Deoxygenative Functionalization of Alcohols and Carbonyl Compounds via Electrochemical Reduction

Abstract

Oxygen-containing functional groups are prevalent motifs in natural products and feedstock chemicals, but direct methods for their deoxygenative transformation remain rare due to the difficult cleavage of the strong C–O bond. Here, we develop a general activation strategy that employs hydrosilanes as activating reagents for alcohols, carbonyls, and esters to afford a common silyl ether intermediate. Electrochemical reduction of the in-situ generated silyl ether results in C–O cleavage to afford a carbanion, which reacts with a number of electrophiles for the construction of C–Si, C–B, C–Ge, and C–Sn bonds.

Graphical Abstract

Despite the abundance of oxygen-containing molecules, no unified approach exists that can activate various oxidation states of oxygenated functional groups for deoxygnenative functionalization. Here, we disclose a tandem, one-pot hydrosilylation–electroreduction sequence that enables the conversion of alcohols, aldehydes, ketones, and esters into nucleophilic reagents widely used in cross-coupling chemistry.

Introduction

Deoxygenation of Alcohols, Carbonyls and Esters

Oxygenated functional groups, including alcohols, aldehydes, ketones, and esters, are ubiquitous in natural products, pharmaceuticals, and feedstock chemicals.^[1] Given their wide abundance, the development of methods for the selective functionalization of these motifs is an attractive objective in synthetic chemistry. The inherent strength of the C–O and C=O bonds^[2] often necessitates the activation of these groups prior to the desired deoxygenative transformation (Figure 1a). Furthermore, owing to their distinct properties, deoxygenative transformations of alcohols, ketones, and esters have traditionally relied on distinct activation strategies, while a unified approach that is amenable to all oxidation states remains elusive.

Figure 1.

Background and introduction. (a) Classical approaches to the deoxygenative functionalization of alcohols, carbonyls and esters. (b) Synthetic strategies for organosilane synthesis. (c) Our previous electrochemical approach to deoxygenative borylation. (d) This work: Deoxygenative functionalization of alcohols, carbonyls, and esters through a silane-mediated activation strategy.

Deoxygenative functionalization of alcohols has emerged as a powerful strategy for rapid complexity generation.^[3–6] In addition to classical substitution chemistry, redox-based approaches have been widely studied in recent years. Oxidative approaches inspired by the canonical Mitsunobu reaction via generation of cationic P(V) or phosphine radical cation intermediates have been successfully employed toward photo- and electrochemical C–O activation.^[7–9] Barton-type strategies using oxalates and xanthates have similarly enabled deoxygenative functionalization.^[10–13] More recently, the use of N-heterocyclic carbene reagents has been developed for alcohol activation via radical intermediates.^[14] In each of these cases, single-electron oxidation results in C–O bond cleavage, which is rendered irreversible by the formation of a strong P=O or C=O bond, or the release of gaseous byproducts.

Among reductive transformations, pioneering work by Lam and Markó demonstrated the feasibility of alcohol deoxygenation via in-situ ester formation, followed by chemical or electrochemical reduction.^[15–16] Recently, Wickens developed a mild photochemical strategy using similar ester activating groups.^[17] Other activating groups, including phosphinate esters,^[18] carbonates,^[19] oxalate esters,^[20–22] and N-hydroxyphthalimide (NHPI) oxalates^[23–24] have since been explored. In these cases, the alcohol is treated as a precursor to a C-centered radical, while strategies to convert alcohols to carbanions remain rare.^[25–29]

Likewise, the deoxygenative functionalization of carbonyls through sequential nucleophilic additions^[30–33] or carbene generation^[34] are well-developed strategies, but the use of carbonyls as precursors to carbon nucleophiles remains relatively underexplored. Wolff-Kishner reduction of hydrazones represents the most common method for formal carbanion generation from ketones and aldehydes.^[35–36] Similarly, the Shapiro reaction generates a vinyl anion through decomposition of hydrazone.^[37–38] While useful, the harsh conditions required often limit their application as general deoxygenative functionalization strategies.

Unlike the above cases, esters are almost universally treated as electrophiles for acylation reactions or reduced to aldehyde, alcohol or methyl groups using metal hydrides.^[39–41] Serial nucleophilic additions, as in the Kulinkovich reaction and others employing organomagnesium or organolithium reagents, also leverage the ester as an electrophile.^[42] A few reports by Shono^[43–44] and others^[45] demonstrate the electroreductive dimerization of esters to diketones and acyloin-type products via radical intermediates, but reductive methods for the conversion of esters to carbanions remain unexplored.

Organosilanes: Challenges and Opportunities

Organosilanes have emerged as a useful class of compounds in organic synthesis, materials chemistry, and medicine.^[46–49] In recent years, there has been particular interest in the development of organosilicon chemistry due to their utility as bench-stable reagents for cross-coupling. Classical methods employing organosilanes, such as Hiyama-Denmark coupling^[50–51] and Hosomi-Sakurai allylation,^[52] have been complemented by a variety of electrochemical^[53–54] and base-mediated approaches.^[55–56] Organosilanes, particularly benzylic and allylic variants, can often be used as masked carbon-nucleophiles in a variety of C–C bond forming reactions beyond cross-coupling.^[57–58] Despite their wide importance, traditional methods to synthesize organosilanes often employ harsh organolithium or organomagnesium reagents. Therefore, the generation of organosilanes under mild conditions with readily available precursors would constitute a significant advance toward increasing the accessibility of organosilicon reagents. To this end, the deoxygenative silylation of alcohol derivatives has emerged as an attractive strategy. Despite numerous advances by Martin,^[59–60] Oestreich,^[61] Shishido,^[62–63] and others,^[64–68] to the best of our knowledge all of these approaches employ expensive or difficult-to-access silyl nucleophiles, or noble metal catalysts (Figure 1b).

The electrochemical synthesis of organosilanes has been disclosed in early reports via electroreduction of benzylic or allylic chlorides or acetates in the presence of chlorosilanes.^[69–71] These results hint at the possibility of a general method for deoxygenative silylation. Indeed, our group previously demonstrated a few examples of the deoxygenative silylation of allylic alcohols and ethers via deep reductive electrochemistry, although yields and selectivity were modest.^[72] Early attempts to extend this reaction to a general deoxygenative functionalization strategy were hampered by competing proton reduction and protonation of the in-situ generated carbanion by alcoholic protons.

To overcome this limitation, here, we explore the use of silyl ethers as activated intermediates for C–O bond cleavage. Silyl ethers are a well-known class of alcohol derivatives with wide utility in organic synthesis as protecting groups due to their structural tunability and stability to nucleophilic attack.^[73] While silyl ethers are commonly used to suppress undesired reactivity by masking reactive alcohol groups, few examples use these reagents to promote deoxygenative transformations. Recent work by Newman has found that silyl ethers may be used as electrophiles in Suzuki-Miyaura couplings under Bi(III) catalysis.^[74–75] Furthermore, aryl silyl ethers have been employed as electrophiles in Ni-catalyzed cross-coupling reactions for the construction of Csp²–Si bonds.^[76–77] In contrast, the deoxygenative transformation of alkyl silyl ethers into carbanion intermediates remains comparatively underdeveloped, with a seminal example employing Li metal to facilitate reductive C–O cleavage.^[78] We envisioned that the direct redox control offered by electrochemistry would provide a general and selective means for the reductive cleavage of the strong C–O bond.^[79–81]

Here, we disclose a method for the electroreductive cleavage of C–O bonds in alcohols, aldehydes, ketones, and esters via in-situ generation and activation of silyl ethers, affording a carbanion for the construction of C–Si as well as C–B, C–Ge, and C–Sn bonds.

Results and Discussion

Reaction Design

Our group previously reported the electrochemical deoxygenative borylation of alcohols, aldehydes, and ketones using pinacolborane (HBpin) as both an in-situ activating reagent and electrophile (Figure 1c).^[82] In that system, HBpin reacts with alcohols and carbonyls to afford a common borate ester intermediate, which can be reduced electrochemically to cleave the strong C–O bond and afford a carbanion. This carbanion can then be quenched by excess HBpin to afford an organoboron product. While it is convenient to use a single reagent as both a substrate activator and a functional group source, this approach limits the type of reaction products to alkylboronic esters. Furthermore, control experiments revealed that the borate ester intermediate itself could serve as an electrophile, complicating the reaction pathway and making it challenging to introduce other functional groups using an additional electrophilic agent. To address these shortcomings, we sought to develop an alternative activation strategy that would be orthogonal to the subsequent carbanion trapping step.

After screening various activating groups, we found silyl ethers to be promising candidates. Unlike our previous method wherein only the pinacol-derived boryl group can serve as an effective activating group, silyl groups can be more readily modified, and they are more stable toward nucleophilic attack at Si. Notably, silyl ethers could be generated from hydrosilanes (Figure 1d), enabling the conversion of alcohols (A), carbonyls (B), and even esters (C) into a unified intermediate (D). Upon cathodic reduction, (D) would undergo C–O bond cleavage to afford carbon-centered radical (E). When radical (E) is adjacent to a carbanion-stabilizing group such as an aryl or allyl system, it can be further reduced to the corresponding carbanion (F), which could then be captured by a silyl electrophile to forge a desired C–Si bond (G). Importantly, the silyl ether intermediate (D) and any residual hydrosilane are only weakly electrophilic, and thus do not serve as competing electrophiles in the carbanion trapping step. This feature allows for the use of a variety of electrophiles to enable diverse deoxygenative transformations under a single mechanistic manifold.

Reaction Optimization

At the outset of our investigation, we sought to optimize conditions for the electroreductive deoxygenative functionalization of tert-butyldimethylsilyl (TBS) ethers (Figure 2a), which are among the most common protecting groups for alcohols. After initial screening, we obtained optimal conditions to convert silyl ether 1 into benzylsilane 2 in 88% yield using Me₃SiCl as the electrophile, ⁿBu₄NClO₄ as the electrolyte, graphite as the cathode and Mg as the sacrificial anode, under a constant current of 10 mA in a mixed solvent of THF and tris(pyrrolidino)phosphine oxide (TPPA) (entry 1).^[83] Control experiments showed that both Zn and Al anodes were inferior to Mg (entries 2 and 3), while other cathodic materials, such as stainless steel and Pt, could be used in place of graphite (see Supporting Information, Table S2). The inclusion of TPPA as a cosolvent is not essential to the success of the reaction (entry 4). When TPPA is removed, however, the cell voltage is much higher and electrode fouling was observed. TPPA has previously been used as a highly polar additive and substitute to toxic HMPA in the electrochemical reduction of arenes and epoxides.^[84–85] DME, an ethereal solvent that has previously been shown to prevent passivation of Mg electrodes by solubilizing Mg²⁺ salts,^[86–87] afforded 2 in diminished yield (entry 5), while the more polar DMF failed to afford any product (entry 6). The reaction could not be reproduced using Mg powder as the reductant in lieu of passing current (entry 7). Furthermore, applying 0.3 F of charge (0.15 reducing equivalents) to activate the Mg anode surface prior to stirring the solution without current resulted in minimal product formation, which precludes the mechanistic possibility of direct silyl ether reduction by metallic Mg (entry 8).

Figure 2.

(a) Optimization of electrochemical conditions Reactions were conducted on 0.5 mmol scale. ^{a 1}H NMR yield. (b) Substrate limitation of TBS ether reduction. ^{b 19}F NMR yield. (c) Computed reduction potentials of a series of silyl ethers. ^c Calculated reduction potentials for cleavage of the C–O bond in THF at the B3LYP level of theory, versus ferrocene/ferrocenium (Fc/Fc+).

While these conditions are effective for the transformation of the model substrate, a tentative survey of other benzyl alcohols revealed unsatisfying functional group tolerance. For instance, when the TBS ether of 4-fluorobenzyl alcohol 3 was employed as substrate, benzylsilane 4 was formed in only 49% yield, with competing defluorination observed (Figure 2b). We reasoned that the introduction of electron-withdrawing alkoxy groups at silicon might enable a more selective process by decreasing the reduction potential required to activate the substrate. Inclusion of a single alkoxy group led to a promising yield increase from 49% to 69%. Substituting the silicon center with a second or third alkoxy group further increased the reaction yield to 72%. This trend was further investigated computationally (Figure 2c) by modelling the reduction potential for single-electron induced cleavage of the benzylic C–O bond. While trialkylsilyl ether Si-I was reduced only under deeply reducing conditions (computed E_red = −3.72 V vs. Fc/Fc⁺), the replacement of alkyl groups for alkoxy led to substantial increases in reduction potential (Si-II, E_red = −3.53 V; Si-III, E_red = −3.41 V; and Si-IV, E_red = −3.26 V). Although the trend provides important insight into the design of silane-based activating groups, it overestimates the potential required; in practice, the reduction potential of intermediate Si-III was measured to be −2.8 V vs Fc/Fc⁺ by cyclic voltammetry (See Figure S1). In the absence of any radical- or anion-stabilizing group, the computed reduction potential shifts more cathodic (E_red = −3.82 V vs Fc/Fc⁺ for the optimal silane), and the desired silylation product was not detected after electrolysis.

Finally, we developed an in-situ silyl ether generation method to improve the operational simplicity of the method. Previously, Oestreich^[88] and Grubbs^[89] demonstrated that alcohols could be protected with hydrosilanes under mild conditions. In our hands, Oestreich’s dehydrogenative coupling protocol allowed silyl ether 6 to be generated rapidly in THF in the presence of catalytic KO^tBu using dimethylmethoxyhydrosilane (DMMS) (Figure 3a) from the corresponding alcohol 5. The inclusion of catalytic base was essential (entry 2), while including the supporting electrolyte during the activation stage did not hinder the reactivity (entry 3). Importantly, these conditions are amenable to the hydrosilylation of carbonyls (7) and esters (8) by extending the reaction time and/or increasing the equivalents of silane (Figure 3b). This activation strategy has several limitations (Figure 3c). Carboxylic acid 10, its potassium salt 11, and anhydride 12 were unreactive in the reaction system with complete recovery of the starting material.

Figure 3.

Optimization of dehydrogenative silylation conditions. Reactions were conducted on 0.5 mmol scale. ^{a 1}H NMR yield. (b) Extension to carbonyl and ester hydrosilylation. (c) Functional groups which did not undergo hydrosilylation under the optimized conditions.

Reaction Scope

With optimal tandem activation–reduction conditions in hand, we finally sought to explore the scope of the transformation. A range of alcohols with varying degrees of substitution (i.e., primary, second, and tertiary) could be efficiently transformed into organosilanes (Scheme 1). Parent benzylsilane (13) in addition to those featuring alkoxy- (2, 16–18), fluoro- (4), thioether (14, 19), and Bpin (15) groups were obtained in high yield, although chloroarenes underwent dechlorination instead of deoxygenation (12). Aromatic heterocycles such as pyrazole (18) were tolerated, along with secondary (21) and tertiary (22) alcohols. Benzyl alcohols bearing ester and silyl ether groups were converted to the corresponding benzylsilane in good yields (23-24); these functional groups, however, were incompatible with the hydrosilane-based activation strategy, necessitating the synthesis and use of TBS ethers in the electrolysis (see Figure 2a for conditions). Furthermore, allylsilanes, which are widely employed as reagents in organic synthesis, were readily accessible from allylic alcohols. Geraniol was converted to the corresponding allylsilane in moderate isolated yield with exclusive selectivity for the terminal position (25). Tertiary allylic alcohols could be converted to the corresponding terminal allylsilanes by transpositional functionalization (26-29). By extending the electrolysis time and increasing the loading of hydrosilane, diols could be efficiently converted to bis-silanes or cyclosilanes via dual deoxygenative functionalization (31-32). Furthermore, alcohols could be converted to hydrosilanes and disilanes (33-34), which can be further elaborated for use in materials chemistry.^[90] Organogermanium (35) and organostannane (36) compounds could also be efficiently prepared, providing facile access to additional common synthons for cross-coupling chemistry.^[91] A scope of incompatible electrophiles can be found in the Supporting Information (Section 4).

Scheme 1.

Substrate scope of alcohols. Yields given are isolated unless noted. ^{a 1}H NMR yield. ^{b 19}F NMR yield. ^c Obtained from electrolysis of the isolated TBS ether. ^d DMMS (2.2 equiv.), electrolysis for 5.5 F/mol. ^e Me₂SiCl₂ (2 equiv) used as electrophile, electrolysis for 5 F/mol. ^f Electrophile (3.2 equiv) used in place of Me₃SiCl.

We next explored the substrate scope of deoxygenative silylation of carbonyls (Scheme 2). From the corresponding aldehyde, silane products with a diverse range of substituents such as methoxy (2, 40), methylthio (14), diphenylamino (37), and alkyl (38-39) could be prepared in good yield. The scalability of the reaction was demonstrated by preparing benzylsilane 40 at 7 mmol scale (1.4 g) in excellent yield. Analogous to the reaction of alcohols, various aliphatic and aromatic heterocycles such as carbamate-protected piperazine (41), alcohol-substituted piperidine (42), pyrazole (43), and pyridine (44) were all tolerated. Both aryl alkyl and diaryl ketones proved to be suitable substrates (45–49), including the fragrance celestolide (50). β-ionone was converted to the corresponding allylsilane, with high regioselectivity for the less hindered site (17:1 r.r.) (51). Furthermore, like alcohols, carbonyls could be functionalized with a range of electrophiles to afford benzyltriethylsilane 52, hydrosilanes 53–54 and organoborane 55 in high yield.

Scheme 2.

Substrate scope for the reductive silylation of aldehydes and ketones. Yields given are isolated unless noted. ^{a 1}H NMR yield. ^b Electrolysis conducted in a 1,2-dimethoxyethane (DME) solution of LiOTf as the electrolyte with a Mg anode and stainless steel cathode. MeOBpin as electrophile. See SI for full details.

We next surveyed benzoate esters as substrates (Scheme 3) and obtained benzylic silanes with ether (2) and heterocyclic groups (20, 41). In addition to methyl esters, an ethyl ester was also efficiently converted to the silane (56). Finally, a synthetic intermediate of the antidepressant vilazodone was transformed to silane 57, though limited solubility led to a low yield.

Scheme 3.

Substrate scope for the reductive silylation of esters. Yields are isolated unless otherwise noted. ^{a 1}H NMR yield.

Organosilanes are often used as bench-stable, masked carbon-nucleophiles in C–C bond formation in the presence of a base activator.^[55–58] To demonstrate the utility of our silane products, we developed a telescoped synthesis of naproxen methyl ester (60) starting from abundant ketone 58 (Scheme 4). Activation and subsequent electrochemical silylation affords benzylsilane 59, which, after filtration through silica, was subjected to carboxylation in the presence of CsF and methylation to afford desired 60 in 57% yield with one chromatographic step.^[92] We next explored the synthesis of an analogue of zatolmilast, which is currently in Phase III clinical trials for the treatment of Fragile-X syndrome.^[93] Coupling of benzylsilane 24 with cyanopyridine 61 using conditions reported by Bandar afforded the drug scaffold in 44% yield (62). Finally, a methylated derivative of the natural product combretastatin was synthesized in a single step by coupling of 40 with aldehyde 63 to afford 64 in 51% yield.

Scheme 4.

Further transformations of organosilane products. Yields are isolated.

Conclusion

Reductive electrochemistry continues to develop as a powerful strategy for the construction of C‒heteroatom bonds. We have disclosed a two-stage, one-pot protocol for the deoxygenative functionalization of alcohols, carbonyls, and esters. Key to the success of this strategy is the use of hydrosilanes as activating reagents to convert each of these functional groups into activated, non-electrophilic silyl ethers. Computation in addition to empirical screening provides a pathway for tuning the electronic structure of the hydrosilane activator to improve the functional group compatibility. Electrochemical reduction of the in-situ generated silyl ether results in cleavage of the strong C–O bond to afford a carbanion, which can be selectively quenched with a range of electrophilic trapping partners to yield a variety of useful cross-coupling partners from ubiquitous starting materials.

Supplementary Material

The authors have cited additional references within the Supporting Information.^[94–113]