Olefination via Cu-mediated Dehydroacylation of Unstrained Ketones

Abstract

The dehydroacylation of ketones to olefins is realized under mild conditions, which exhibits a unique reaction pathway involving aromatization-driven C–C cleavage to remove the acyl moiety, followed by Cu-mediated oxidative elimination to form an alkene between the α and β carbons. The newly-adopted N’-methylpicolinohydrazonamide (MPHA) reagent is key to enable efficient cleavage of ketone C–C bonds at room temperature. Diverse alkyl- and aryl-substituted olefins, dienes, and special alkenes are generated with broad functional group tolerance. Strategic applications of this method are also demonstrated.

Graphical Abstract

Olefin represents one of the most versatile and widely useful functional groups (FGs) in organic synthesis, as it can be easily converted to a large variety of other structural motifs,¹ e.g., alcohols, amines, alkyl halides, ethers, and carbonyl compounds. Conventionally, functionalized olefins are prepared via constructive approaches, namely adding complexity to simpler starting materials, such as olefination of carbonyls,^1,2 hydrofunctionalization of alkynes,³ or derivatization of less substituted olefins.⁴ Alternatively, a complementary strategy is to synthesize substituted olefins from more complex but readily available substrates by removing a small fragment (Fig. 1A). Such so-called deconstructive transformations,⁵ compared to the constructive approaches, have been much less developed. Recently, elegant examples have been demonstrated to form olefins using aldehydes,⁶ nitriles,⁷ carboxylic acids⁸ and their activated derivatives⁹ via decarbonylation, decarboxylation or C–CN bond activation. However, deconstructive olefin synthesis from regular unstrained ketones remains an elusive transformation,¹⁰ likely due to the fact that unstrained ketone C–C bonds are more resistant to manipulations.¹¹ In addition, from the strategic planning viewpoint, this dehydroacylation reaction parallels known ketone-to-olefin reactions¹ and could feature unique utilities because of its deconstructive nature (Fig. 1B). First, ketones can serve as a convenient traceless handle to install α, β,¹² or remote substituents¹³ that would eventually be brought into the olefin products after the dehydroacylation process. Second, it is often easier to alkylate substrates with α,β-unsaturated ketones via radical or anionic conjugate addition;¹⁴ thus, the proposed dehydroacylation reaction can provide a complementary approach to known alkenylations. Third, given that terminal alkenes can be converted to ketones straightforwardly via Wacker oxidation,¹⁵ the coupling with this dehydroacylation process can lead to an unusual iterative two-carbon-deletion of terminal olefins. Fourth, this transformation could be useful for late-stage modification of natural or complex bioactive compounds containing linear ketones. Here we describe the initial development of facile dehydroacylation of unstrained ketones to access functionalized olefins via aromatization-driven C–C cleavage (Fig. 1C). The reaction is enabled by a newly-adopted, easily accessible N’-methylpicolinohydrazonamide (MPHA) reagent that allows for copper-mediated efficient cleavage of ketone α-C–C bonds¹⁶ under mild conditions.

Figure 1.

Synthetic potential and reaction design of the dehydroacylation of ketones.

Aromatization has emerged as an important driving force for cleavage of unstrained C–C bonds.^{17, 18} To date, a few systems were reported for ketone C–C cleavage using this approach but are associated with limitations (Scheme 1). For instance, in our previous works^17e,g the generation of pre-aromatic intermediate (PAI) 2 from ketone 1 needed a complex sequence: hydrazone formation, [3+2] cycloaddition with excess 1,3-dienes, and Ir-catalyzed olefin migration; additionally, a high reaction temperature (160 °C) was used, which compromised the substrate scope. Alternatively, 2-aminobenzamide^17b,f has been employed to condense with ketone 1 for subsequent C–C cleavage and then alkylation. Likely owing to the low nucleophilicity of the amide moiety, the condensation step often gave moderate efficiency and required purification prior to the C–C cleavage. In addition, cleavage of unactivated primary alkyl moieties has been difficult,^17b possibly because formation of the less aromatic quinazolinone products from PAI 3 requires a higher energy barrier for the C–C cleavage step. Moreover, dehydroacylation, i.e., olefin formation, has not been demonstrated using this strategy, to the best of our knowledge.

Scheme 1.

Aromatization-Driven Ketone C–C Cleavage

Hence, the need for a new-generation activating reagent for efficient aromatization-driven C–C cleavage of ketones is warranted. The reagent has to be inexpensive, react rapidly and cleanly with ketones to form PAI without need of purification, and enable C–C cleavage of diverse substrates under mild conditions. To meet these criteria, hydrazonamides (A), a class of unique compounds, could be suitable candidates. They can be easily prepared from nitriles¹⁹ (vide infra), and their increased nucleophilicity (comparing to an amide) should promote efficient condensation with ketones to give PAI 4 cleanly. In addition, single-electron oxidation of species like 4 is known,²⁰ which may trigger facile C–C cleavage. Moreover, a strong driving force to form triazole is expected to enable generation of unstabilized carbon radicals. Such a driving force has been utilized in an elegant recent demonstration of deconstructive oxygenation of cycloalkanamines by Han and coworkers.¹⁸ To realize the desired dehydroacylation of ketones, we further hypothesized that an inexpensive copper (II) salt can be employed to serve two roles in: first triggering single-electron oxidation of 4 to promote C–C cleavage, and second converting the resulting alkyl radical to the corresponding olefins via oxidative elimination, first studied by Kochi.²¹

To test this hypothesis, 4-(3,4-dimethoxyphenyl)butan-2-one (1a) was used as the model substrate, and hydrazonamides A1-A10 were tested as activating reagents under various conditions (Scheme 2). The reaction was generally conducted in one pot by first mixing 1a, a hydrazonamide and the dehydrating reagent for a certain period, followed by addition of a copper salt. After careful optimization, the desired dehydroacylation product (5a) was obtained in 76% yield using MPHA (A4) as the activating reagent, Cu(OAc)₂ as the oxidant, and Al₂O₃ as the dehydration additive²² (Scheme 2A). The initially-tested N’-phenylbenzohydrazonamide (A1) was incapable of condensing with ketone 1a; replacing the N’-phenyl moiety with a smaller, more electron-donating N’-Me group (A2), the yield of 5a increased significantly to 34%. Reducing the electron density on the phenyl moiety generally decreased the efficiency (e.g., A3); however, changing the phenyl to a 2-pyridyl (MPHA, A4) led to a superior outcome, with virtually full conversion from 1a to PAI 4 and a more selective transformation to 5a, in which the undesired C–H formation or substrate hydrolysis pathways were minimized. While the exact role of the 2-pyridyl moiety is still unclear, the improved efficiency may link to its coordination ability, possibly through directing the Cu(II) in closer proximity to the reaction center. For example, reagents A5 and A6 with electronically modified pyridines gave similar yields (around 70%), while the introduction of an ortho t-butyl to the 2-pyridyl (A7) or the use of 4-pyridyl-substituted one (A8) gave much lower efficiency. Interestingly, A9, with the N-Me replaced by N-H, can fully condense with 1a but cannot promote C–C cleavage. N’-Hydroxypicolinimidamide (A10) was also examined but failed to condense with 1a.

Scheme 2. Reaction Optimization.

^aThe yields are based on ¹H NMR analysis of the crude reaction mixture. ^bIsolated yield.

The impact of other reaction parameters was also studied (Scheme 2B). The copper salt was found inevitable (entry 1), and removal of Al₂O₃ or use of only1.0 equivalent of MPHA slightly decreased the yield (entries 2 and 3). The reaction appears to be less sensitive to solvents, as replacing the DMSO/H₂O mixed-solvent with either benzene or MeCN still furnished 5a in good yields (entries 4 and 5). Other copper salts, including Cu(OPiv)₂, CuSO₄·5H₂O, Cu(NO₃)₂·3H₂O and CuCl₂ gave much lower efficiency than inexpensive Cu(OAc)₂ (entries 6–9). In addition, attempts to use a catalytic amount of Cu(OAc)₂ with adding an external oxidant was unfruitful at this stage, often resulting in complicated mixtures (entry 10 and Supporting Information), possibly due to competing radical-trapping pathways. Finally, the 2-pyridyl motif in MPHA not only promotes the key C–C cleavage reaction, but also simplifies the reagent synthesis. As a bench-stable solid, MPHA was easily prepared in one step from inexpensive picolinonitrile and methylhydrazine in good yield (Scheme 2C).

The substrate scope was then investigated. Both aryl-(Scheme 3A) and alkyl-substituted olefins (Scheme 3B) were generated in good yields with broad structural diversity covering terminal, 1,1-disubstituted and 1,2-disubstituted olefins. The C–C cleavage generally exhibits over 20:1 site-selectivity at the non-methyl side, and can be run at a gram scale with comparable yield (5a). Besides methyl ketones, other linear ketones, such as phenyl alkyl ketone, also reacted.²³ One merit of this method, comparing to the metal-hydride-based approaches,¹⁰ is that no styrene oligomerization or olefin migration was observed. Endo- (5v-w) and exo-cyclic olefins (5x) can both be constructed, and α-fluoroalkenes were also accessible from the corresponding readily available α-fluoroketones, albeit in lower yields (5l). Various FGs were tolerated, including aryl bromides and iodides (5k and 5d), boronic esters (5g), silanes (5b), phenols (5j), esters (5e), ketals (5ah), and Weinreb amides (5h). Orthogonal handles for other types of deconstructive functionalization, including aliphatic nitriles (5ab) and carboxylic acids (5ac), were preserved as well, which also implies complementary reactivity of ketone moieties. No oxidation side-products were observed with electron-rich arenes (5j and 5aj); heteroarenes and polyarenes were compatible, including indoles (5ag), benzofurans (5r), quinazolinones (5ak), xanthines (5ai), anthracenes (5af), and pyrenes (5ad). Beyond regular olefins, conjugated and electron-poor and -rich olefins can also be produced (Scheme 3C). Note that acrylates (5ao) and acrylamides (5an) were produced in good yields and intact, highlighting not only the generality of the reaction, but also its selectivity against uncontrolled radical additions. Additionally, multiple-substituted internal alkenes can also be synthesized (5m, 5n, 5v, 5w, 5ap), albeit in lower yields, due to the challenge of forming the dihydro 1,2,4-triazole intermediates with these bulkier substrates.²⁴ Using aza-Michael additions with butenone followed by the dehydroacylation, a unique strategy was realized for the synthesis of vinyl indole (5aq) and enamide (5ar).

Scheme 3. Substrate Scope^a,b.

^aUnless otherwise mentioned, all reactions were conducted on a 0.2 mmol scale. ^bIsolated yields. ^c0.4 mmol scale. ^d0.5 mmol scale. ^e0.1 mmol scale. ^fDMSO instead of DMSO:H₂O (4:1). ^gThe E/Z ratio was determined by ¹H NMR analysis. brsm, (yield) based on recovered starting material.

The synthetic potential of this method was further explored. As illustrated in Scheme 4A, merging of the dehydroacylation with diverse ketone-generating reactions can enable rapid construction of complex olefin products (otherwise challenging to introduce) via a simple two-step protocol. For example, the metal hydride H-atom transfer (MHAT) coupling between a 1,1-disubstituted olefin and butanone,²⁵ followed by this method, led to the formal hydroalkenylation of 7, which is a non-trivial transformation for non-aromatic olefin substrates.²⁶ The photocatalytic addition of C–H bonds to butanone²⁷ and then dehydroacylation introduced a vinyl group at the conventionally hard-to-do-so position with tolerance of tertial anilines (12). Further, facile construction of a homoallylic quaternary center (15) was realized through a sequence of oxidative Heck²⁸ and this method. Besides, vinylated isochromenones, such as 18, was modularly forged via combining tandem annulation²⁹ and dehydroacylation. On the other hand, late-stage deconstructive derivatizations of ketone-containing terpenes and drugs were further showcased in Scheme 4B. Famesylacetone (19), geranylacetone (21), pentoxyifylline (23), α-ionone (25), and dihydro-β-ionone (27) were all transformed to the corresponding lower olefins or 1,3-dienes. A more complex rotenone derivative (29) also worked smoothly despite the presence of several other FGs. Finally, by combining this method with Wacker oxidation, controlled iterative “−2C” and “−4C” deletion was further demonstrated with alkene 31 (derived from feedstock chemical oleic acid) as a model compound; formally, two methylene groups were removed via this two-step sequence in one iteration (Scheme 4C).

Scheme 4. Further Applications^a,b.

^a Unless otherwise mentioned, all reactions were conducted on a 0.2 mmol scale. ^bIsolated yields. ^c0.1 mmol scale. ^d0.4 mmol scale. ^e0.5 mmol scale. ^f2.0 mmol scale. ^gThe E/Z ratio was determined by ¹H NMR analysis. For detailed experimental procedures, see the Supporting Information.

A plausible reaction pathway was further proposed (Scheme S1A). First, the condensation between ketone 1 and MPHA generates PAI 4 that can be isolated and fully characterized (step A), which is expected to be subsequently oxidized by Cu(OAc)₂ to form a delocalized radical (R1) via either single-electron oxidation then deprotonation, or a proton-coupled electron transfer pathway²¹ (step B). The following aromatization-driven C–C fragmentation of R1 should give an alkyl radical (R2) and a triazole (6) (step C).¹⁷ⁱ The final oxidative elimination of radical R2 by another equivalent of Cu(OAc)₂ is anticipated to yield the desired olefin (5) with the concurrent formation of Cu(I) and HOAc (step D).²¹

This proposed pathway is consistent with a collection of experimental and literature evidence. For example, PAI 4 gives similar yields of 5 and 6 as of from 1 when subjected to the standard condition, which supports the intermediacy of PAI 4. It is also documented that oxidation of compounds similar to 4 can give intermediate R1 that can be isolated with certain substituents; the dark brown color observed during this dehydroacylation reaction matches the reported color of R1.²⁰ Therefore, we are prone to propose a similar oxidation process of PAI 4 by Cu(II), followed by aromatization-driven C–C cleavage. Moreover, the generation of alkyl radicals is implied by the formation of a cyclization product (38) with substrate 36 (Scheme S1B, Eq. S1). Finally, an intramolecular KIE value of 2.3 was observed at the β position with β-D-1c (Eq. S2), which matches with the literature value for the Cu-mediated oxidative elimination.^21b A much smaller, close-to-one KIE was further observed through the intermolecular competition experiment (Eq. S3), indicating that the rate-determining step occurs prior to the oxidative elimination, e.g., the C–C cleavage (for more details, see supporting information). Currently, investigation on the detailed mechanism and the development of other deconstructive methods with unstrained ketones using MPHAs are ongoing.