Iron-Catalyzed Asymmetric Hydrosilylation of Vinylcyclopropanes via Stereospecific C-C Bond Cleavage.

An iron-catalyzed highly anti-Markovnikov selective, enantioselective hydrosilylation of vinylcyclopropanes with PhSiH3 was reported for the preparation of valuable chiral allylic silanes via stereospecific C-C bond cleavage. Simultaneously, difficultly prepared chiral VCPs could be also obtained with moderate to excellent enantioselectivity via this kinetic resolution pathway. The chiral Z-allylic silanes could be converted to various chiral allylic derivatives. A possible mechanism via an iron-silyl species was proposed based on experimental and computational studies.

Introduction

Vinylcyclopropane (VCP) has been widely used in organic synthesis, particularly as a versatile organic synthon in cycloadditions (Lautens et al., 1996, Reissig and Zimmer, 2003, Rubin et al., 2007, Aïssa, 2011, Jiao and Yu, 2013, Wang and Yu, 2015, Fumagalli et al., 2017). Although the regio- and stereoselective ring-opening hydrofunctionalization reactions of VCPs could theoretically afford various potentially useful products, poor regioselectivity with eight possible regioisomers extremely limits their utility. Traditionally, in order to control the regioselectivity, gem-substituted electron-deficient groups have to be used to control the 1,4- or 3,4-regioselectivity via the cleavage of more substituted carbon-carbon bond (Crossley et al., 2016, Souillart and Cramer, 2015, Nairoukh et al., 2017, Wu and Zhu, 2019, Shigehisa et al., 2013, Sartori et al., 1983, Burgess, 1987, Sebelius et al., 2005, Sumida et al., 2008, Li et al., 2009, Dieskau et al., 2012, Wu et al., 2015, Zell et al., 2016, Meyer et al., 2017, Wang et al., 2019). To the best of our knowledge, asymmetric ring-opening hydrofunctionalization reactions of vinylcyclopropanes are still quite limited. Using donor-acceptor (D-A) vinylcyclopropanes, Krische and co-workers reported the iridium-catalyzed asymmetric alkylations with aldehydes or alcohols as electrophiles, affording the 3,4-regioselective hydrocarbonation products (Moran et al., 2011). Trost group recently disclosed the palladium-catalyzed asymmetric 1,4-hydrocarbonation of D-A vinylcyclopropanes with 3-substituted indoles as nucleophiles (Trost et al., 2018) (Figure 1A). Recently, our group reported an iron-catalyzed asymmetric 5,1-hydroboration of non-donor-acceptor (non-D-A) vinylcyclopropanes via iron-hydride intermediate (Chen et al., 2017). However, overstoichiometic amounts of VCPs were used to be converted to chiral boronic esters with less than 90% ee, and the ee values of recovered VCPs were extremely poor. So, the development of various regio- and enantioselective hydrofunctionalizations of vinylcyclopropanes via stereospecific C-C bond cleavage for the synthesis of chiral products is highly desirable.

Figure 1

Highly Enantioselective Hydrofunctionalizations of VCPs via C-C Bond Cleavage

Allylic silane is one of the most useful allylic reagents for organic transformations, such as Hosomi-Sakurai reaction (Masse and Panek, 1995, Fleming et al., 1997, Barbero and Pulido, 2004, Díez-Poza and Barbero, 2017). Chiral allylic silanes were used to be synthesized using stoichiometric chiral reagents via diastereoselective or stereospecific transformations (Sparks and Panek, 1991, Panek and Clark, 1992, Suginome et al., 1996, Bourque et al., 2007, Binanzer et al., 2010, Aggarwal et al., 2011). It is still a challenge for the catalytic enantioselective synthesis of chiral allylic silanes, which used to focus on the formation of chiral silicon-substituent carbon center (Hayashi et al., 1982, Hofstra et al., 2018, Wu et al., 2010, Ohmura et al., 2006, Da et al., 2018, Shintani et al., 2007, Lee et al., 2012, Kacprzynski et al., 2007, Sang et al., 2018, Wen et al., 2019).

Herein, we reported an iron-catalyzed 1,5-selective asymmetric hydrosilylation of vinylcyclopropanes via stereospecific C-C bond cleavage, affording chiral allylic silanes with excellent enantioselectivity and recovering VCPs with moderate to excellent enantioselectivity (Figure 1B).

Results and Discussion

Initially, we chose the 1-phenyl-2-(1′-phenyl)vinylcyclopropane 1a as a model substrate (Table 1). The reaction of 1a with 1.5 equivalent of PhSiH3 using 5 mol % of iron precatalyst L1·FeCl2 and 15 mol % of NaBHEt3 in a solution of dioxane (0.5 M) at ambient temperature for 24 h was carried out to afford the desired product 2a in 6% yield (entry 1), whereas the reaction with Ph2SiH2 did not afford the hydrosilylation product (entry 2). The observation of 2a illustrated the possibility of the generation of iron-silyl species. In our previous studies, free hydroxyl group was found to be tolerated in cobalt-catalyzed alkyne hydrosilylation in which cobalt-silyl species was proposed (Guo and Lu, 2016, Zuo et al., 2016). Ge and coworkers found that the addition of phenol could inhibit the Z/E-isomerization in Co-catalyzed alkyne hydrosilylation (Teo et al., 2017). These examples demonstrated that the addition of free hydroxyl group could promote the formation of cobalt silyl species. However, the proposed iron silyl species has not been reported for the tolerance of free hydroxyl group (Cheng et al., 2018, Hu et al., 2018, Chen et al., 2018, Obligacion and Chirik, 2018). With 15 mol % of water, the reaction afforded 2a in an increasing yield (12%), which indicated that the addition of free hydroxyl group could also promote the formation of iron silyl species (entry 3). Using the phenol as an additive, the reaction was conducted for 2 h to afford 2a in 28% yield with 98% ee, as well as 1a in 65% recovery with 34% ee (entry 4). By further investigation of various phenols, 3-chlorophenol was found to accelerate reaction efficiently (entries 5–8). The reactivities were increased by increasing the steric hindrance of the group on oxazoline (Me, Bn, Cy) (entries 8–10). However, when L4·FeCl2 was used as a precatalyst, the reaction was inhibited, which might be due to the overlarge steric hindrance (entry 11). When the 4-MeO-phenyl group was introduced into para-position of aniline, the reaction using L5·FeCl2 as a precatalyst could be further accelerated to access 2a in 48% yield with 13/1 Z/E and 98% ee, simultaneously with 28% recovery of chiral 1a in 86% ee (entry 12). It should be noted that some dienes and hydrogenation products were observed as side products, which decreased the recovery of starting materials (see in Supplemental Information). The standard conditions were identified as 0.5 mmol of racemic vinylcyclopropane, 0.75 mmol of silane, 5 mol % of L5·FeCl2, 15 mol % of NaBHEt3, and 15 mol % of 3-chlorophenol in 1 mL of dioxane for 2 h.

Table 1

Optimization of the Reaction Conditions

Entry	L	Additives	Yield of 2a (%)a	ee of 2a (%)a	Recovery of 1a (%)a	ee of 1a (%)a
1b	L1	–	6	–	89	–
2b,c	L1	–	0	–	48	–
3b	L1	H2O	12	–	86	–
4	L1	C6H5OH	28	98	65	34
5	L1	4-MeOC6H4OH	18	97	76	21
6	L1	4-ClC6H4OH	30	98	62	38
7	L1	3-ClC6H4OH	35	97	59	41
8	L1	2-ClC6H4OH	15	98	82	14
9	L2	3-ClC6H4OH	32	96	60	37
10	L3	3-ClC6H4OH	41	98	45	63
11	L4	3-ClC6H4OH	2	–	90	–
12d	L5	3-ClC6H4OH	48	98	28	86

PMP, 4-methoxyphenyl. See also Table S1.

Using 1a (0.5 mmol), PhSiH3 (0.75 mmol), L·FeCl2 (5 mol %), NaBHEt3 (15 mol %), additives (15 mol %) and dioxane (0.5 M).

Yields were determined by 1H NMR analysis based on 1a and ee value was determined by chiral HPLC. Unless noted, the ratio of Z/E was around 20/1.

24 h.

Using Ph2SiH2 instead of PhSiH3, affording 14% yield of dienes without silicon group and 37% yield of hydrogenation products.

Z/E = 13/1.

With the optimal conditions in hand, the substrate scope was explored in Table 2. The scope of various substituents on alkene was first explored. The allylic silanes 2b-2j with functional groups (methoxyl, methylthio, fluoro, chloro, bromo) at the para-, meta-, or ortho-position on the phenyl ring could be obtained in 39%–47% yields and 92%–98% ee, and chiral VCPs 1b-1j could be observed in 15%–40% recoveries with 63%–97% ee. 2-Naphthyl substrate 1k was also suitable, providing 2k in 47% yield with 95% ee and 11% recovery of chiral 1k in 99% ee. The reaction of alkyl-substituted VCPs did not occur. The scope of substituents on cyclopropanes was next examined. VCPs bearing electron-donating (1l, 1m) or electron-withdrawing (1n-1s) groups at the different positions on the phenyl ring gave the corresponding products in 40%–48% yields with 78%–98% ee, and chiral VCPs 1l-1s were observed in 13%–39% recoveries with 64%–97% ee. Additionally, 1-naphthyl, 2-naphthyl, 2-furyl, and 2-thienyl VCPs 1t-1w were suitable for the catalytic system, affording 2t-2w in 40%–47% yields with 64%–98% ee and chiral 1t-1w in 21%–43% recoveries with 64%–96% ee. Owing to the sterically hindered groups, the selectivities of 1s and 1u were decreased. Notably, the reactions of 1o-, 2o-, 3o-alkyl substrates afforded 2x-2ac in 44%–50% yields with 80%–95% ee and chiral 1x-1ac in 37%–46% recoveries with 74%–96% ee.

Table 2

Substrate Scope

Reaction conditions: 1 (0.5 mmol), PhSiH3 (0.75 mmol), L5·FeCl2 (5 mol %), NaBHEt3 (15 mol %), 3-ClC6H4OH (15 mol %) and dioxane (0.5 M), r.t. Isolated yield of Z-isomer. The ratio of Z/E and recovery of 1 were determined by 1H NMR analysis and ee value was determined by chiral HPLC. Unless noted, the ratio of Z/E was better than 10/1.

aZ/E = 9/1.

bZ/E = 8/1.

cZ/E = 4/1.

To expand the utility of this strategy, a broad range of aryl and alkyl silanes were tested. Various aryl silanes were used in the reaction, affording 2ad-2ag in 38%–45% with 96%–99% ee and chiral 1a in 28%–41% recoveries with 62%–94% ee. When alkyl silanes were employed, the transformation also performed well, affording 2ah-2ak in 37%–42% yields with 87%–98% ee and 12%–31% recoveries with 78%–94% ee. In general, VCPs bearing aryl or alkyl groups proceeded through a kinetic resolution (KR) pathway to deliver the desired chiral allylic silanes in good to excellent yields, moderate to excellent stereoselectivities and excellent enantioselectivities, and recover chiral VCPs in moderate to excellent yields with moderate to excellent enantioselectivities (Keith et al., 2001, Vedejs and Jure, 2005, Muller and Schreiner, 2011, Miller and Sarpong, 2011, Gao et al., 2014, Xiao et al., 2016, Hu et al., 2016, Das et al., 2017, Shimoda and Yamamoto, 2017, Jones et al., 2019, Brauns and Cramer, 2019, Zheng et al., 2019, Deng et al., 2019, Wu et al., 2019, Rajkumar et al., 2019).

To demonstrate the utility of this method, a gram-scale reaction could be carried out to afford 2ag in 41% yield with 14/1 Z/E, 95% ee and recover chiral 1a in 13% yield with 98% ee (Scheme 1A). The further derivatizations of products were illustrated in Scheme 1B. The Fleming-Tamao oxidation of 2a delivered chiral allylic alcohol 3 in 90% yield with 97% ee (Wen et al., 2019). The absolute configuration was confirmed by X-ray diffraction of 3. Obviously, the Hosomi-Sakurai reaction of 2a with selectfluor and meta-chloroperoxybenzoic acid gave 4 and 5 in 85% and 88% yields with 1.2/1 and 6/1 dr, respectively (Hayashi et al., 1984, Thibaudeau and Gouverneur, 2003, Tredwell et al., 2008). Moreover, the hydrosilylation of n-hexyne with silane 2a accessed a tertiary silane with silicon-stereocenter 6 in 88% yield with 7/1 dr (Cheng et al., 2017). The cross-coupling reaction of 2a with phenylmagnesium bromide afforded 7 in 90% yield with 98% ee (Hirone et al., 2010).

Scheme 1

Gram-Scale Reaction and Further Transformations

Two control experiments were conducted to elucidate the possible reaction pathway (Scheme 2, eq. 1 and 2). The reaction of ( under standard conditions afforded ( in 94% yield with >99% ee (Scheme 2, eq. 1), whereas the reaction of ( under standard conditions using ·FeCl2 as precatalyst afforded 2a in 12% yield with a 2/3 ratio of Z/E and 46% recovery of (. The diene ( was observed as a side product, which also demonstrated the reason of poor recovery (scheme 2, eq. 2). These two control experiments well indicated the phenomenon of “matched and mismatched” between the substrate and iron precatalyst.

Scheme 2

Control Experiments

Inspired by the control experiments and the previously reported literatures on metal-silyl species (Wei and Darcel, 2019, Randolph and Wrighton, 1986, Lee et al., 2010, Lee and Peters, 2011, Tondreau et al., 2012, Zhang et al., 2014, Jia and Huang, 2016) and hydrosilylation (Chen et al., 2015, Chen et al., 2019, Xi and Lu, 2016, Guo et al., 2017, Guo et al., 2019, Cheng et al., 2019, Sun and Deng, 2016, Du and Huang, 2017, Trost and Ball, 2001, Mo et al., 2014, Ding et al., 2015, Du et al., 2016, Schuster et al., 2016, Pappas et al., 2016, Wang et al., 2017, Gribble et al., 2017, Liu et al., 2018, Wen et al., 2018, Zhan et al., 2018, Hu et al., 2019), we proposed that the iron-silyl species might promote this new 1,5-selective type of hydrosilylation of VCPs, which was different from the hydroboration of VCPs through iron-hydride species. The possible mechanism was proposed (Figure 2). The iron-silyl species is generated from OIP·FeCl2 in the presence of NaBHEt3, 3-chlorophenol, and silanes (Teo et al., 2017). In the matched catalytic cycle, 1,2-insertion of ( into the iron-silicon bond delivers the tertiary alkyl iron species A. Subsequent β-carbon elimination generates the primary alkyl iron species B. Iron species B undergoes δ-bond metathesis with hydrosilane to access the product and regenerate iron silyl species. The mismatched catalytic cycle was discussed in Supplemental Information.

Figure 2

Proposed Mechanism

The proposed mechanism is also supported by density functional theory (DFT) calculations. The computed free energy changes of the most favorable pathway for iron-catalyzed 1,5-selective hydrosilylation of VCP 1a with OIP ligand L2 are shown in Figure 3. From the (OIP)Fe-silyl active species int1, alkene coordinates to allow the subsequent alkene insertion via TS3. This generates the tertiary alkyl-iron intermediate int4, which undergoes the proposed β-carbon elimination through TS5 to give the primary alkyl-iron intermediate int6. From int6, the 1,5-selective hydrosilylation product 2a is produced through a σ-bond metathesis via TS7, accompanied by the regeneration of int1. Based on the free energy changes of the whole catalytic cycle, the irreversible alkene insertion determines the overall regioselectivity of hydrosilylation.

Figure 3

Free Energy Diagram for the Most Favorable Reaction Pathway for 1,5-Selective Hydrosilylation of 1a

We next explored the regioselectivity of alkene insertion, and the energies and optimized structures of the competing insertion transition states (TS3 and TS8) are shown in Figure 4. TS3 is 13.8 kcal/mol more favorable than TS8, which is consistent with the experimental observations of the exclusive 1,5-selective hydrosilylation. The distortion/interaction analysis (Bickelhaupt and Houk, 2017) is performed to further elucidate the origins of regioselectivity. The distortion energies ΔEdist are the energy penalty associated with the geometric change from the ground state to the corresponding geometry in alkene insertion transition state. ΔEint reflects the strength of interaction between the two distorted fragments in the transition state. ΔEdist-sub is the leading cause that differentiates the two transition states. This is due to the steric repulsions between the inserting VCP substrate and OIP ligand in TS8. The closet H-H distances between the substrate and ligand are highlighted in Figure 4. Therefore, the sterically demanding OIP ligand forces the anti-Markovnikov insertion of VCP, leading to the excellent regioselectivity of hydrosilylation.

Figure 4

Optimized Structures, Energies, and Distortion/Interaction Analysis of the Regioisomeric Alkene Insertion Transition States

Energy barriers in kcal/mol are compared with int1 and 1a. Trivial hydrogens are omitted for clarity.

Conclusion

In summary, we have first developed an iron-catalyzed asymmetric anti-Markovnikov-selective hydrosilylation of vinylcyclopropanes via stereospecific C-C bond cleavage. The addition of 3-chlorophenol has been found to promote the formation of proposed iron silyl species to accelerate the reaction. The transformation with a wide range of VCPs and silanes proceeds through a kinetic resolution pathway to generate the chiral allylic silanes in excellent ee and recover chiral VCPs in moderate to excellent ee. The chiral allylic silanes can be used in the Hosomi-Sakurai reaction, hydrosilylation, and cross-coupling reaction. An iron-silyl species was proposed based on experimental and computational studies. The studies on novel difunctionalization of VCPs are underway in our laboratory.

Limitations of Study

In this study, the scope of substrate is well explored but the vinylcyclopropanes with alkyl-substituents on alkene did not work. It is still a challenge for the highly enantioselective earth abundant metal-catalyzed hydrosilylation of 1,1-disubstituted alkenes with alkyl substituents.

Methods

All methods can be found in the accompanying Transparent Methods supplemental file.