Decarboxylative sulfoximination of benzoic acids enabled by photoinduced ligand-to-copper charge transfer

Sulfoximines are synthetically important scaffolds and serve important roles in drug discovery. Currently, there is no solution to decarboxylative sulfoximination of benzoic acids; although thoroughly investigated, limited substrate scope and harsh reaction conditions still hold back traditional thermal aromatic decarboxylative functionalization. Herein, we realize the first decarboxylative sulfoximination of benzoic acids via photo-induced ligand to copper charge transfer (copper-LMCT)-enabled decarboxylative carbometalation. The transformation proceeds under mild reaction conditions, has a broad substrate scope, and can be applied to late-stage functionalization of complex small molecules.


Introduction
Coordination of substrates including alcohols, 1 halides, 2 azides, 3 and alkyl carboxylates 4 to abundant 3d metal salts like Fe, Ni, or Cu can form photoactive metal complexes. Promoted to their excited states upon irradiation can result in intramolecular ligand-to-metal charge transfer (LMCT) within the excited state complexes to generate reactive open-shell radical intermediates with reactivity hardly reachable in the ground state. 1-5 Conventional metal-catalysed or -mediated thermal decarboxylative cross-coupling reactions normally require high reaction temperature and ortho-substituents. 6 Radical aromatic decarboxylation proceeds about three orders of magnitude slower than from aliphatic carboxyl radicals, 7,8 which generally leads to undesirable side reactions such as hydrogen atom abstraction for benzoyl radicals. 7 As a consequence, many aromatic decarboxylative bond forming reactions, including sulfoximination, are still out of reach for conventional reaction chemistry. Here we report the rst decarboxylative sulfoximination of benzoic acids enabled by photo-induced ligand to copper charge transfer. Photoactive copper(II) carboxylates undergo a low-barrier radical CO 2 extrusion upon irradiation, with the putative formed aryl radicals subsequently captured by copper complexes to generate CuAr(III) species for C-N reductive elimination. The synthetic utility of this method was exempli-ed by late-stage decarboxylative sulfoximination of several complex small-molecule benzoic acids, which are abundantly available from nature.
Photonic excitation of copper(II) complexes have been known as an effective platform for generating reactive radicals for decades. Kochi rst studied the addition and C sp 3 -H abstraction reactivity of chlorine radical generated by the photo-irradiation of CuCl 2 in different organic solvents. 2a Based on this initial nding, Wan and co-workers developed a vicinal dichlorination of alkenes catalyzed by CuCl 2 under air, 2b and the Rovis group realized a copper catalyzed olenation of unactivated C sp 3 -H bonds. 2c In addition to chlorine radicals, copper-LMCT is also suitable for N-or C-centered radical generation. In 2018, Rehbein and Reiser found that copper-LMCT was effective for azide radical generation, 3 and Wang and Xu suggested the formation of N-centered radical cations via intramolecular LMCT of quinolinyl-8-glycinate ester coordinated alkyl-Cu(III) adducts. 9 For C-centered radicals, Gong and co-workers proposed the generation of alkyl radical intermediates via photolysis of Cu(II)-alkyl complexes. 10 Although copper-LMCT in copper(II) carboxylate complexes was rst described by DeGraff and co-workers during their research on the photolysis of copper(II)-malonate, 4a,b it was not until 2021 that our group applied this copper-LMCT reactivity to synthetic applications. 11 We realized the rst aromatic decarboxylative uorination 11a and the decarboxylative hydroxylation 11b of benzoic acids. Concurrently, the MacMillan group explored the copper-LMCT reactivity in aromatic decarboxylative borylation 12a and halogenation, 12b and achieved copper catalysis for transformations with single electron oxidants such as 1-uoro-2,4,6trimethylpyridinium tetrauoroborate (NFTPT). Stoichiometric copper is still required for nucleophiles such as uoride. 11a Nearly at the same time, the Yoon group developed a coppermediated oxidative decarboxylative functionalization of aliphatic carboxylic acids. 4e Most recently, Reiser and co-workers reported a copper-catalysed aliphatic decarboxylative oxygenation methodology, where oxygen was applied as the oxidant. 4i While we and others have developed the concept of photo-induced copper-LMCT-enabled aromatic radical decarboxylation to achieve previously unknown reactivity, the coupling counterparts are generally limited to halides, carboxylates or boronate esters but strong coordinating NHnucleophiles have not been shown to react (Fig. 1A). 11,12 Given that copper-LMCT relies on the coordination of carboxylates to copper(II) species, strong coordination of NH-nucleophiles such as NH-sulfoximines will compete with carboxylates and form undesired copper species that can diminish the reaction efficiency. Though thermal aromatic decarboxylative C-N cross couplings under high reaction temperatures (normally more than 140°C) have been explored by Jia, Gooben, and Xie, electron-decient ortho-substituted benzoic acids are required for efficient CO 2 extrusion (Fig. 1A). 13 Thus, a general aromatic decarboxylative C-N cross coupling still remains elusive.
The study of sulfoximines dates back as far as 1949 when methionine sulfoximine was rst synthesized by Bentley and Whitehead. 14 Today, sulfoximines play a remarkable role in synthetic chemistry 15 and drug discovery; 16 they have been applied as chiral auxiliaries, 17 chiral ligands, 18 asymmetric organocatalysts 19 and building blocks. 15,20 N-arylated sulfoximines are mainly prepared by transition-metal-catalyzed direct N-arylation of NH-sulfoximines with aryl (pseudo)halides, 21 arylboronic acids, 22 aryl siloxanes, 23 acyl peroxides, 24 arylsulnates, 25 and aryl hydrazides 26 but cannot be accessed from benzoic acids (Fig. 1B). Herein, we present the rst decarboxylative sulfoximination of benzoic acids by applying a photoinduced carboxylate-to-copper charge transfer strategy (Fig. 1C). We found that lithium carboxylates with 2,6-di-tertbutylpyridine (DTBP) and LiOMe as additives was able to overcome the challenging low reaction efficiency associated with copper-LMCT-enabled aromatic decarboxylative sulfoximination.

Results and discussion
Because of the enhanced N-H acidity, NH-sulfoximines can undergo facile deprotonation and readily coordinate with copper(II) species. 16c In the copper-LMCT process, competing coordination of sulfoximines to copper(II) might hinder the formation of key copper(II) carboxylate intermediates and, in turn, decrease the reaction efficiency. We hypothesized that initial deprotonation of the benzoic acids to their carboxylate salts might facilitate the generation of photoactive copper(II) carboxylates, while the careful screening of additives can hinder the formation of undesired sulfoximine-ligated copper(II) species.
As shown in Table 1, we veried our assumption via the decarboxylative sulfoximination of lithium 4-uorobenzoate (1). A series of reaction condition optimization (see ESI for more details †) reveals that purple light irradiation of a mixture of 1, NH-sulfoximine 2, Cu(OTf) 2 , LiOMe and 2,6-di-tert-butylpyridine (DTBP) in MeCN can afford N-arylated sulfoximine 3 in 66% yield, together with side product ester 3a in 8% yield (Table  1, entry 1). Compared with 4-uorobenzoic acid, 4-uorobenzoate salts underwent more efficient decarboxylation, and the lithium salts gave the best yield (entries 1-3). The result is consistent with our hypothesis that the use of benzoate salts can promote the formation of copper carboxylates and increase the reaction efficiency. Notably, the combination of bulky DTBP and LiOMe is crucial to obtain a high yield. Replacing DTBP/  LiOMe with other inorganic or organic bases generally led to lower yields, and a large amount of side oxydecarboxylation product 3a was detected (entries 4-6). Strong coordinating bi-or tridentate pyridine-based ligands or bidentate PyBOX ligand were also tested, yet no better yields were observed (see ESI for more details †). We assume that weak ligation of bulky DTBP to copper might favour C-N reductive elimination over C-O reductive elimination or help to form photoactive copper(II) carboxylates species. The role of LiOMe is not very clear; we propose that the addition of LiOMe might help decrease the concentration of free sulfoximines by forming poorly soluble sulfoximine lithium salts and in turn, accelerate the formation of copper(II) carboxylate species. Low conversion of starting substrates 1 and 2 was observed when copper sources including Cu(OAc) 2 were used instead of Cu(OTf) 2 (entry 7). Interestingly, only MeCN as solvent was productive, and DCM only afforded 12% protodecarboxylation side-product uorobenzene (entry 8).
Preliminary investigations to study the reaction mechanism are consistent with the mechanism shown in Fig. 3. In the UVvis absorption spectrum of the mixture of lithium 4-uorobenzoate (1) and Cu(OTf) 2 , a strong absorbance (370-470 nm) attributed to the LMCT band of copper(II) carboxylates was detected (Fig. 3A). 28 The LMCT band overlaps with the purple LED emission spectrum, consistent with the excitation of copper(II) carboxylates under the reaction conditions. The coordination of sulfoximines to copper(II) and the coordination of 2,6di-tert-butylpyridine (DTBP) to copper(II) are in agreement with the observation of an absorbance (370-470 nm) of a mixture of sulfoximine 2 and Cu(OTf) 2 , and a mixture of DTBP and Cu(OTf) 2 (Fig. 3A). All copper(II)-containing mixtures display a broad d-d transitions absorbance at 550−900 nm, which decreased monotonously upon purple LED irradiation, consistent with the reduction of Cu(II) to Cu(I) (Fig. 3B). The formation of Cu(I) was conrmed by the observation of a characteristic purple [Cu I (biq) 2 ] + complex when 2,2 ′ -biquinoline (biq) was added to the irradiated reaction mixture (see ESI for more detail †). 29 Additional experiments were then performed to explore the transformation of carboxylate ions upon irradiation. In radical trapping experiments, aryl carboxyl radical adduct phenyl 3-methoxybenzoate and aryl radical adduct 3-methoxy-1,1 ′ -biphenyl were separated, which indicated the formation of aryl carboxyl and aryl radicals (see ESI for more details †). The presence of aryl carboxyl radical and aryl radical was further conrmed by 6-endo-trig intramolecular radical cyclisation (see ESI for more details †) and radical deuterodecarboxylation (Fig. 3C), respectively. N-Phenyl-sulfoximine, possibly formed by radical addition of sulfoximinyl radical to benzene, was also identied in the radical trapping experiments. We hypothesize that the sulfoximinyl radical is formed via a nitrogen to copper charge transfer in the copper(II) sulfoximine complex, with consumption of the copper(II) species. This result is consistent with competing coordination of sulfoximines to copper(II) and may also explain the low reaction reactivity caused by the competing coordination. Based on the above mechanistic investigation, we propose a mechanism as depicted in Fig. 3D. Photo-induced carboxylate to copper(II) charge transfer in copper(II) carboxylates I affords aryl carboxyl radical intermediates II, which then undergo low-barrier radical decarboxylation to afford aryl radicals III. Subsequent copper-assisted aryl radical capture generates arylcopper(III) intermediates IV that nally undergo C-N reductive elimination to afford N-arylated sulfoximines (Fig. 3D). Additionally, we found that other N-nucleophiles, such as ortho-sulphobenzamide, could also be coupled by the copper-LMCT approach (Fig. 3E).

Conclusions
Copper-LMCT based radical aromatic decarboxylative carbometalation enabled the rst decarboxylative sulfoximination of benzoic acids. The broad substrate scope and good functional group tolerance demonstrate the generality of the copper-LMCT concept in aromatic decarboxylative sulfoximination. Conceptually, the success of this transformation demonstrates the expansion of the copper-LMCT concept for aromatic decarboxylative cross-couplings to reactions with strongly coordinating nucleophiles.

Data availability
Procedures and compound characterization are provided in the ESI. †

Author contributions
P. X. initiated the project and performed experiments and analyzed the data. W. S. performed and analyzed experiments regarding the mechanism. T. R. directed the project.

Conflicts of interest
The authors declare no competing interests.