Electron Donor–Acceptor Complex-Enabled Autoinductive Conversion of Acylnitromethanes to Acylnitrile Oxides in a Photochemical Machetti–De Sarlo Reaction: Synthesis of 5-Substituted 3-Acylisoxazoles

Abstract

A photochemical Machetti–De Sarlo reaction has been developed for preparing 5-substituted 3-acylisoxazoles from acylnitromethanes and terminal alkynes. This photochemical protocol utilizes an innovative electron donor–acceptor complex, generated in situ from acylnitromethanes, catalytic LiO^tBu, and 1,1,1,3,3,3-hexafluoro-2-propanol, as a photosensitizer to promote rapid conversion with a broad substrate scope in up to 80% efficiency. A sigmoidal autoinductive kinetic profile is revealed, demonstrating the novel and unique dual catalysis in the first photochemical approach of this reaction.

Renewed interest in synthetic photochemistry has spurred the use of light as a “traceless” promoter in organic transformations,¹ utilizing transition metal complexes and organic dyes as photosensitizers. Recently, the use of charge transfer (CT) complexes,² particularly electron donor–acceptor (EDA) complexes,³ as photosensitizers in organic synthesis has gained significant traction. These complexes, consisting of an electron donor and acceptor, exhibit a unique bathochromic (red) shift in the ultraviolet–visible (UV–vis) absorption spectrum, a property not observed in the individual components. Several types of EDA complexes, formed in situ and catalytic in nature, have been summarized in a recent review.^3a This principle has expanded to include EDA complexes generated from added electron donors or acceptors that form complexes with reaction substrates, acting as photosensitizers for their own conversions.^3b,3f

In 2005, Machetti and De Sarlo demonstrated that activated primary nitromethanes undergo dehydration under thermal conditions in the presence of 1,4-diazabicyclo[2.2.2]octane (DABCO) to yield isoxazoline derivatives, formal cycloadducts of nitrile oxides.^4a This transformation, now known as the Machetti–De Sarlo (MDS) reaction, has been expanded to employ imidazole- and 4-dimethylaminopyridine (DMAP)-type bases^4b,4c as well as catalytic Cu(OAc)₂^4d and is catalyzed in EtOH or water.^4e,4f A more recent example included a (2,2,6,6-tetramethylpiperidin-1-yl)oxyl (TEMPO)-catalyzed protocol under aerobic conditions.^4g

Despite recent advancements, a photochemical approach to this transformation has yet to be realized. Herein, we introduce a novel EDA photocatalyst, generated in situ from acylnitromethanes, catalytic LiO^tBu, and 1,1,1,3,3,3-hexafluoro-2-propanol (HFIP) for the synthesis of isoxazoles via the MDS reaction (Scheme 1). The protocol represents the first photochemical MDS reaction, thus achieving the previously elusive protocol. Kinetic studies of the reaction revealed an autoinduction kinetic,⁵ thereby confirming the consistent kinetic character of the reaction under both thermal and photochemical conditions. We proposed a mechanism involving the generation of the catalytic nitrite anion, which subsequently catalyzed the transformation with the entire process facilitated by the EDA-catalyzed photochemistry. This mechanism highlighted the uniqueness of this protocol, in that the transformation was simultaneously catalyzed by two distinct classes of catalytic species.

Scheme 1. First Photochemical Machetti–De Sarlo Reaction.

The exceptional ability of HFIP⁶ to stabilize complex organic intermediates led us to employ it for reaction optimization, using benzoylnitromethane (1a) and phenylacetylene (2a). Extensive optimization studies⁷ led to the identification of the optimal conditions, as shown in panels a and b of Scheme 2. Catalytic LiO^tBu, HFIP, and 390-nm light-emitting diode (LED) were all confirmed necessary for the reaction. Importantly, the exclusion of oxygen from the reaction mixture was not necessary as the reaction was found to provide similar results, thus making the current protocol more practical. The protocol was validated in the reactions of compound 1a with alkynes 2a–2y (Scheme 2a). Although the reaction yielded product 3aa in 52%, the reaction of 4-tolylacetylene (2b) resulted in a lower yield (28%), while the reactions of 4-ⁿBu-phenylacetylene (2c) and 4-^tBu-phenylacetylene (2d) produced slightly better yields (39%) for both products 3ac and 3ad.

Scheme 2. Scope of the Reaction.

Yields of isolated products.

Complex mixture.

The reaction of 3-anisylacetylene (2e) produced isoxazole (3ae) in 61% yield, with the 3-OMe group having minimal impact on the reaction. However, 4-anisylacetylene (2f) yielded only trace amounts of product 3af. Interestingly, compound 2f was relatively stable under the reaction conditions without compound 1a. The observed decomposition was hypothesized to occur through an unknown mechanism facilitated by the EDA photocatalyst under photoirradiation, which inhibited the formation of the desired product and resulted in a complex reaction mixture. In contrast, alkyne 2g (R¹ = 4-OCF₃-Ph) yielded product 3ag in 51%, with the 4-OCF₃ group’s weaker electron-donating ability contributing to the increased stability of compound 2g. In a series of arylalkynes with halogen atoms on the phenyl ring, reactions generally yielded products in moderate to good yields (57–80%). Substrates with 2-halophenyl groups, such as alkynes 2h (R¹ = 2-F-Ph), 2i (R¹ = 2-Cl-Ph), and 2l (R¹ = 2-Br-Ph), produced the corresponding products 3ah, 3ai, and 3al in moderate yields (57–59%). The yield improved with 3-halogenated arylalkyne 2j (R¹ = 3-Cl-Ph), giving product 3aj in 65%, similar to product 3ae (R¹ = 3-OMe-Ph). For 4-halogenated arylalkynes 2k (R¹ = 4-Cl-Ph) and 2m (R¹ = 4-Br-Ph), products 3ak and 3am were obtained in good yields (77–80%). A trend was observed where 2-halogenated products had lower yields than 3-halogenated products, which, in turn, had lower yields than 4-halogenated products. Alkynes with inductively electron-withdrawing groups at the 4 position, such as compounds 2n (R¹ = 4-CF₃-Ph) and 2o (R¹ = 4-CN-Ph), also yielded products in good yields (67 and 69%). Notably, alkyne 2p(8a) (R¹ = 2-NHBz-Ph) produced product 3ap in 65%, with the benzoyl group likely reducing electron donation and steric hindrance from the anilide group, thus reducing the co-planarity, contributing to the good yield.

The scope of reactions between compound 1a and alkynes substituted with heterocycles was explored. The reaction with 2-ethynylpyridine (2q) yielded product 3aq in 28%, while the reaction with 3-ethynylpyridine (2r) resulted in a complex mixture with no isolated product 3ar, likely due to N-oxidation of the pyridine ring. Similarly, the reaction with 3-ethynylthiophene (2s) yielded only 14% of product 3as, likely because thiophene’s electron delocalization into alkyne led to side reactions as well as the possible sulfur oxidation. Alkylalkynes 2t (R¹ = ^tBu) and 2u (R¹ = Cy, cyclohexyl) produced products 3at and 3au in 62 and 58% yield, respectively, although requiring longer reaction times. Overall, the reactions were effective for a variety of R¹ substituents, including aromatic, aliphatic, and alicyclic groups, yielding isoxazoles in moderate to good yields. Notably, reactions with 4-haloarylalkynes 2k (R¹ = 4-Cl-Ph) and 2m (R¹ = 4-Br-Ph) were particularly successful. However, alkynes with R¹ = 4-alkyl-Ph (2b–2d) and R¹ = heteroaryl (2q–2s) groups yielded low or no products.

Reactions with terminal alkynes containing various functional groups were evaluated, including ethyl propiolate (2v), 4-pentyn-1-ol (2w), propargyl bromide (2x), and N⁷-propargylated 8-bromotheophylline (2y).^8b These reactions proceeded smoothly, yielding products in moderate to good yields: 3av (R¹ = CO₂Et) at 72%, 3aw (R¹ = (CH₂)₃OH) at 52%, 3ax (R¹ = CH₂Br) at 65%, and 3ay (R¹ = CH₂-N⁷-8-bromotheophylline) at 73%. Most reactions were completed within 7–8 h, except for compound 2w, which required a longer time. The reaction with compound 2y, containing a theophylline scaffold, a purine base, highlights potential applications of the current method in biologically relevant molecules.⁹

Phenylacetylene (2a) was tested with various acylnitromethanes (1b–1q) bearing different R² substituents. Both compounds 1b (R² = 2-OMe-Ph) and 1c (R² = 3-OMe-Ph) reacted with compound 2a, yielding products 3ba and 3ca in 56%. The reaction of compound 1d (R² = 4-OMe-Ph) with compound 2a provided isoxazole 3da in a slightly lower yield (50%), with reaction times varying: compounds 1b and 1d required 10 h, while compound 1c completed in 6 h. The yield of product 3ea (R² = 4-OCF₃-Ph) further decreased to 41%. Substrate 1f (R² = 2-F-Ph) yielded product 3fa in 31%, and substrate 1g (R² = 4-Cl-Ph) produced product 3ga in 29%. The reaction with compound 1h (R² = 2-Br-Ph) yielded product 3ha in 20% yield after 11 h, while compound 1i (R² = 3-Br-Ph) was more efficient, yielding product 3ia in 47%. For compound 1j (R² = 4-Br-Ph), the yield was 12%, and compound 1k (R² = 2-I-Ph) produced product 3ka in 34%. These results revealed a correlation between the halogen positions and product yields, with 2- or 4-position halogens generally leading to lower yields and longer reaction times due to electronic or steric effects, while a halogen at the 3 position, as in compound 1i, showed better efficiency (3ia, 47%). This pattern was also observed with product 3ca (R² = 3-OMe-Ph, 56%), where the 3-OMe group enhanced the efficiency. Electron-donating groups at the 2- or 4-position moderately facilitated the reaction, as seen in products 3ba (R² = 2-OMe-Ph, 56%), 3da (R² = 4-OMe-Ph, 50%), and 3ea (R² = 4-OCF₃-Ph, 41%). The reactions of 2,5-dihalogenated aryl substrates 1l (R² = 2-Br-5-F-Ph) and 1m (R² = 2-Br-5-Cl-Ph) proceeded slowly (14 h) with low yields of products 3la (37%) and 3ma (25%). Substrates with electron-withdrawing groups at the 4-position, such as compounds 1n (R² = 4-CF₃-Ph), 1o (R² = 4-CO₂Me-Ph), and 1p (R² = 4-NO₂-Ph), produced the corresponding products in low to modest yields, with the highly electron-withdrawing NO₂ group in compound 1p yielding only 23%. Notably, the CO₂Me and NO₂ groups required shorter reaction times (5 h) compared to the CF₃ group (12 h). Lastly, acylnitromethane 1q (R² = (CH₂)₂Ph), representing an alkyl R², yielded product 3qa in a modest 34% yield but required a longer reaction time (24 h). Overall, these results demonstrate that the R² substituents of acylnitromethanes 1 have a more significant impact on the reaction outcomes than the R¹ substituents of alkynes 2.

In addition to aryloylnitromethanes and terminal alkynes, ethyl nitroacetate and other dipolarophiles, styrene and diphenylacetylene, were investigated (Scheme 3a). The reactions of benzoylnitromethane (1a) with styrene and diphenylacetylene yielded only trace amounts of products 4 and 5, along with complex mixtures observed in both cases. We suspected that both styrene and diphenylacetylene may have been decomposed by the EDA complex under the photoirradiation. For ethyl nitroacetate, the reaction with compound 2a proceeded less effectively; after 12 h, the reaction was incomplete while yielding the desired product (6) in 38% yield, along with 41% of unreacted ethyl nitroacetate. As the MDS reaction is typically conducted using activated nitromethanes in excess, our protocol was conducted with reversed stoichiometries for comparison. As shown in Scheme 3b, the reactions were conducted between 1.0 equiv of compound 2a and 2.0 equiv of compounds 1a, 1b, and 1p under optimal conditions. providing products in moderate to good yields: 3aa (83%), 3ba (62%), and 3pa (51%), within various reaction times (5–10 h). In this work, acylnitromethanes (1) were employed as the limiting agent due to their requirement for preparation, whereas most terminal alkynes (2) were commercially available. Nevertheless, these experiments demonstrated the flexibility of our photochemical MDS reaction. The synthesis of product 3aa was carried out on a millimole scale (Scheme 3c) to yield the desired product in a moderate yield (55%), comparable to the smaller scale reaction. Notably, only 1 mL of HFIP was required to conduct this synthesis. The results in reactions b and c of Scheme 3 underscore the practicality and scalability of this procedure. Furthermore, the existence of radical species was probed in radical-trapping experiments using TEMPO, as depicted in Scheme 3d. The results of the photochemical reactions of compound 1a under the standard conditions in the presence of 1.5 equiv of TEMPO, with and without alkyne 2a, corroborated the presence of radical intermediates. In the absence of compound 2a, the reaction proceeded to completion within 1 h, yielding TEMPO-trapped compound 7(10a) as a minor product (10%) and TEMPO ester 8(10b) as a major product (31%). In contrast, the reaction with alkyne 2a (2.0 equiv) yielded product 7 as a major product (18%) and the desired product (3aa) in a trace amount (5%). Compound 7 could be readily generated from the reaction of compound 1a, LiO^tBu, and TEMPO in HFIP at room temperature. Only when compound 7 was photoirradiated did the conversion occur to compound 8. Similar oxidative conversions of nitromethanes to esters, amides, and related carbonyl derivatives have been reported under various conditions,¹¹ although systematic studies to attain the fully optimized photochemical conditions have yet to be conducted and realized. Nevertheless, this study demonstrated for the first time that conversion of compound 1a to a compound of type 8 could take place photochemically.

Scheme 3. (a) Reactions of Acylnitromethanes 1 with Other Dipolarophiles, (b) Reactions with Reversed Stoichiometries of Compounds 1 and 2, (c) Reaction in Millimole Scale, and (d) Radical-Trapping Experiments.

Yields of isolated products.

Complex mixture.

Reaction was incomplete. BRSM = based on recovered starting material.

The formation of the catalytic EDA complex, acting as a photosensitizer in this transformation, was investigated through UV–vis absorption measurements. UV–vis absorption spectra were obtained for different compositions of compound 1a, compound 2a, and LiO^tBu in HFIP (Figure 1). A distinct bathochromic shift in the compound 1a + LiO^tBu mixture was observed, indicating EDA complex formation. The consistent λ_max value was observed in the composition of the actual reaction mixture (compound 1a + compound 2a + LiO^tBu), confirming the presence of the same complex in the reaction, along with a visible increase in turbidity of the reaction mixture upon the addition of LiO^tBu. Previous kinetic studies of the thermal MDS reaction^4d,5 revealed an autoinduction kinetics, which was also observed in our own photochemical reaction.⁷ Additionally, substrate 1a was converted more rapidly in our photochemical reaction; >98% of compound 1a was converted in comparison to 43% conversion under thermal conditions after 5 h. These findings demonstrate a significant acceleration affected by our photochemical conditions in the synthesis of isoxazoles 3.

Figure 1.

UV–vis absorption spectra of various compositions in HFIP.⁷

Scheme 4 proposes the mechanism. UV–vis data (bathochromic shift) suggest the formation of an EDA complex between nitronate 1′ and Li·HFIP (generated by LiO^tBu deprotonation). Kinetic studies, along with prior reports of autoinduction via substrate 1 decomposition to a nitrite anion catalyst,⁵ suggest that the EDA complex promotes photochemical decomposition. The nitrite catalyst then reacts with compound 1 to form intermediate A, which dehydrates to intermediate B. Intermediate B converts to nitrile oxide C (regenerating the nitrite catalyst) and undergoes a 1,3-dipolar cycloaddition with alkyne to yield isoxazole 3. Detection of furoxan 9a(12) in the reaction between compounds 1a and 2a confirms nitrile oxide formation. Notably, competitive nitrile oxide dimerization to give furoxan 9a is minimized in our reaction possibly because of the faster reaction with excess alkyne 2a. The formation of furoxan 9 and its subsequent transformations^4b,12c could not be completely ruled out and may have contributed to low yields of desired products in some cases.

Scheme 4. Proposed Mechanism of the Photochemical Machetti–De Sarlo Reaction.

In summary, we successfully developed the first photochemical Machetti–De Sarlo reaction to synthesize 3-acylisoxazoles from acylnitromethanes and terminal alkynes using a 390-nm LED. This reaction involves a novel EDA complex formed from catalytic LiO^tBu, HFIP, and acylnitromethane substrates, confirmed by UV–vis absorption spectra, showing a significant bathochromic shift. Kinetic studies using ¹H nuclear magnetic resonance (NMR) revealed an autoinduction kinetic, matching the thermal variant and suggesting the generation of the catalytic nitrite anion for the conversion of acylnitromethanes to acylnitrile oxides. The reaction’s dual catalytic events are a distinctive feature. This method is applicable to a broad range of substrates and can be performed at high concentrations (up to 1.0 M) of acylnitromethane substrates with a shorter induction period and faster conversion, achieving modest to good yields of isoxazole products under milder conditions than the thermal alternative.

Data Availability Statement

The data underlying this study are available in the published article and its Supporting Information.

Supporting Information Available

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acs.orglett.4c02708.

• Reaction optimization studies, kinetic experiments, UV–vis absorption measurements, experimental procedures, and copies of ¹H, ¹³C, and ¹⁹F NMR spectra (PDF)

The authors declare no competing financial interest.