We began our investigation by using acid 1 as the substrate and examined different conditions, as shown in table 1. First we tried Ir-1 as the photocatalyst and there was little product observed (Table 1, Entry 1). When Cu(OAc)2 was added, we were able to isolate the desired product with 26% yield (Table 1, Entry 2). The yield was further improved to 48% percent when ligand L1 was used (Table 1, Entry 3). A further screening of other ligands revealed L4 as a better choice (Table 1, Entries 3-8). We then evaluated different copper sources (Table 1, Entries 9-11). When the reaction was carried out at 30oC, the yield jumped to 81% after 40 hours (Table 1, Entry 12). Given the fact the whole transformation is composed of a few steps, the average yield for each step is impressingly high. At last, we tested different photocatalysts and Ir-1 proved to be the best (Table 1, Entries 12-15). Further control experiments revealed no reactions occurred in the absence of photocatalyst, Cs2CO3, or blue LEDs (for a detailed account of the optimization studies, see Supplementary Table S1-S5).
With the optimized conditions in hand, we proceeded to investigate the scope of this transformation. We first evaluated this method with different aromatic substituted piperidine-4-carboxylic acids (Table 2). N-substituted piperidine derivatives bearing tert-butoxycarbonyl and benzoyl groups were well tolerated (Table 2, substrates 1-2). A diverse range of electron-withdrawing and electron-donating functional groups were entirely compatible and delivered the products smoothly (Table 2, substrates 3-8). F, Cl, t-Bu, and Ph groups furnished the products in good yields (Table 2, substrates 3, 4, 7, 8). However, strong electron-withdrawing group such as CF3 and strong electron-donating group such as OMe only gave the product in moderate yields (Table 2, substrates 5 and 6). Other aromatic group such as thiophene also worked well (Table 2, substrate 9).
Reaction conditions: substrate (0.5 mmol), Ir-1 (0.015 mmol), Cu(OAc)2 (0.1 mmol), L4 (0.125 mmol), Cs2CO3 (0.75 mmol), DCM (10 mL), 45W blue LEDs, 30oC, 40 h.
We then examined aliphatic substituted piperidine-4-carboxylic acids (Table 3). For these compounds, Ir-2 turned out to be the better catalyst. Methyl and ethyl groups provided the products in moderate yields (Table 3, substrates 10 and 11). However, for allyl, benzyl and isopropyl substituted substrates, oxidation of these functionalities happened and the main isolated products were 4-piperidinone (Table 3, substrates 12-14). Four-membered azetidinone substrate gave α-amino ketone product in 58% yield (Table 3, substrate 15). For five-membered pyrrolidine-3-carboxylic acid substrate 16, β-amino ketone product was isolated in 64% yield. As expected, piperidine-3-carboxylic acids furnished γ-amino ketone products in good yields (Table 3, substrate 17, 18, 20). However, due to the oxidation of the benzyl group, the yield for substrate 19 is low.
We further applied this method to all carbon cyclic acids (Table 4). For cyclobutanecarboxylic acid substrates, 1,4-dicarbonyl compounds were isolated in good yields (Table 4, substrates 21-23). When cyclopentanecarboxylic acid was used, 1,5-dicarbonyl compound was obtained instead in 70% yield (Table 4, substrates 24).
We were pleased to find out that this method worked with acyclic acids (Table 5). For 2,2,2-triphenylacetic acid and 2,2-diphenylpentanoic acid, benzophenone was isolated as the main product in moderate yields (Table 5, substrates 25-26). However, for 2,2-diphenylpropanoic acid, the main product was acetophenone (Table 5, substrates 27). Different substituted 2-methyl-2-phenylpropanoic acids also furnished acetophenone type products (Table 5, substrates 28-31). Accordingly, substrate 32 provided propiophenone as the main product in good yield. It's noteworthy that 3-phenylpropanoic acids could also be compatible and benzaldehyde type products were isolated (Table 5, substrates 33-35). For 3-phenylbutanoic acid, acetophenone was isolated in 66% yield (Table 5, substrate 36).
We also evaluated this method with β-hydroxy acids (Table 6). Interestingly, the C-C cleavage tended to happen at the α-β position, probably because the radical intermediates were stabilized by the β-hydroxy group. Thus, diketones were usually provided as the main products. For substrates which contain tertiary β-hydroxy groups (Table 6, substrates 37-39), the products were isolated in good yields. However, substrates with secondary β-hydroxy groups delivered the products only in moderate yields (Table 6, substrates 40-41). For substrate 42, the C-C cleavage happened at both positions with about 3:1 ratio.
Based on our proposed mechanism ( Figure 2), radical intermediate IV was formed during the process. To capture this intermediate, we tried to add additives. Eventually, we found the addition of Selectfluor successfully delivered ketone-alcohol as the main product. Thus, under the optimized condition, cyclobutanecarboxylic acid substrates provided 4-hydroxybutyrophenones in good yields (Table 7, substrates 43- 45). When cyclopentanecarboxylic acid was used, 5-hydroxybutyrophenone was provided instead (Table 7, substrates 46). For cyclohexanecarboxylic acid, 6-hydroxyhexaphenone was isolated in 75% yield (Table 7, substrates 47). We also tested other six-member cyclic acids, all of them worked smoothly and delivered the products in moderate to good yields (Table 7, substrates 48-52).
Synthetic utilities. This reaction provides a direct method to construct different diketones, which are versatile building blocks in the synthesis of natural products and bioactive compounds. Thus, we performed the synthesis of Primaperone, Melperone and Haloperidol. These drugs could be accessed in one step from product 22b via reductive amination in good yields. We also did late-stage modification of commercial drug and complex natural products. For Sertraline derivative 53a, the reaction worked smoothly and the product was isolated in 83% yield. For steroids 54a and 55a, the regioselectivities were good and we only isolated one product. However, the reactions were sluggish and much of the starting materials were recovered.