylation, showed a significant preference for the less sterically hindered position. Furthermore, hydrogermanes 3j and 3k, substituted with two phenyl groups and one alkyl group, were effectively transformed into products 4j and 4k with high efficiency. However, under these updated conditions, the reaction of a hydrogermane, containing one phenyl and two isopropyl substituents, completely failed. Instead, we only observed the byproduct resulting from σ-bond metathesis (not shown in the Figure).

Fig. 3: Optimization of Ge-directed C−H borylation and the reaction scope of hydrogermanes.
figure 3

Reaction conditions: [Ir(cod)OMe]2 (5.0 mol%), L1 (1.5 mmol), 3 (0.50 mmol), HBpin (1.25 mmol), KOAc (1.25 mmol) in 3.0 mL of THF at 115 °C for 24 h under argon. aYield was determined by GCMS. bIsolated yield.

Full size image

Product derivatization

The adept synthesis of mixed-metalloid compounds has paved the way for more in-depth investigations into their reactivity (Fig. 4). Our initial studies focused on the reactivity of ortho-boronated arylhydrosilanes (Fig. 4a). For instance, compound 2a can efficiently undergo a Pd-catalyzed Suzuki–Miyaura coupling reaction with iodoarenes 5, leading to the rapid formation of biaryl product 6 in a yield of 54%. Subsequently, in the presence of Wilkinson’s catalyst, compound 6 can be further transformed through dehydrogenative cyclization, yielding silole 7 with an impressive 88% yield51. Moreover, the reaction of compound 2a with ClCH2I and nBuLi results in the in situ generation of the CH2BrLi reagent, which ultimately leads to the formation of boronate 8, now featuring an additional methylene unit52. Notably, the methylation of the Si–H bond in 2a, achieved via treatment with CH2I2 and Et2Zn, delivers product 9 with full conversion. This compound can then be effectively oxidized using NaBO3·4H2O, providing phenol 10 in nearly quantitative yield. Further diversification of compound 9 is achieved by subjecting it to NIS in the presence of CuI, resulting in the iodination product 11 with a yield of 73%. Additionally, we developed a Rh-catalyzed cyclization protocol wherein compound 9 reacts with alkyne 12, enabling the synthesis of benzosilole 13 with a 63% yield53. We also explored the reactivity of ortho-boronated arylgermanes, using reagent 3a as the starting material (Fig. 4b). Through a sequential C–H borylation and methylation of the Ge atom, compound 14 can be synthesized in a one-pot process with a yield of 64%. Utilizing the above rhodium-catalyzed cyclization conditions, we successfully obtained benzogermole 15 with a good yield from compound 14 and alkyne 12.

Fig. 4: Synthetic applications.
figure 4

a Follow-up transformations of compound 2a. b One-pot C–H borylation and methylation of 3a and further cyclization.

Full size image

Mechanistic investigation

A series of deuterium labeling experiments were initially conducted in the absence of HBpin to elucidate the potential reaction intermediates (Fig. 5). When the reaction utilizing Ph3Si–D (1a–d) was performed without KOAc, 76% deuterium incorporation was observed at the ortho position of 1a, while only 5% D was detected at the silicon atom (Fig. 5a). The observed D incorporation suggests that the oxidative addition of Si–H bond to iridium center is irreversible and that the reaction may proceed through a four-membered silacyclic metal intermediate. The ortho C–D incorporation increased upon the addition of KOAc, indicating that KOAc accelerates the ortho C–H metalation. Furthermore, subsequent addition of D2O to this system resulted in significant deuterium labeling at the Si atom (43% D), suggesting the formation of a potassium silylide within the system, which can undergo hydrolysis with water. Importantly, no deuterium incorporation was observed in 1a with D2O in the absence of the catalyst (Fig. 5b). This discovery further confirms the essential presence of the iridium catalyst for the formation of four-membered silacyclic-iridium intermediate. A series of deuterium labeling experiments were then conducted under standard conditions using HBpin. During the work-up stage, deuterium was introduced into the reaction mixture of 1a and HBpin, resulting in 68% Si–D incorporation in the final product 2a (Fig. 5c). This observation confirms that the final product 2a is derived from a potassium silyl precursor. To further elucidate the mechanistic details of the Ir-catalyzed C–H borylation process, two kinetic isotope effect (KIE) experiments were performed. In the first experiment, parallel reactions were carried out using Ph3SiH (1a) and Ph3Si–D (1a–d1) with HBpin, yielding a KIE value of 1.29 (Fig. 5d). In the second experiment, parallel reactions were conducted using Ph3SiH (1a) and (C6D5)3Si-H (1a–d15) with HBpin, revealing a KIE value of 1.52 (Fig. 5e). These findings suggest that the cleavage of the Si–H bond and the ortho C–H bond are not the rate-determining steps in the reaction54.

Fig. 5: Deuterium labeling experiments.
figure 5

a Deuterium experiment of 1a–d in the absence of HBpin. b Deuterium incorporation of 1a in the absence of catalyst. c Deuterium experiment of 1a with D2O under standard conditions. d KIE experiment of 1a and 1a–d. e KIE experiment of 1a and 1a–15.

Full size image

DFT calculations

Based on the above results, density functional theory (DFT) calculations were conducted to elucidate the reaction pathway of triphenylsilane 1a in the absence of HBpin, aiming to gain a deeper understanding the formation of strained silametallacycles (Fig. 6). The catalytic cycle initiates with the dissociation of the dimeric iridium catalyst into the monomeric iridium intermediate INT1A. Following this, the Si−H bond of 1a undergoes oxidative addition to the iridium center of INT1A through the transition state TS2A, with an activation energy of 13.6 kcal mol−1 relative to the zero-energy point. In the absence of KOAc, the formed intermediate INT2A undergoes a reversible intramolecular oxidative addition of ortho C–H bond, forming a four-membered strained silametallacycles, designated as INT3A. In the presence of KOAc, intermediate INT2A activates the ortho C–H bond via a base-assisted internal electrophilic-type substitution (BIES) mechanism55,56. This process proceeds through the transition state TS3A-OAc, characterized by an activation energy of 18.1 kcal mol−1. Remarkably, this pathway is substantially more energetically favorable compared to the oxidative addition route through TS3A, which exhibits a significantly higher activation energy of 33.7 kcal mol−1. The reduced activation energy associated with the BIES mechanism underscores the efficiency of KOAc in facilitating the C–H activation, thereby promoting the formation of the strained silametallacyclic intermediate. Intermediate INT4A further undergoes reductive elimination via TS5A-Si with an activation energy of 25.3 kcal mol−1, transferring the hydrogen atom from silicon to the ortho-position of silicon center. This finding is consistent with experimental observations of deuterium incorporation.

Fig. 6: DFT-computed reaction pathways for the formation of strained silametallacycles in the absence of HBpin.
figure 6

a The energy profile for the generation of intermediate INT4A. b The calculated key transition states. M06-D3/6–311 + G(d, p)-SDD(Ir)/SMD(tetrahydrofuran)//B3LYP-D3/6–31 G(d)-LANL2DZ(Ir).

Full size image

We further investigated the competition reaction between silane 1a and HBpin with iridium catalyst (Fig. 7). The calculated energy barrier for the oxidative addition of the H–B bond to iridium is 18.8 kcal mol−1 relative to the energy of Ir-cat, significantly higher than that of the transition state TS2A for the oxidative addition of the H–Si bond (18.8 kcal mol−1 for TS2A-B vs 13.6 kcal mol−1 for TS2A). This energy difference suggests that silane 1a preferentially reacts with the iridium catalyst to form the strained silametallacyclic intermediate INT4A. This mechanistic pathway deviates notably from the previously directed aromatic C−H borylation triggered by Ir-Bpin species (see Supporting Information Figs. S1–6 for details)57,58,59.

Fig. 7: DFT-computed reaction pathways for the competition reaction between iridium catalyst and HBpin.
figure 7

a Calculated transition state for the oxidative addition of the H–B bond to iridium center. b Key transition state of TS2A-B.

Full size image

Based on the above results, we examined the pathway leading to the final products by DFT calculations (Fig. 8). A molecule of HBpin coordinates to the Ir center in INT4A, undergoing oxidative addition via transition state TS6A, with an energy barrier of 1.5 kcal mol−1, to form the Ir-H intermediate INT6A. Subsequently, reductive elimination and H2 release generate the penta-coordinated intermediate INT7A, with an activation energy of 2.3 kcal mol−1. A competitive pathway involving direct reductive elimination from INT6A to form the C–B bond was also considered but ruled out due to its prohibitively high energy barrier. The reductive elimination of INT7A proceeds through transition state TS8A, which has an activation energy of 20.3 kcal mol−1, leading to the formation of the desired C–B bond. In TS8A, the forming C–B bond length is 1.68 Å, indicating a well-defined transition state for this critical step. The generated intermediate INT8A then undergoes a transmetalation process with excess KOAc in the system, forming a silicate-potassium salt, pro-K. The structural information of transition state TS9A reveals the forming K–Si bond length to be 3.19 Å and the breaking Ir–Si bond length to be 3.45 Å. The activation energy barrier for this step is calculated to be 34.0 kcal mol−1 relative to INT6A, making it the highest energy barrier in the calculated free energy profile and thus the rate-determining step. During the work-up with D2O, the formed pro-K can further undergo hydrolysis, resulting in the formation of products 2a–d, as shown in the experimental results in Scheme 4a. Concurrently, the Ir intermediate INT10A, which is also generated, participates in an oxidative addition with another silane 1a, followed by a BIES-type C−H bond activation. This sequence of events culminates in the generation of the strained silametallacyclic intermediate INT4A, thereby completing the catalytic cycle efficiently.

Fig. 8: DFT investigations for the reaction pathways (M06-D3/6–311 + G(d, p)-SDD(Ir)/SMD(tetrahydrofuran)//B3LYP-D3/6–31 G(d)-LANL2DZ(Ir)).
figure 8

a The further transformation of intermediate INT4A. b The calculated key transition states.

Full size image

Discussion

In conclusion, we have successfully developed a highly efficient and practical method for the synthesis of mixed-metalloid products through C−H borylation, enabled by iridium catalysts. This breakthrough addresses several long-standing challenges in the field of C−H activation, particularly the catalytic formation of strained benzo-fused metallacycles and the use of metalloid directing atom. The dual functionality of the formed mixed-metalloid compounds is a significant advantage, as it enables the synthesis of a wide range of valuable molecules with high versatility. The findings represent a significant advancement in organic synthesis, opening new avenues for the creation of complex and functionalized organic molecules with improved efficiency and selectivity.

Methods

General procedure for C−H borylation of hydrosilanes

For hydrosilanes 1a–1h

In a glove box, a 25 mL dried Schlenk tube equipped with a stirring bar was charged with silane (0.5 mmol), [Ir(cod)OMe]2 (8.3 mg; 0.0125 mmol, 2.5 mol%), dtbpy (6.7 mg; 0.025 mmol, 5 mol%), AcOK (123 mg; 1.25 mmol, 2.5 equiv.) and HBpin (160 mg; 1.25 mmol, 2.5 equiv.). The Schlenk tube was then removed from the glove box into air, and THF (1.5 mL, super dry) was added under a nitrogen flow. The mixture was heated to 125 °C and stirred for 6 h. The solvent was removed under reduced pressure, and the residue was purified by silica gel column chromatography to afford the products 2b–2h.

For hydrosilanes 1i–1z

In a glove box, a 25 mL dried Schlenk tube equipped with a stirring bar was charged with silane (0.5 mmol), [Ir(cod)OMe]2 (16.6 mg; 0.025 mmol, 5 mol%), dtbpy (13.4 mg; 0.05 mmol, 10 mol%), AcONa (102.5 mg; 1.25 mmol, 2.5 equiv.), HBpin (19.2 mg; 0.15 mmol, 30 mol%) and B2pin2 (190.6 mg; 0.75 mmol, 1.5 equiv.). The Schlenk tube was then removed from the glove box into air, and THF (1.5 mL, super dry) was added under a nitrogen flow. The mixture was heated to 150 °C and stirred for 6 h. The solvent was removed under reduced pressure, and the residue was purified by silica gel column chromatography to afford the products 2i–2z.

General procedure for C−H borylation of hydrogermanes

For hydrogermanes 3a–3k

In a glove box, a 25 mL Schlenk tube equipped with a stirring bar was charged with [Ir(cod)OMe]2 (6.64 mg; 0.01 mmol; 5 mol%), 2,2′:6′,2″-terpyridine (139.8 mg; 0.6 mmol; 3 equiv.), and AcOK (49 mg; 0.5 mmol; 2.5 equiv.). Then, the Schlenk tube was removed from the glove box into air. Germane (0.2 mmol), HBpin (64 mg; 0.5 mmol; 2.5 equiv.) and THF (3 mL) were added to the Schlenk tube under a nitrogen flow. The reaction mixture was then heated to 115 °C and stirred for 24 h. The solvent was removed under reduced pressure, and the residue was purified by silica gel column chromatography to afford the products 4b–4k.