Umpolung α-Silylation of Cyclopropyl Acetates via Low-Temperature Catalytic C–C Activation

Abstract

We report a redox-neutral, catalytic C–C activation of cyclopropyl acetates to produce silicon-containing five-membered heterocycles in a highly region-and chemoselective fashion. The umpolung α-selective silylation leading to dioxasilolanes is opposed to contemporary β-selective C–C functionalization protocols of cyclopropanols. Lewis base activation of dioxasilolanes as α-silyl carbinol equivalents undergoes the unconventional [1,2]-Brook rearrangement to form tertiary alcohols. Notably, mechanistic studies indicate that an electrophilic metal-π interaction harnessing highly fluorinated${\text{Tp}}^{{{\text{(CF}}_{\text{3}}\text{)}}_{\text{2}}}\text{Rh(nbd)}$catalyst permitted a low-temperature C–C activation.

Graphical Abstract

INTRODUCTION

Carbon–carbon (C–C) functionalization approaches offer new synthetic strategies involving traditionally unimaginable bond disconnection to build important molecular architectures rapidly for chemistry, biology, and medicine.^1–14 Current strategies of C–C σ-bond metal insertion and succeeding functionalization have centered on new C–C formation. For example, carbonylation,^9,14,15 an insertion of alkene,¹⁶ alkynes,^15,17 and C=N,¹⁸ and others^7,19 to C–C single bond has been well demonstrated. Emerging interests in synthetic chemistry are to develop new synthetic strategies to construct carbon–heteroatom (e.g., B, F, Si) bonds. Among them, carbon–silicon (C–Si) bond formation is attractive because of the environmental sustainability and ready diversification nature of organosilanes, aiming at rapid preparation of a range of bioactive molecules and new materials.^20,21 Despite their established synthetic benefits, catalytic C–C silylation has been underdeveloped, primarily because of a lack of efficient catalytic protocol to effect the crucial bond scission and formation. Recently, Murakami and co-workers have demonstrated the first C–C silylation reactions through unique Pd-catalyzed cross metathesis of C–C and C–Si bonds within cyclobutanones and silacyclobutanes, respectively, which builds structurally unique silacycles.^22–24

Another important issue in C–C activation^14–17 has arisen in regioselective bond functionalization. For example, conventional C–C activation approaches of cyclopropanol derivatives^25–29 are limited to the β-C–C functionalization to provide ketones 2 via metallohomoenolate 3 (Figure 1a).³⁰ However, to the best of our knowledge, catalytic α-functionalization of cyclopropanols has not been reported. Herein, we report umpolung α-C–C silylation of cyclopropyl acetates with highly fluorinated ${\text{Tp}}^{{{\text{(CF}}_{\text{3}}\text{)}}_{\text{2}}}\text{Rh(nbd)}$catalyst. The redox-neutral, catalytic C–C activation provides dioxasilolanes 6 in a highly region- and chemoselective fashion at low temperature (e.g., room temperature) (Figure 1b).

Figure 1.

Catalytic C–C activation of cyclopropanol derivatives: Complementary α- and β-regioselective functionalization.

DISCUSSION

To address the synthetic challenge associated with regioselective C–C silylation of cyclopropanols, we envisioned Rh-catalyzed redox-neutral, umpolung α-selective C–C silylation of cyclopropyl acetates (Figure 1b). Specifically, loading of metal catalyst onto cyclopropyl hydrosilyl acetals 5, accessed via Ir-catalyzed ester hydrosilylation,³¹ followed by cross σ-bond metathesis between C–C and Si–M bonds in 7 provides regioselective C(α)–Si bond formation and leads to dioxasilolanes 6 via 8 (Figure 2a). The resulting dioxasilolanes 6 can serve as stable, α-silylcarbinol equivalents. Although they can be accessed by a silyl anion addition approach to ketones, the potential setbacks with this stoichiometric approach include a 1,2-silyl migration from carbon to oxygen (Brook rearrangement) of the resulting oxyanion after the nucleophilic addition,³² enolization, and preferential 1,4-silyl nucleophile addition to α,β-unsaturated ketones to afford β-silyl ketones.^33–35

Figure 2.

Proposed mechanism for region- and chemoselective C–C silylation.

Furthermore, chemoselective metal insertion into C–C bonds in the presence of largely abundant C–H bonds in close proximity is challenging.¹⁷ To address this issue, our strategy for chemoselective C–C silylation is predicated on the transition-state-guided reaction design, harnessing analysis of each putative cyclometalated structure generated through either C–H or C–C activation. We recognized the importance of metalated transition state (TS) to activate and functionalize specific chemical bonds successfully and selectively. Our hypothesis is that the highly chemoselective Rh-catalyzed C–C activation of 5 is achievable with adopting the five-membered TS geometry, depicted in 7 (Figure 2a). In contrast, two proximal arene and cyclopropyl C–H activation^36,37 requires unfavorable eight-membered TS for 9 and seven-membered TS for 10, respectively (Figure 2b).

As a proof-of-concept of performing region- and chemoselective catalytic C–C silylation, we prepared phenyl-cyclopropyl acetate 4a (Table 1). This substrate includes proximal arene and cyclopropyl C–H bonds and cyclopropyl C–C bond. To assess the feasibility of the α-selective C–C silylation, 4a was subjected to a sequence of Ir-catalyzed ester hydrosilylation and Rh-catalyzed C–C silylation conditions in a single pot. A series of supporting ligands including pentamethylcyclopentadienyl (Cp*) (entry 1) and tris-(pyrazolyl)borate (Tp) (entries 2–7) screened showed that the C–C silylation was viable to afford a pair of constitutional isomers (6a and 11a) (Table 1). Among them, the electron-poor Tp ligand, ${{\text{[Tp}}^{{{\text{(CF}}_{\text{3}}\text{)}}_{\text{2}}}]}^{-}$led to the highest yield of 6a (entry 4).^38,39 Among a number of cyclic olefin promotors having different ring strains (entries 4–7),⁴⁰ the norbornene was found to be the most effective and generally applicable for the C–C silylation (see the Supporting Information). In any case the competing proximal arene and cyclopropyl C–H silylations were not observed, as rationalized by the TS analysis shown in Figure 2.

Table 1.

Reaction Optimization of Rhodium-Catalyzed C–C Silylation^a

entry	supporting ligand	promotorb	6a (%)b	11a (%)b
1	[Cp*]−c	nbe	73	6
2	${\text{Tp}}^{{\text{(CH}}_{\text{3}}\text{)}}{}^{{}_{\text{2}}}\text{K}$	nbe	70	9
3	${\text{Tp}}^{{\text{(CH}}_{\text{3}}\text{)Ph}}\text{Na(THF)}$	nbe	76	11
4	${\text{Tp}}^{{{\text{(CH}}_{\text{3}}\text{)}}_{\text{2}}}\text{Na(THF)}$	nbe	78	8
5	${\text{Tp}}^{{{\text{(CH}}_{\text{3}}\text{)}}_{\text{2}}}\text{Na(THF)}$	nbd	37
6	${\text{Tp}}^{{{\text{(CH}}_{\text{3}}\text{)}}_{\text{2}}}\text{Na(THF)}$	cod	20
7	${\text{Tp}}^{{{\text{(CH}}_{\text{3}}\text{)}}_{\text{2}}}\text{Na(THF)}$	coe	18

^aConditions: (i) cyclopropanoacetate 4a (0.2 mmol), [lr(coe)₂ Cl]₂ (0.1 mol %), H₂SiEt₂ (3 equiv), THF (0.33 M). (ii) [Rh(nbd)Cl]₂ (0.8 mol %), supporting ligand (1.6 mol %), promotor (2 equiv), THF (0.33 M).

^bIsolation yield. The major diastereomer of 6a is shown; in all cases, the minor diastereomer of 6a was observed as determined by ¹H NMR and GC-MS spectrometry analyses of the crude reaction mixture. A diastereomeric ratio of 11a was not determined.

^c[RhCp*Cl₂]₂ was used. nbe, norbornene; nbd, norbonadiene; cod, 1,5-cyclooctadiene; coe, cis-cyclooctene.

${{\text{[Tp}}^{{{\text{(CF}}_{\text{3}}\text{)}}_{\text{2}}}]}^{-}$coordinates to Rh(I) in bidentate fashion as evident from the X-ray structure of ${\text{Tp}}^{{{\text{(CF}}_{\text{3}}\text{)}}_{\text{2}}}\text{Rh(nbd)}$ is shown in Figure 3.⁴¹ The use of isolated ${\text{Tp}}^{{{\text{(CF}}_{\text{3}}\text{)}}_{\text{2}}}\text{Rh(nbd)}$ for the C–C silylation under essentially identical conditions led to 6a in 76% yield and 12:1 diastereoselectivity. A substoichiomeric amount of nbe was indeed required for efficient reaction (cf., 39% yield without nbe), inferring that nbe is not innocent for an overall catalytic cycle.

Figure 3.

Isolation and X-ray structure of ${\text{Tp}}^{{{\text{(CF}}_{\text{3}}\text{)}}_{\text{2}}}\text{Rh(nbd)a}$ catalyst for redox-neutral, catalytic α-C–C silylation of cyclopropyl hydrosilyl acetal 5a.

Unexpectedly, we found that the ${\text{Tp}}^{{{\text{(CF}}_{\text{3}}\text{)}}_{\text{2}}}\text{Rh(nbd)}$ catalyst permitted the room temperature C–C silylation of aryl-, alkenyl-, and alkynyl-substituted cyclopropyl hydrosilyl acetals (Figure 4).⁴² The lower yield observed in the C–C silylation of 5a at room temperature is attributed to partial instability of 5a during the extended reaction time (24 h). In contrast to the sluggish kinetics in the formation of 6a, the reactions of the alkene and alkyne-substituted cyclopropyl hydrosilyl acetals, 5b and 5c, respectively, were remarkably fast to afford 6b and 6c within 40 and 30 min at room temperature, with good yields (87% and 83%, respectively) and excellent diastereoselectivity (both >20:1). We rationalize that increasing electrophilicity of the rhodium center by highly fluorinated tris(pyrazolyl)borate ligand facilitates the donor–acceptor interaction between the neighboring π-donor and the electrophilic rhodium and thus develops the Rh–(C–C) agostic interaction^8,43 for the ready C–C activation, leading to facile C(α)–Si bond formation. When alkyl-substituted cyclopropyl hydrosilyl acetals in the absence of a neighboring π donor were subjected to the catalytic conditions, the C–C silylation reactions ultimately failed and the starting materials were recovered. This result indicates the importance of the Rh-π interaction for the successful catalytic C–C activation of cyclopropyl hydrosilyl acetals.

Figure 4.

Room temperature, C–C silylation with electron-poor ${\text{Tp}}^{{{\text{(CF}}_{\text{3}}\text{)}}_{\text{2}}}\text{Rh}$ catalyst.

Under the optimized reaction conditions for a single-pot, redox-neutral C–C silylation, a range of a new class of silicon-containing heterocycles 6 were prepared from 4 with moderate to excellent yield and diastereoselectivity (Table 2). Dioxasilolanes 6 presented in Table 2 were chromatographically stable and a range of common functional groups including alkenes (6b, 6d), alkyne (6c), amine (6h), ethers (6i–6p), halides (6m, 6n), trifluoromethyl (6o), 3,5-dimethyl (6p), acetal (6q), naphthalene (6r), furan (6s), thiophene (6t), and indole (6u) were tolerated either at rt or 100 °C. However, a reaction with allene (6e) gave a complex mixture. The reactions with substrates holding substituted cyclopropyl ring also proceeded to furnish 6v–6x with good yield and excellent diastereoselectivity. In these cases, we observed the preferential activation of a more kinetically accessible C–C bond. An NOE experiment with 6v revealed that the diastereoselection was consistent with the proposed TS model 7 (Figure 2a and see the Supporting Information), which permitted high diastereoselectivity.

Table 2.

Scope of Redox-Neutral Catalytic C–C Silylation of Cyclopropyl Acetates^a

^aYields are for isolated material over a single-pot, two-step sequence from 4 (0.2 mmol). Diastereomeric ratio (dr) was determined by ¹H NMR spectroscopy and GC-MS spectrometry analyses of the crude reaction mixture.

^b[Rh(nbd)Cl]₂ (5 mol %), ${\text{Tp}}^{{{\text{(CF}}_{\text{3}}\text{)}}_{\text{2}}}\text{Na(THF)}$ (10 mol %), norbornene (10 mol %), THF (0.33 M), rt.

^c1 mmol of 4a was used.

The cyclic silyl acetals 6 are chromatographically stable, yet willing acceptors of organometallic nucleophiles.^31,44,45 Upon a nucleophilic attack to 6 leading to a putative, pentacoordinated silicon species 14, the high thermodynamic driving force drives them rapidly and irreversibly to 15 in equilibrium with a strained, pentacoordinated species 16 (Figure 5a). This spring-loaded nature makes 6 suitable for rapid diversification to tertiary alcohol derivatives. To demonstrate this aspect, we first performed the [1,2]-Brook rearrangement of 6a as α-silylcarbinol equivalents.³² Specifically, a ring opening of 6a with MeLi releases acetaldehyde, which can engage in additional MeLi, and a spontaneous ring closure of 15 produces 16. Upon addition of electrophiles and raising of the temperature, the reactions furnished tertiary alcohols capped with a diethylmethylsilyl group, 12 and 13, respectively. When the reaction was quenched at −78 °C, α-silylcarbinol 17 was produced (88% yield). This protocol was applied to a complex molecular setting, where estrone-derived cyclopropyl acetate 18 was converted into silyl ether 19 (85% yield, a three-step single pot from 18) (Figure 5b).

Figure 5.

Synthetic applications of dioxasilolanes 6.

To gain insight into the mechanism of this reaction, we conducted a series of mechanistic experiments. First, we performed the Hammett study to understand a linear free-energy relationship of a redox-neutral C–C silylation mechanism (Figure 6a). The rate acceleration by electron-withdrawing groups was observed in the process. Specifically, the ρ value of 0.80 indicates no appreciable accumulation of the electron density in the transition state of the C–C cleavage step. The results infer that the M-π coordination is not likely the turnover-determining step, but increasing the electron-withdrawing nature of the aryl moiety would rather facilitate the cleavage of the neighboring cyclopropyl C–C bond. Labeling experiments with tetradeuterated cyclopropyl hydrosilyl acetal and a fully deuterated phenyl group-substituted hydrosilyl acetal conclude that no obvious cyclopropyl and arene C–H cleavage occurs in the course of the C–C silylation (Figure 6b,c, respectively). We then conducted the kinetic isotope effect (KIE) study with 5a-CP-H₄ and 5a-CP-D₄ in a single vessel (Figure 6d). The reaction afforded inverse secondary KIE. These results along with the Hammett studies described above suggest that the turnover-determining step is likely the C–C cleavage.

Figure 6.

Mechanism studies. (a) Hammett plot. (b) Deuterium-labeling studies with tetradeuterated cyclopropane. (c) Deuterium-labeling studies with a pentadeuterated phenyl group. (d) Kinetic isotope effect (KIE) study.

On the basis of our preliminary mechanistic studies along with literature precedents,^46–49 the proposed mechanisms accounting for formation of 6 and 11 proceeds through a common bis-silyl rhodium intermediate C (Figure 7). Oxidative addition of cyclopropyl hydrosilyl acetal 5 to ${\text{Tp}}^{{{\text{(CF}}_{\text{3}}\text{)}}_{\text{2}}}\text{Rh(nbd)}$ generates Rh-silyl hydride A, which was identified as a major species observed through ¹H NMR spectroscopy (see the Supporting Information).^48,49 Norbornene is indispensable in producing C efficiently; olefin migratory insertion to form B, followed by oxidative addition of 5 and reductive elimination, could generate norbornane (observed by ¹H NMR) and C, which is in equilibrium with D. Considering the bonding similarities between the Cp* and Tp ligands, structurally similar Cp*Rh(SiEt₃)₂ species were proposed by Berry and co-workers in the context of C–H silylation.⁴⁸ Once the rhodium-π interaction that facilitates a tightly organized geometry and thus would develop putative C–C agostic interaction with the metal center are established within D, direct cross metathesis of Si–Rh and C–C within D, which is a turnover-determining step, can give rise to the intermediates E (major) and F (minor). We rationalize that F requires a significantly distorted geometry for the putative intramolecular bond exchange process. Because the steric bulk from the proximal, fully substituted carbon center is present next to the metal, σ-bond metathesis between hydrosilyl acetal 5 and E furnishes 6, where the β-hydride elimination adduct G was not observed. The crossover experiment between protiosilyl acetal and deuteriosilyl acetal established that the final hydrogen transfer occurs intramolecularly (see the Supporting Information). On the other hand, the reaction between the alkylrhodium species F holding the bidentate RhTp and 5 leads to a minor constitutional isomer 11 (H was not observed). We made an observation on the alkene transposition adduct 20 (9% yield) along with 6d (79% yield) from 4d, suggesting that formation of the fully substituted alkylrhodium species 21 underwent [1,3]-rhodium shift²⁵ to produce less hindered 22 and thus F is responsible for formation of dioxasilepanes 11.

Figure 7.

Proposed mechanism.

CONCLUSION

In summary, we have developed umpolung α-silylation of cyclopropyl acetates via redox-neutral catalytic C–C activation to produce silicon-containing five-membered heterocycles with high region- and chemoselectivities. Synthetic applicability of the methods was demonstrated through the rapid preparation of tertiary alcohols vis Lewis base-mediated unconventional [1,2]-Brook rearrangement of the resulting chromatographically stable dioxasilolanes. Finally, preliminary mechanistic studies suggest that the electrophilic metal center present in highly fluorinated ${\text{Tp}}^{{{\text{(CF}}_{\text{3}}\text{)}}_{\text{2}}}\text{Rh(nbd)}$ catalyst is a key to the facile, low-temperature C–C silylation. Studies toward an enantioselective variant of this reaction are underway and will be reported in due course.