Al
Hannam‡
a,
Phinyada
Kankraisri‡
a,
Karan R.
Thombare‡
b,
Prahallad
Meher
b,
Alexandre
Jean
c,
Stephen T.
Hilton
d,
Sandip
Murarka
*b and
Stellios
Arseniyadis
*a
aDepartment of Chemistry, Queen Mary University of London, Mile End Road, E1 4NS, London, UK. E-mail: s.arseniyadis@qmul.ac.uk
bDepartment of Chemistry, Indian Institute of Technology Jodhpur, Karwar-342037, Rajasthan, India. E-mail: sandipmurarka@iitj.ac.in
cIndustrial Research Centre, Oril Industrie, 13 rue Desgenétais, 76210, Bolbec, France
dUCL School of Pharmacy, University College London, 29-39 Brunswick Square, WC1N 1AX, London, UK
First published on 3rd July 2024
We report here a practical and cost-effective method for the synthesis of CHF2-containing benzimidazo- and indolo[2,1,a]-isoquinolin-6(5H)-ones through a visible light-mediated difluoromethylation/cyclization cascade. The method, which affords functionalized multifused N-heterocyclic scaffolds in moderate to high yields under mild reaction conditions, is also easily scalable using low-cost 3D printed photoflow reactors.
![]() | ||
Fig. 1 Representative syntheses of fluoroalkylated benzimidazo[2,1-a]isoquinolin-6(5H)-one derivatives. |
We initiated our study by conducting a first set of reactions in batch using 1a as a model substrate and difluoromethyl triphenylphosphonium bromide as the CHF2 radical precursor. The latter, first introduced by Burton,14 was found to be readily available and easy to handle compared to all the other difluoromethylation reagents. We evaluated four different photocatalysts as well as several reaction conditions as shown in Table 1. As a general trend, the best result was obtained when running the reaction in a 1:
1 mixture of MeCN and DCM at rt under light irradiation (440 nm Kessil lamp) using 2 equiv. of PPh3CHF2Br, 1 equiv. of 2,6-lutidine and 2 mol% of fac-Ir(ppy)3 (68%, Table 1, entry 4). In sharp contrast, the use of Ru(bpy)3, 4CzlPN and Eosin Y under otherwise identical conditions did not provide the desired product (Table 1, entries 5–7). This disparity can be attributed to the large difference in the excited-state reduction potentials between [Ru(bpy)3Cl2] (−0.81 V vs. SCE), 4-CzIPN (−1.04 V vs. SCE), Eosin Y (−1.15 V vs. SCE) and fac[Ir(ppy)3] (−1.73 V vs. SCE),15 the latter being the only one capable of generating the CHF2 radical from PPh3CHF2Br (−1.2 V vs. SCE) via single-electron-transfer (SET). The use of other solvents, such as MeCN, THF or DMSO (Table 1, entries 1–3), or other bases, such as DIPEA, NEt3, K2CO3 or DABCO (Table 1, entries 8–11), led to the desired product, albeit in lower yields. Replacing PPh3CHF2Br by Akita's (difluoromethyl)bis(2,5-dimethylphenyl)sulfonium salt16 failed to provide the product (result not shown). Unsurprisingly, running the reaction in the absence of photocatalyst or light proved unproductive. Finally, as decreasing the amount of CHF2 radical precursor from 2 equiv. to 1.2 equiv. did not drastically impact the yield (64%, Table 1, entry 4), we maintained this stoichiometry for the rest of the study. Also, while the use of the MeCN and DCM seems counterintuitive, it proved to work well when running our oxy-difluoromethylation reactions under continuous flow using 3D printed PhotoFlow reactors,13a which we also wish to implement here.
Entry | Photocatalyst | Solvent | Base | Yielda (%) |
---|---|---|---|---|
a Determined by 19F NMR using trifluorotoluene as an internal standard. b Reaction run with 1.2 equiv. of PPh3CHF2⊕Br⊖. c Reaction run either in the dark or without photocatalyst. | ||||
1 | fac-[Ir(ppy)3] | MeCN | 2,6-Lutidine | 54 |
2 | fac-[Ir(ppy)3] | THF | 2,6-Lutidine | 34 |
3 | fac-[Ir(ppy)3] | DMSO | 2,6-Lutidine | 63 |
4 | fac-[Ir(ppy)3] | MeCN/DCM (1![]() ![]() |
2,6-Lutidine | 68 (64b) (—c) |
5 | Ru(bpy)3 | MeCN/DCM (1![]() ![]() |
2,6-Lutidine | — |
6 | 4CzlPN | MeCN/DCM (1![]() ![]() |
2,6-Lutidine | — |
7 | Eosin Y | MeCN/DCM (1![]() ![]() |
2,6-Lutidine | — |
8 | fac-[Ir(ppy)3] | MeCN/DCM (1![]() ![]() |
DIPEA | 35 |
9 | fac-[Ir(ppy)3] | MeCN/DCM (1![]() ![]() |
NEt3 | 39 |
10 | fac-[Ir(ppy)3] | MeCN/DCM (1![]() ![]() |
K2CO3 | 51 |
11 | fac-[Ir(ppy)3] | MeCN/DCM (1![]() ![]() |
DABCO | 67 |
![]() |
After identifying the best reaction conditions, we proceeded to examine the substrate scope by evaluating a broad range of N-substituted 2-aryl benzimidazoles (Fig. 2). As a general trend, the reaction appeared to be tolerant to substrates bearing both electron-withdrawing and electron-donating groups on the phenyl ring. Hence, the para-methyl (1b) and para-isopropyl (1c) derivatives were converted to the corresponding CHF2-containing benzimidazo[2,1-a] isoquinolin-6(5H)-ones 2b and 2c in 61% and 69% yield, respectively. Unsurprisingly, the meta-methoxy derivative 1d provided two regioisomers (2d and 2d′) in a 1:
1 ratio and 44% combined yield. Likewise, the tri-methoxy derivative 1e afforded the difluoromethylation/cyclization cascade product 2e in 47% isolated yield. Interestingly, the 1-naphthyl and 2-naphthyl derivatives 1f and 1g were converted in 66% and 46% yield respectively. As mentioned previously, substrates bearing an electron-withdrawing group on the phenyl ring were also readily converted. Hence, the para-fluoro (2h, 49%), para-chloro (2i, 43%), para-bromo (2j, 53%), para-trifluoromethyl (2k, 41%), ortho-fluoro (2l, 43%), and ortho-bromo (2m, 23%) derivatives were all obtained, albeit in slightly lower yields. The method was also successfully applied to 2-phenyl benzimidazoles bearing various substituents on the benzimidazole ring. Again, both electron-donating and electron-withdrawing substituents were well tolerated as showcased by the moderate to good yields ranging from 36% to 72% obtained with all the substrates tested. Comparatively, substrates bearing an electron-donating group, such as the 4-methyl (2n, 61%) and the 5,6-dimethyl (2o, 72%) derivatives, were obtained in higher yields that the ones bearing an electron-withdrawing group such as the 5-bromo (2p, 37%), the 5,6-dichloro (2q, 38%) and the 5-phenylmethanone (2r, 36%) derivatives. We also explored the scope of the reaction with regards to the substituent on the N-acryloyl moiety. Interestingly, replacing the methyl group by a benzyl group did not hamper the reaction as the corresponding product 2s was obtained in 52% yield.
Following these results, we decided to extend the method to another family of multifused N-heterocyclic scaffolds, namely the indolo[2,1-a]isoquinolin-6(5H)-ones, which are widely found in pharmaceuticals, natural products and functional materials.17 Pleasingly, the method could also be used to access these interesting compounds as showcased by the conversion of the 2-phenyl derivative 3a to the corresponding CHF2-containing indolo[2,1-a]isoquinolin-6(5H)-one 4a in 78% yield. Once again, the method tolerated both electron-donating (3b and 3c) and electron-withdrawing (3d and 3e) substituents around the phenyl and the indole rings.
To confirm the mechanism, we conducted a fluorescence quenching and a radical trapping experiment using TEMPO (Fig. 3A). The latter resulted in the formation of 36% of the CHF2-containing benzimidazo[2,1-a]isoquinolin-6(5H)-one 2a along with 34% of the TEMPO-CHF2 adduct, which strongly supports a one-electron reduction of PPh3CHF2Br and subsequent decomposition releasing the CHF2 radical. The “ON–OFF” experiment (Fig. 3B) and quantum yield calculation (Φ = 0.615) clearly show that the reaction doesn’t proceed through a radical chain mechanism (see ESI† for more details). With these results in hand, we propose the following mechanism where the excited *Ir(ppy)3 undergoes SET to the triphenyl phosphonium bromide, which leads to the release of a CHF2 radical (Fig. 3C). This radical subsequently adds onto the acryloyl moiety to form a radical intermediate, which then undergoes intramolecular cyclization, oxidation and deprotonation to form the desired product and thus complete the photocatalytic cycle.
![]() | ||
Fig. 3 (A) Radical-trapping experiment. (B) ON–OFF experiment. (C) Plausible reaction mechanism. (D) Scale-up using a 3D-printed PhotoFlow reactor. |
Finally, to demonstrate the scalability of the method, a 1 and a 5 mmol scale difluoromethylation/cyclization cascade were carried out on 1a under continuous flow conditions using our standardized 3D printed PhotoFlow reactors, which offer guaranties in terms of reproducibility (Fig. 3D).18,19 The reaction proved easy to set up and the product was isolated in 57% and 58% yield, respectively. Similarly, the scale-up of the indole derivative 3a afforded the corresponding indolo[2,1-a]isoquinolin-6(5H)-one 4a in 75% and 77% isolated yield.
In summary, we have developed a scalable, operationally trivial and highly straightforward access to CHF2-containing benzimidazo[2,1,a]-isoquinolin-6(5H)-ones through a visible light-mediated difluoromethylation/cyclization cascade. The method can also be used to synthesize CHF2-containing indolo[2,1-a]isoquinolin-6(5H)-ones starting from the corresponding 2-aryl indole precursors. Interestingly, the use of standardized, low-cost, 3D printed PhotoFlow reactors facilitates reproducibility and scalability through a better control of the reaction parameters.
We would like to thank Dr Rodolphe Tamion and Dr Jean Fournier at Oril Industrie for fruitful discussions. We also would like to thank Dr Lucile Vaysse-Ludot from Oril Industrie affiliated to “Les Laboratoires Servier”, SERB [CRG/2022/000470], CSIR [02(0426)/21/EMR-II] and Queen Mary University of London for financial support, and DST-FIST [SR/FST/CS-II/2019/119(C)] for the HRMS facility at IIT Jodhpur.
Footnotes |
† Electronic supplementary information (ESI) available: Detailed synthetic procedures and spectroscopic data. See DOI: https://doi.org/10.1039/d4cc02557a |
‡ These authors contributed equally to this work. |
This journal is © The Royal Society of Chemistry 2024 |