Iron-catalyzed sequential hydrosilylation

Abstract

Highly regio-, diastereo- and enantioselective iron-catalyzed sequential hydrosilylation of o-alk-n-enyl-phenyl silanes with alkynes is reported for various 5-, 6-, and 7-membered benzosilacycles in 60-94% yields with up to 95:5 rr, 95:5 dr, and 99% ee. Chiral fully carbon-substituted silicon-stereogenic benzosilacycles could also be obtained via triple hydrosilylation reactions. The unique electronic effect of ligands is observed while adjusting the regioselectivity and enantioselectivity in hydrosilylation reactions. A possible mechanism has been proposed by variable time normalization analysis (VTNA) and H/D exchange experiment.

Benzosilacycles are of potential interest in drug discovery and materials science, and they are often synthesized through transition-metal cyclizations. Here, the authors report iron-catalyzed sequential hydrosilylations of alkynes to access various 5-, 6-, 7- and 10-membered benzosilacycles in high yield and selectivity.

Introduction

Silasubstitution strategy was regarded as an innovative strategy to develop new materials and pesticides, and has found interesting applications in many fields (Fig. 1A)^1–16. A series of 5-, 6-, and 7-membered benzosilacycles had been constructed using transition metal catalysis, such as Rh^17–29, Pd^30–37, Ru³⁸, Ir³⁹, Sc⁴⁰, and Ni⁴¹ etc. (Fig. 1B). It should be noted that neither the catalytic cyclization for the synthesis of benzosilacycles containing two Si-H bonds which could be further derivatized, nor the construction of vicinal chiral carbon and silicon-centers in benzosilacycles has been reported. Although intramolecular hydrosilylation of alkenes containing mono-substituted silanes could be proposed to deliver the benzosilacycles containing two Si–H bonds which could undergo further sequential reactions⁴², there were several challenges: (1) The side reactions, such as hydrogenation, intermolecular hydrosilylation, and over functionalization of Si–H bonds, could significantly affect the production of benzosilacycles containing two Si–H bonds. (2) Compared to the C–C bond, the transition state was more flexible due to the longer C–Si bond which prompted more hardship in differentiating two enantiotopic faces. (3) For the sequential reactions, the difficulty to achieve the compatibility and control the sequence was increased in the presence of the multiple active Si–H bond and similar unsaturated C–C bonds in one pot. Additionally, although iron was the most earth-abundant transition metal⁴³, neither the iron-catalyzed hydrosilylation for the construction of benzosilacycles nor sequential hydrofunctionalization have been reported until now. Here, we reported an iron-catalyzed sequential hydrosilylation of n-alkenyl-phenyl silanes with alkynes to afford 5-, 6-, and 7- membered various benzosilacycles in good yields with excellent diastereo- and enantioselectivitiy, and sequential reactions in one pot for the construction of fully carbon-substituted silicon center, as shown in Fig. 1C. In this process, while changing the electronic effects on ligands, the regioselectivity and enantioselectivity of hydrosilylation reactions could be controlled.

Fig. 1. Benzosilacycles and their synthesis.

A Silasubstitution strategy, B Metal-catalyzed construction of benzosilacycles, C This work: Construction of 5-, 6-, 7-membered benzosilacycles by iron catalysts. The green colour is about the construction of 5-membered benzosilacycles, blue colour is aboult the construction of 6-membered benzosilacycles and purple colour is about the construction of 7-membered benzosilacycles. The pink colour is about “Si”.

Results: the iron-catalyzed sequential hydrosilylation to construct benzosilacycles

Results 1: the construction of 5- and 6-membered various benzosilacycles benzosilacycles

At the beginning of the screening, simple (2-vinylphenyl)silane (1a) and 1,2-diphenylethyne (2a) were chosen as model substrates. Lithium tert-butoxide and toluene were used as a mild activator and solvent, respectively. The reaction using bidentate bipyridine iron dichloride complex did not provide hydrosilylation products (entry 1, Table 1). The PDI ligand promoted the intramolecular hydrosilylation to afford 3a′ in 35% yield without sequential hydrosilylation product 3a (entry 2). Both Pybox and OIP ligands could promote the sequential reaction to deliver 3a in 44% and 54% yield, respectively. Increasing the steric hindrance on oxazoline and imine could improve the reactivity (entries 5 and 6). The reaction using 1.2 equiv of 1a afforded 3a in 96% yield (entry 7). Finally, the standard conditions were identified as using silane 1a (1.2 equiv), 1,2-diphenylethyne 2a (0.5 mmol), Fe-3 (5 mol%), and LiOtBu (15 mol%) in a solution of toluene (0.5 M) stirring at room temperature for 24 h. It should be noted that the catalyst Fe-3 could distinguish the intermolecular hydrosilylation and intramolecular hydrosilylation, and also raise the activity of the reactions. More details are shown in supplementary information (Table S2)

Table 1.

Optimizations for sequential hydrosilylation

Entry	Catalyst	Yield of 3a (%)a	Yield of 3a′ (%)	Yield of 3a′′ (%)
1	L1-Fe	–	–	–
2	L2-Fe	–	35	–
3	L3-Fe	44	–	3
4	Fe-1	53	–	–
5	Fe-2	79	–	–
6	Fe-3	84	–	–
7b	Fe-3	96 (90)c	–	–

^aReaction conditions: 1a (0.5 mmol), 2a (0.5 mmol), cat. (0.025 mmol), LiOtBu (0.075 mmol), toluene (1 mL).

^bUsing 1.2 equiv. of 1a.

^cIsolated yield in the parentheses.

Under the optimized conditions, the substrate scope was explored in Fig. 2A. Reactions of various diaryl alkynes containing ethers, halogens, heterocycles, diheteroaryl and dialkyl alkynes efficiently proceeded to afford desired anti-Markovnikov products 3a–3o in 68-93% yields. The 5-methyl and 4-chloro group on aryl silane could be tolerated to access 3q and 3r in 85 and 81% yield, respectively. Time course is shown in supplementary information (SI Figs. 1 and 2)

Fig. 2. The scope of 2-vinylphenyl silanes and 2-allylphenyl silanes.

^aStandard conditions: 1a (0.60 mmol), 2a (0.50 mmol), cat. Fe-3 (0.025 mol), LiOtBu (0.075 mmol), toluene (1 mL). ^bStandard conditions: 4a (0.3 mmol), 2a (0.3 mmol), cat. Fe-4 (0.015 mmol), LiOtBu (0.045 mmol), toluene (0.6 mL). ^cStandard conditions: 4a (0.30 mmol), 2a (0.30 mmol), cat. Fe-5 (0.015 mmol), LiOtBu (0.045 mmol), toluene (0.6 mL). ^dIsolated yield in the parentheses. ^eYields were determined by ¹H NMR using TMSPh as an internal standard. A The scope of 2-vinylphenyl silanes; B The scope of 2-allylphenyl silanes.

As shown in supporting information Table S4, the reaction using Fe-7 and LiOtBu in a solution of toluene at room temperature for 24 h afforded the mixed product in 78% yield with 50:50 rr. While using Fe-16 with minor hindrance of oxazoline, only 22% yield and 50:50 rr are provided. While adjusting the steric hindrance of imine could realize the control of Markovnikov selectivity to deliver the 5-membered benzosilacycle 5a in 91% yield with >95:5 rr and 92:8 dr. Then the use of Fe-17 which was modified with the electronic donating group could increase the anti-Markovnikov regioselectivity to 15:85. The use of more sterically hindered on oxazoline Fe-5 could improve the regioselectivity to afford 6-membered benzosilacycle 6a in 75% yield with 95:5 rr (Fig. 2B).

Results 2: The construction of 6- and 7-membered various benzosilacycles benzosilacycles

When (2-(but-3-en-1-yl)phenyl)silane (7a) was used as a substrate, the asymmetric iron-catalyzed intramolecular hydrosilylation could occur smoothly to deliver 6-membered dihydro-benzosilacycle 8a. As shown in Table 2, the use of Fe-8 afford the product with 79% ee. While electronic donating group was added to the ligand, 90% ee was provided. Electronic effects on the ligand might have a positive impact on enantioselectivity of intramolecular hydrosilylation reactions. The reactions of substrates with 4-methory, 3-fluoro, and 4-TBSO- groups were pushed on smoothly to afford products 8b–8d in 87–90% yields with 93:7 to >95:5 rr, and 90–97% ee (Fig. 3A). The absolute configuration was consistent with that previously reported by our group⁴³.

Table 2.

Optimizations for intramolecular hydrosilylation

Entry	Catalyst	x	Base	Solvent	Time (h)	Yieldb (%)	ee of 8ac (%)
1a	Fe-1	5	NaOtBu	THF	12	60	58
2	Fe-6	5	NaOtBu	THF	12	62	60
3	Fe-7	5	NaOtBu	THF	12	66	61
4	Fe-8	5	NaOtBu	THF	12	62 (83:17 rr)	75
5	Fe-8	5	NaBHEt3	THF	12	90	74
6	Fe-8	5	NaBHEt3	Neat	12	95	79
7	Fe-5	5	NaBHEt3	Neat	12	73	90
8	Fe-5	2	NaBHEt3	Neat	12	90	90
9	Fe-5	2	NaBHEt3	Neat	3	91	91
10	Fe-9	2	NaBHEt3	Neat	3	90	87

^aStandard conditions: 7a (0.50 mmol), cat. Fe (0.025 mol), NaOtBu (0.075 mmol), THF (1 mL).

^bYields were determined by ¹H NMR using TMSPh as an internal standard.

^cThe enantiomeric excess ratio (ee) was determined by chiral HPLC.

Fig. 3. The scope of chiral benzosilacycles.

^aStandard conditions: 7a (0.50 mmol), cat. Fe-5 (0.010 mmol), NaBHEt₃ (0.030 mmol). ^bStandard conditions: 7a (0.30 mmol), cat. Fe-5 (0.006 mmol), NaBHEt₃ (0.018 mmol), cat. Fe-3 (0.015 mmol), 2a (0.30 mmol), LiOtBu (0.045 mmol), toluene (0.6 mL). ^cStandard conditions: 11a (0.30 mmol), L (0.006 mmol), NaBHEt₃ (0.018 mmol), THF (0.6 mL). ^dStandard conditions: 7a (0.36 mmol). ^eYields were determined by ¹H NMR using TMSPh as an internal standard. ^fIsolated yield in the parentheses. A The scope of benzosilacycles with two Si–H bonds; B The scope of benzosilacycles with vicinal chiral carbon and silicon-centers; C The scope of 7-membered benzosilacycles.

After the first intramolecular hydrosilylation completed, the complex Fe-3 (5 mol%), LiOtBu (15 mol%), alkynes (1 eq.) and toluene (0.5 M) were added in the system. The sequential reactions could be carried out to provide the 6-membered monohydro-benzosilacycle 10a with vicinal chiral carbon and silicon centers in 65% yield with >95:5 rr, >95:5 dr, and 93% ee (Fig. 3B)^44,45. The sequential reactions of substrates with 4-methory, 3-fluoro, and 4-TBSO- groups were carried out smoothly to afford 10b–e in 60–70% yields with excellent regio-, diastereoselectivity (>95:5 rr, >95:5 dr), and 86-93% ee. The substrates with heterocycle and alkyl alkyne could also be pushed on efficiently. For the unsymmetrical alkyne but-1-yn-1-ylbenzene, sequential reaction afforded the α-syn-selective product 10j in 75% yield. The complex molecular empagliflozin ntermediate was also easily transformed into the corresponding chiral silanes 10k with uncompromised 80% yield with 93% ee. The absolute configuration of the resulting carbon and silicon-stereogenic center in 10l was unambiguously determined by an X-ray diffraction. Excitedly, 7-membered carbon-stereogenic benzosilacycle 12a could also be constructed in 73% yield with 99% ee while using (2-(pent-4-en-1-yl)phenyl)silane (11a) in the presence of iron catalyst (Fig. 3C).

The reaction of 1a with 2a could be carried out in a gram scale to afford 3a with 1.28 g in 82% yield (Fig. 4A, eq.1). The reaction of 7a could be performed on gram scale, providing 6-membered benzosilacycles 10a with 1.06 g in 62% yield (Fig. 4A, eq.2). Treatment of 3a with 1-ethynyl-4-methoxybenzene in the presence of a dicobalt carbonyl N-heterocyclic carbene [(IPr) Co₂(CO)₇] (IPr = 1,3-di(2,6-diisopropylphenyl)imidazol-2-ylidene) complex provided 5-membered heterocycles derivatives 14a in 74% yield (Fig. 4Beq.3)^46–49. Terminal alkene, terminal alkyne and α, β unsaturated ketone could be used as partners to convert to fully carbon-substituted silicon-stereogenic heterocycles derivatives (15a, 16a and 17a) in the presence of Karstedt’s catalyst (Fig. 4B, eq.4–6)^50–52. It is interesting that three stereocenters benzosilacycles 17a could be obtained, which was quite difficult to be afforded by using other methods.

Fig. 4. Gram scale reaction and further derivatization.

^aeq.1 Gram scale reaction of 1a, ^beq.2 Gram scale reaction of 7a, ^ceq.3 further derivatization of 3a, ^deq.4-6 further derivatization of 10a. A. Gram scale reaction of 1a and 7a. B further derivatization of 3a and 10a.

Deuterium-labeling experiments were conducted to explore the mechanism in those asymmetric hydrosilylation processes as shown in Fig. 5. The silicon-deuterated product 19a-d₃ or 21a-d₃ was afforded by the asymmetric sequence hydrosilylation of unsaturated C-C bonds with silane-d₃ (18a or 20a) catalyzed by iron catalyst. The result suggested that insertion was regioselective and irreversible (eq.7 and eq.8)⁵³. The fact that the product 21a′ with deuterium incorporation into the α-carbon was not generated indicated the impossibility of a Fe-H or Fe-D intermediate in the catalytic cycle, because facile H/D exchange between C_α–H and Fe–D would otherwise occur through reversible olefin insertion into the Fe-D bond^54,55. When intramolecular hydrosilylation with a 1:1 mixture of silane-d₃ (20a) and silane 7b was carried out and followed by the intermolecular hydrosilylation, silicon-deuterated products 21a and 21b were obtained (in Supporting Information). The construction of silicon-deuterated product 21b and the next deuterium-labeling experiment in eq.9, giving 24a-d₂ and 25a-d₂, indicated that Si-Fe species might participate in the catalytic cycle.

Fig. 5. Mechanistic experiments to explore the active intermediate.

eq.7-8 Deuterium-labeling experiments.

Mechanistic interrogation was realized through kinetic studies using variable time normalization analysis (VTNA)^56–62. VTNA was used to determined the reagent orders of the reaction components, from which we hypothesized the turnover-limiting step. This method was used to determined the kinetic order of the reaction components in the second step. Two different slopes were observed in Fig. 6a, where the kinetic profile of the reaction with two different iron catalyst concentrations was shown. When a first-order factor of catalyst concentration (t [Fe]¹) was used (Fig. 6c), two reaction progress curves failed to overlap. While using a 0.5-order correlation (t [Fe]^0.5), two reaction progress curves overlapped (Fig. 6b), indicating that the kinetic order for iron is a 0.5-order rather than the first order. It was speculated that the active intermediates of iron catalyst might exist in the form of dimers⁶³. It also meant that the rate of reaction was relatively insensitive to changes in the concentration of the reactants, but not completely unaffected. Next, a superior reaction was observed when increasing substrate 8a loading (Fig. 6d–f), corresponding to a first order. Same with substrate 8a, the result that a first order for substrate 2a was obtained from VTNA (Fig. 6g–i). The observation of [8a]¹ and [2a]¹ suggested that one molecular 8a and one molecular 2a were involved in the turnover limiting step. More information were shown in (SI Figs. 3–5)

Fig. 6. variable time normalization analysis.

a–c VTNA analysis of catalyst, d-f VTNA analysis of silane, g–i VTNA analysis of alkyne.

Based on the above mechanisms, the proposed mechanism was shown in Fig. 7. The Fe–Si species (A) was proposed through reducing Fe-5 by NaBHEt₃ and silane^64–67. Then alkene inserted in Fe–Si bond to generate iron alkyl species (B), which wonder go δ-bond metathesis with silane to regenerate species (A) and afford the hydrosilylation product (8a). The Fe–Si (C) obtained from (8a). Because the kinetic order for iron is a 0.5-order rather than the first order, the active intermediates Fe–Si (C) might exist in the form of dimers, which was proposed and this proposal would be tested in our future work. Then alkyne inserted into Fe–Si bond to generate iron alkenyl species (D), which reacted with silane to regenerate Fe–Si species (C) and afforded the final product (10a). More studies were continuously under-going in our laboratory to gain an accurate understanding of the mechanism. The mechanism of Fe(0) species cannot be completely ruled out.

Fig. 7. The proposed mechanism in sequential hydrosilylation.

The proposed mechanism starts from intramolecular hydrosilylation, followed by intermolecular hydrosilylation.

The primary models to predict the regiochemical outcome about the construction of 5- and 6-membered benzosilacycle were shown in Fig. 8. The primary models PM(A) and (B) provided similar possibility of Markovnikov and anti-Markovnikov hydrosilylation products 5a and 6a. While increasing the steric hindrance of imine, the primary models PM(C) was disfavored. While electronic donating groups added to ligand, electronegativity on Fe turned over. Then the primary models PM(E) become the dominant conformation and anti-Markovnikov product was provided. In the primary models, electronic donating groups on the ligand may exist in a resonant form, resulting in the electronegativity of Fe turning over. This resonant form might alter the size of ligand cavity, thus making the transition state more compact and improving the enantioselectivity of the reaction Fig. 8. More information were shown in (SI Figs. 6 and 7).

Fig. 8. The primary models to predict the regiochemical outcome^a.

A–H The primary models to predict the regiochemical outcome to explain the selectivity in the construction of 5- and 6-membered benzosilacycles.

Discussion

In this work, we developed a mild iron-catalyzed system for the construction of 5-, 6-, and 7-membered benzosilacycles with highly regio-, diastereo- and seteroselectivity. Eight kinds of benzosilacycles including silacycles with vicinal chiral carbon and silicon-centers were successfully synthesized. The regioselectivity of hydrosilylation reactions could be controlled by iron catalyst step by step, and the kinetic order of the reaction components in the second step was determined by using variable time normalization analysis (VTNA). The development of more asymmetric iron-catalyzed reactions is ongoing in our laboratory.

Methods

A 25 mL Schlenk flask equipped with a magnetic stirrer and a flanging rubber plug is dried with flame. When cooled to ambient temperature, it is transferred into glove box. Then Fe-3 complex (0.025 mmol, 5 mol%), LiOtBu (0.075 mmol, 15 mol%), 1a (0.6 mmol, 1.2 equiv), alkynes (0.5 mmol), toluene (1 mL, 0.5 M) are added sequentially. The mixture is stirred at room temperature for 24 h. Then the resulting solution is transferred into room and quenched with 10 mL of PE and filtered through a pad of silica gel, washed with PE/AcOEt (10/1) (or other suitable solvent) (5 × 10 mL). The combined filtrate is concentrated and the yield is monitored by ¹H NMR analysis using TMSPh as an internal standard. The resulting solution is purified by flash column chromatography to give the corresponding product.

Author contributions

Conceptualization, Z.L.; Methodology, X.W, J.J.Z., D.Y.W. L.D.; Investigation, X.W., and J.J.Z.; Writing-original draft, Z.L. and X.W.; Writing-review & editing, Z.L., and X.W.; Funding Acquisition, Z.L. and L.D.; Supervision, Z.L. and L.D.

Peer review

Peer review information

Nature Communications thanks Liangbin Huang, and the other, anonymous, reviewer for their contribution to the peer review of this work. A peer review file is available.

Data availability

The authors declare that the data Supplementary the findings of this study are available within the paper and its Supplementary Information file. The X-ray crystallographic coordinates for the structure of compound 10I reported in this study have been deposited at the Cambridge Crystallographic Data Centre (CCDC), under deposition number 2383336. These data can be obtained free of charge from https://www.ccdc.cam.ac.uk. The Cambridge Crystallographic Data Centre via The experimental procedures and characterization of all new compounds are provided in the Supplementary Information. Data supporting the findings of this manuscript are also available from the authors upon request.

Competing interests

The authors declare no competing interest.

Supplementary information

The online version contains supplementary material available at 10.1038/s41467-025-59364-3.