Geminal Difunctionalizations of Aliphatic Ketones and Aldehydes Enabled by Thioacetal Activation and Deep Electroreduction

Abstract

In this work, we identify dithioacetals as synthetically accessible, redox-active, and traceless activating groups that enable the electroreductive transformation of aliphatic ketones and aldehydes. This strategy effectively converts these carbonyl compounds into dianion equivalents to undergo deoxygenative gem-difunctionalization with various electrophiles. Faradaic control was achieved over reaction chemoselectivity, allowing for the combinatorial construction of densely functionalized silanes, boronates, and germanes.

Carbonyls are prevalent functional groups in nature and ubiquitous across commodity chemicals, materials, and pharmaceuticals. Their central role as synthetic linchpins has profoundly shaped the development of organic synthesis. Canonical carbonyl functionalization reactionsincluding Grignard additions, Wittig olefinations, and metal hydride reductionsremain among the most widely practiced synthetic operations. In contrast, far fewer methods enable the net deoxygenative geminal difunctionalization of carbonyl compounds, a strategy that would convert planar carbonyl motifs into densely substituted, electronically distinct C­(sp³)-centers. Such methods would provide direct access to diverse chemical space from abundant carbonyl feedstocks, with broad implications for synthetic and medicinal chemistry.

Existing approaches for carbonyl difunctionalization most often treat carbonyl compounds either as formal dications undergoing successive nucleophilic addition or as dipolar synthons that engage in both nucleophilic addition and electrophilic trapping (Scheme A). By contrast, an umpolung strategy that renders carbonyls as formal dianions capable of successive electrophilic substitution would unlock fundamentally new disconnections. Recent contributions from König, Li, and our laboratory have established methods to generate a carbanion equivalent from carbonyl-derived intermediates by means of metal catalysis, photochemistry, or electrochemistry, enabling monofunctionalization (Scheme B, upper). Given the instability of unconjugated carbanions, existing methods primarily focus on the transformation of conjugated carbonyl compounds, with few examples capable of converting ketones to the corresponding carbanions. Moreover, a strategy enabling carbonyl groups to serve as dianion equivalents via successive activations has not yet been realized.

1. Background and Proposed Transformation.

Recently, our group reported the net difunctionalization of unactivated aldehydes and ketones through cyanophosphate activation, wherein cyanohydrin O-phosphates undergo cathodic reduction to generate stabilized nitrile anions competent in coupling with a second electrophile (Scheme B, lower). Consequently, this strategy requires installation of an anion-stabilizing group to facilitate successive electroreduction, incorporating the cyano group into the product. The substrate scope and accessible products are therefore intrinsically linked to the activation mode. To decouple the activation strategy from the synthetic target, we sought to develop an electroreductive cross-electrophile coupling platform that employs a traceless carbonyl activation mode capable of sustaining iterative reduction, enabling a diverse scope of geminal difunctionalizations of electronically unactivated ketones and aldehydes (Scheme C).

To this end, we first evaluated a series of aldehyde derivatives in a model electroreductive gem-disilylation reaction (Scheme ). Electrolysis of native benzaldehyde (A1) delivered gem-disilane 1 in only 24% yield, with significant competitive pinacol coupling observed. In contrast, both cyclic and acyclic acetal derivatives (C1–E1) afforded 1 in improved yield (61–76%). We also investigated dithioacetal analogs (B1 and F1) and identified thiophenol-derived F1 as the optimal substrate, giving rise to 1 in quantitative yield. Notably, we found that this activation strategy was also applicable to electronically unactivated 3-phenylpropionaldehyde. Under the same conditions, all derivatives (A2–E2) furnished gem-disilane 2 in poor yield (5–19%) with the exception of thiophenol derivative F2, which furnished 2 in 77% yield.

2. Activation Strategy.

^a Yields determined by ¹H NMR of crude reaction mixtures. TPPA, tris­(pyrrol­idino)­phos­phor­amide.

Dithioacetals are bench-stable compounds that are commonly used as protecting groups for aldehydes and ketones. They are conveniently prepared via a single condensation step, which we later demonstrate can be performed in situ before applying electrolysis. Dithioacetals are often employed in umpolung-type alkylation reactions due to their ability to stabilize α-anionic intermediates. This stabilization is attributed to strong polarization and hyperconjugation. In addition, C–S bonds in thioethers are generally considered to be reductively labile owing to the low-lying σ* _C–S, which have been shown to cleave to form stabilized carbanions using stoichiometric metal reductants or electrochemistry from electronically activated substrates (e.g., benzylic thioethers). These features make dithioacetals ideal activating groups for the desired deoxygenative difunctionalization of carbonyl compounds, allowing for successive reductive C–S cleavage and substitution. This strategy overcomes the need for native anion-stabilizing groups such as aryl, vinyl, or cyano, substantially expanding the scope of suitable substrates.

To enhance the operational simplicity, we develop a streamlined one-pot sequence to generate gem-disilane products directly from commercially available carbonyls (Scheme ). Using THF as solvent and Me₃SiOTf as the catalyst, benzaldehyde afforded dithioacetal rapidly using 2.2 equiv PhSSiMe₃ donor, whereas longer reaction times were required for aliphatic aldehydes as well as ketones. Upon addition of trimethylsilyl chloride (TMSCl), direct electrolysis of the resultant mixture afforded the desired benzylic gem-disilane 1 in 93% yield.

3. Disilylation Reaction Using In Situ Activation.

^a Reaction using isolated dithioacetal. Isolated yields are reported in this scheme, with ¹H NMR yields denoted with parentheses.

We then carried this one-pot disilylation using a panel of carbonyl-containing compounds. Cyclic and acyclic aliphatic ketones afforded gem-disilanes 3 and 4 in 70% and 89% yield, respectively. Benzaldehyde derivatives containing N-methyl-indole (5), carbamate (6), ferrocene (7), and trimethoxy aryl ether (8) functionalities were suitable substrates as well. Notably, disilane 8, which was isolated in 60% yield, was used by O’Shea in a diastereoselective olefination to complete the synthesis of marine natural product combretastatin A4 and investigational compound DMU-212 by leveraging the reactivity of in situ generated α-silyl carbanions. Using modified conditions disclosed by Bandar, we also achieved three-component coupling of disilane 8 to afford polyfunctionalized pyridine 9 in 21% yield.

We propose the following stepwise activation pathway for the disilylation (Scheme A, Y = SPh). First, dithioacetal I-A undergoes reductive cleavage of the C–S bond to unveil α-S-stabilized radical I-A-I. This radical is further reduced to α-S-stabilized anion I-A-II, which reacts with a silyl electrophile to generate II-B. Subsequent reductive cleavage of the remaining C–S bond and radical-polar crossover affords a α-Si-stabilized carbanion II-B-II, which can be trapped by a second equivalent of silyl electrophile to afford the gem-difunctionalized product II-C.

4. Mechanistic Proposal and Reaction Progress Analysis.

^a Assay yield determined by GC-FID. Reduction potentials determined by square wave voltammetry.

To further investigate the stepwise desulfurative cross coupling of dithioacetals with TMSCl, cyclic and square wave voltammetry were conducted; dithioacetal F1 was shown to be reduced at −2.1 V (current onset potential vs Fc/Fc⁺), while putative intermediate α-silyl thioether 10 was reduced at a more cathodic potential (−2.5 V). Given this marked potential difference between F1 and 10, we reason that the selectivity of the difunctionalization process is likely under faradaic control, wherein conversion of F1 to 10 would precede that of 10 to 1. Indeed, reaction monitoring revealed that passing 1.8 F/mol of charge resulted in the selective formation of 10 in 91% yield with little gem-disilane 1; the latter was observed as the main product in 92% yield only after passing additional charge for a total of ≥5 F/mol (Scheme B). Armed with this information, we reasoned that selective, two-step synthesis of mixed gem-difunctionalization products could be achieved using two different electrophiles, contingent upon the development of selective monodesulfurative functionalization reactions.

To this end, we first developed a desulfurative silylation to prepare secondary and tertiary α-silyl thioethers (Scheme A). Employing aliphatic dithioketal 11 as a model substrate, we obtained optimal conditions to afford 12 in 92% yield with 10% over-reduced byproduct 3, using a sacrificial Mg anode and Pt cathode, tetrabutylammonium perchlorate as electrolyte, and THF:TPPA (7:1) as the solvent mixture (entry 1). Inclusion of TPPA was essential, as when conducted in neat THF, the reaction suffered from high, unstable cell potentials and poor conversion (entry 2). The use of a platinum cathode delivered higher conversion than graphite (entry 3). Notably, the use of LiClO₄ electrolyte resulted in no reaction, possibly via promotion of silyl chloride reduction via Lewis acid activation (entry 4). No reaction occurred in the absence of electricity with and without activated magnesium metal, supporting that this transformation is largely electrochemically driven and not a result of an in situ generated reducing metal species (entries 5 and 6).

5. Mono-Silylation Reaction .

^a Yields in panel A were determined by ¹H NMR of crude reaction mixtures. Isolated yields are reported in panel B with crude ¹H NMR yields denoted with parentheses. n.d. denotes product was not detected.

We then evaluated the scope and functional group tolerance of the mono-silylation reaction (Scheme B). Cyclic ketones of various ring sizes (13, 17, 22, 24) underwent reductive silylation smoothly, featuring pyran (24), ester with enolizable α-positions (13), and Weinreb amide (17). Furthermore, aldehyde-derived substrates bearing N-Me-indole (21), benzothiophene (23), boronate (14), chloroarene (16), and carbamate (18–20) functionalities were tolerated. Exploration of other electrophiles including dimethylchlorosilane (24), phenyl­dimethyl­chloro­silane (19), and trimethylgermanium chloride (20) also demonstrated generalizability. α-Silyl thioether 10, which has been shown to undergo TBAF-catalyzed Michael addition (26), can also undergo a novel electrochemical Peterson-type olefination to yield 25 in currently moderate yield.

Recognizing the synthetic utility of organoboronates, we next developed a complementary set of electrochemical conditions to extend this electroreductive strategy to desulfurative mono-borylation (Scheme A). Through de novo reaction discovery and optimization (SI Section 3), we obtained the desired α-boryl thioether 27 in 77% yield using a stainless-steel cathode, magnesium sacrificial anode, and lithium bis­(tri­fluoro­methane­sulfonyl)­amide (LiTFSI) in dimethoxyethane (DME) electrolyte solution (entry 1). HBpin was found to be the optimal electrophile; MeOBpin performed poorly while B₂pin₂ was unreactive under these conditions (entries 2 and 3). This reaction was found to be sensitive to current density, where lower current of 2.5 mA gave poor conversion (entry 4). Running the reaction without passing electricity or with activated Mg powder alone gave no conversion, and additional no-electricity controls with Li metal as a chemical reductant led to full consumption of F2 and generation of diborylated byproduct 28 in 37% yield (entries 5–7). This distinct chemoselectivity under chemical versus electrochemical reduction conditions (entries 5 vs 1) highlights the unique capability of faradaic control in the latter.

6. Mono-Borylation Reaction.

^a 5 F/mol charge applied. Yields in panel A were determined by ¹H NMR of crude reaction mixtures. Isolated yields are reported in panel B with crude ¹H NMR yields denoted with parentheses.

A representative scope of α-boryl thioethers was subsequently obtained (Scheme B). Aldehyde-derived substrates containing α-cyclopropyl (37), carbamate (37), alkynyl silane (31), siloxy (32), pyran (39), and chloroarene (29) functionalities underwent the transformation with synthetically useful yield. Cyclic and acyclic ketone-derived substrates containing 1,1-difluoro (34), aryl ether (35), spiroketal (33), and thioether (30) moieties were tolerated. α-Boryl thioethers 37 and 39 underwent conversion to the corresponding BF₃K salt in 84% and 95% yield, respectively. By passing excess charge, exhaustive borylation of N-aryl morpholine-containing dithioacetal afforded diboronate 36 in 68% yield.

Upon establishing faradaically controlled, selective monofunctionalizations and exhaustive bisfunctionalizations, we then explored the potential of sequential cross coupling to access gem-heterodifunctionalized compounds from dithioacetals (Scheme A). Silylation of α-silyl thioether 15 with dimethylchlorosilane generated aliphatic gem-heterobissilane 41 in 61% yield. Analogous reaction with trimethylgermanium chloride afforded α-silylgermane 42 in 64% yield. Silylation of α-boryl thioether 27 with TMSCl afforded α-silyl boronate 43 in 79% yield. Treatment of 27 with trimethylgermanium chloride afforded α-germylboronate 44 in 94% yield.

7. Heterodifunctionalization Sequences.

^a 10 F/mol charge applied. Isolated yields are reported in panels A and B, with crude ¹H NMR yields denoted with parentheses.

α-Silyl boronates and related gem-bismetalloids are useful reagents with limited synthetic access, and fully α-substituted variants are especially challenging to obtain. To address this challenge, we applied our strategy to various ketone-derived substrates (Scheme B). For example, α-boryl thioether 33 underwent silylation to furnish tetrasubstituted silyl boronate 45. Here, another carbonyl protecting group, a cyclic acetal, stayed intact. α-Germyl hydrosilane 46, N-Boc piperidine-containing α-silyl boronate 47, gem-heterobissilane 48, α-germyl boronate 49, and trimetalloids 50 and 51 can each be accessed via tandem desulfurative difunctionalization in synthetically useful yield. Notably, all of these products arose from a common retrosynthetic disconnection, showing our electroreductive platform as a general entry to combinatorial, diversity-oriented synthesis.

During reaction development, we found that TPPA as a cosolvent was crucial for achieving high yields in reactions conducted in THF (Scheme A, entries 1 and 2) but was unnecessary when DME was used as the solvent. Chronopotentiometry studies showed that the inclusion of TPPA in THF provided lower and more stable cathodic and anodic potentials (Figure A), in line with previous findings that TPPA prevents electrode passivation. We studied interactions of TPPA with anodically generated Mg²⁺ by ³¹P NMR, revealing a slight chemical shift change postelectrolysis, consistent with coordination to a Lewis acid (SI Section 5). Furthermore, colorless crystals deposited on both electrodes postelectrolysis, which were identified by X-ray crystallography as MgCl₂(TPPA)₂ (Figure B). Coordination by TPPA would render Mg²⁺ more difficult to reduce. Indeed, investigation of the electroreduction of Mg­(OTf)₂ by linear sweep voltammetry validated this hypothesis (Figure C). Thus, TPPA suppresses the electroreduction of anodically generated metal salts at the cathode surface, deepening the effective reductive potential accessible in THF. In contrast, DME alone can chelate with Mg²⁺, leading to more stable electrode voltages and efficient electrolysis.

1.

Role of TPPA in deep electrolysis in THF.

In conclusion, we disclose an electrochemical strategy for the formal deoxygenative gem-difunctionalization of ketones and aldehydes via dithioacetal activation. We anticipate that this unified strategy to access aliphatic metalloids from carbonyl compounds will enable applications across synthetic, materials, and biological contexts. More broadly, this work establishes redox activation of aliphatic carbonyls as a general platform for modular, faradaically controlled, and iterative difunctionalization and opens new opportunities for reductive electrochemistry on electronically unbiased substrates.

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/jacs.6c09867.

• Experimental procedures, optimization of reaction conditions, characterization data, and copies of NMR spectra (PDF)

The authors declare no competing financial interest.