1,4-Aryl migration in ketene-derived enolates by a polar-radical-crossover cascade

The arylation of carboxylic acid derivatives via Smiles rearrangement has gained great interest in recent years. Both radical and ionic approaches, as well as radical-polar crossover concepts, have been developed. In contrast, a reversed polar-radical crossover approach remains underexplored. Here we report a simple, efficient and scalable method for the preparation of sterically hindered and valuable α-quaternary amides via a polar-radical crossover-enolate oxidation-aryl migration pathway. A variety of easily accessible N-alkyl and N-arylsulfonamides are reacted with disubstituted ketenes to give the corresponding amide enolates, which undergo upon single electron transfer oxidation, a 1,4-aryl migration, desulfonylation, hydrogen atom transfer cascade to provide α-quaternary amides in good to excellent yields. Various mono- and di-substituted heteroatom-containing and polycyclic arenes engage in the aryl migration reaction. Functional group tolerance is excellent and substrates as well as reagents are readily available rendering the method broadly applicable. The α-arylation of amides via aryl migration has attracted considerable interest in recent years. Here, the authors report a method for the preparation of bulky α-quaternary amides via a polar-radical crossover enolate oxidation-aryl migration cascade.

Complementary to radical aryl migrations, anionic 1,4-aryl migrations can also be applied for C(sp 2 )−C(sp 3 ) bond formation. Dohmori and coworkers observed anionic aryl migration in sulfonamide enolates as early as the 1950s [69][70][71][72][73][74][75][76] . More recently, Clayden's group developed a valuable method for the stereoselective arylation of the α-C(sp 3 )−H bond in imidazolidinones (Fig. 1b) 77 . These reactions proceed via deprotonation (enolate formation) and subsequent anionic 1,4-aryl migration from nitrogen to carbon. The same group recently introduced an elegant method for the preparation of α-arylated amines via radicalpolar crossover (Fig. 1c) 78 . The cascade is initiated by a radical addition to an eneurea with subsequent reduction of the adduct radical to the corresponding anion followed by ionic 1,4-aryl migration. The switching from the radical to the anionic mode in such cross-over processes may offer advantages and will also open new opportunities in reaction design.
Keeping the crossover benefits in mind, we herein present a reversed polar-radical crossover strategy that uses an anionic addition as the initial step of the cascade with subsequent SEToxidation and radical translocation, harvesting the great potential of the radical aryl migration reaction. As substrates, we are using ketenes that can readily be prepared by established methodology in a one-pot sequence via acid chlorides easily obtained from commercially available carboxylic acids. As discussed above, ketenes serve as efficient enolate precursors. Hence, the addition of a lithiated arylsulfonamide to a ketene leads to the corresponding sterically demanding α-alkyl-α-aryl or α,α-dialkyl amide enolate (Fig. 1d). The polar-radical crossover is achieved by enolate SET-oxidation which is followed by a radical 1,4-aryl migration. SO 2 -extrusion and hydrogen atom abstraction will complete the sequence. Of note, the sulfonamides are prepared in high yield in an easy and scalable reaction from inexpensive amines and aryl sulfonyl chlorides, which are commercially available in great variety. Therefore, this process allows to prepare a large number of all carbon α-quaternary amides in a simple one-pot process starting from easily accessible precursors.

Results and discussion
Optimization study. We chose N-isopropylbenzenesulfonamide 1a and ethyl phenyl ketene 2a as model substrates for reaction optimization. The sulfonamide 1a was first deprotonated with n-BuLi (1.1 eq.) in THF and then mixed with the preformed ketene 2a. SET oxidation of the enolate was initially attempted by irradiation (456 and 467 nm) in the presence of a photoredox catalyst, [Ir(dF(CF 3 )ppy) 2 (dtbpy)]PF 6 (5 mol%) or [Ru(bpy) 3 ](PF 6 ) 2 (5 mol%) ( Table 1, entry 1 and 2). Disappointingly, the desired αarylated amide 3aa was not formed in both cases. We next tested   the SET-oxidation of the intermediate enolate with stoichiometric CuCl 2 (1.2 eq.) and ferrocenium tetrafluoroborate (FcBF 4 , 1.2 eq.). Pleasingly, the targeted amide 3aa was obtained in encouraging 22% and 20% yield, respectively (Table 1, entry 3 and 4). As a side product, the ketene dimer was observed in these transformations. To suppress dimer formation, the concentration of the reaction mixture was decreased to 0.01 M and yield improved to 38% with CuCl 2 (1.2 eq.) and further to 60% with FcBF 4 (1.2 eq.) ( Table 1, entry 5 and 6). When the ketene is slowly added as a solution in THF (0.06 M) over half an hour by syringe pump, the yield minimally increased to 61% with 1.0 eq. of FcBF 4 (Table 1, entry 7). By using a sub-stoichiometric amount FcBF 4 , only traces of the product 3aa were obtained ( Table 1, entry 8 and 9). Next, the influence of the solvent was investigated and the desired 3aa was not formed in toluene, whereas in benzene the yield decreased to 32% (Table 1, entry 10 and 11). To our delight, in diethyl ether under otherwise identical conditions, a quantitative conversion was achieved, as analyzed by GC-FID and the amide 3aa was isolated in an excellent 94% yield (Table 1, entry 12). The reduced oxidant, ferrocene, which can readily be re-oxidized to the ferrocenium ion, was recovered in 96% yield. In a control experiment in the absence of oxidant, product 3aa was not identified, which indicates the radical nature of the aryl migration excluding an anionic arylation pathway (Table 1, entry 13). A further control experiment in absence of base also resulted in no product formation (Table 1, entry 14). Substitution of n-BuLi with milder and non-nucleophilic bases such as NaH and DBU also shuts down the reaction (Table 1, entry 15 and 16). Only with LiHMDS was product formation observed, but with a reduced yield of 81% (Table 1, entry 17), suggesting that the Lienolate is a crucial intermediate in the cascade described herein.
Even though the inexpensive and commercially available stoichiometric FcBF 4 can be near quantitatively recovered and subsequently recycled, further attempts were made to develop a variant of the reaction using catalytic amounts of an oxidant. Initial screenings using photoredox catalysis in the absence of any cooxidants (Table 1, entry 1 and 2) failed. Further addressing such a process, we tested other photocatalysts with both stronger reducing and oxidizing power, but none of them led to the formation of the desired product (Table 1, entries [18][19][20][21][22]. We then envisioned to establish a dual catalysis cycle using photoredox catalysis to in situ recycle the ferrocenium oxidant. The intermediate amidyl radical formed after the aryl migration sequence might regenerate the photoredox catalyst under formation of the corresponding N-anion. Various photocatalysts in combination with 20 mol% of the ferrocenium oxidant were screened and a maximum yield of 8% was achieved (   added over a period of two hours by syringe pump the yield was further improved to 48% ( Table 2, entry 8). Lowering the catalyst loading of the FcBF 4 led to a reduced yield ( Table 2, entry 9), and increasing the amount of FcBF 4 to 30 mol% also did not affect the reaction outcome (Table 2, entry 10). We next tried stochiometric co-oxidants as additives (BrCCl 3 , tert-butyl hydroperoxide (TBHP) and K 2 S 2 O 8 ) but were not able to further increase the yield ( Table 2, entries 11-13). The addition of (TMS) 3 SiH or Bu 3 SnSnBu 3 as H-atom donors or trapping reagents for the amidyl radical to form silyl or stannyl radicals, respectively, which can subsequently be reduced by the photoredox catalyst to close the catalysis cycle, did also not increase the yield of 3aa (Table 2, entries 14 and 15).
Substrate scope. We first investigated the scalability of the method. For this purpose, the model substrate 1a and ethyl phenyl ketene 2a were converted to the amide 3aa under the optimized conditions on a 5.0 mmol scale. Compared to the small-scale experiment (0.2 mmol), a comparatively excellent yield of 91% (1.28 g) was obtained (Fig. 2). Next, we investigated the substrate scope keeping 2a as the ketene component. First, different substituents on the N-atom of the benzenesulfonamide were investigated. In this series, N-alkylbenzene-sulfonamides 1af were found to be the most efficient substrates and bulkier alkyl substituents on the N-atom lead to higher yields most likely due to conformational effects. Thus, for the sterically least demanding N-methyl derivative 3ab a yield of 57% was noted, whereas the Nbenzyl and N-trifluoroethyl derivatives 3ac and 3ad gave a 78% yield each. In case of the N-methyl derivative 3ab protonated enolate was observed as the side product. The best result (98% yield) was obtained for the N-cyclohexyl derivative 3ae and the N-dioxanyl derivative 3af could be isolated in 54% yield. The substrate bearing the sterically demanding N-tert-butyl group provided the aryl migration product 3ag in only 33% yield and unreacted sulfonamide 1g was observed. Thus, the high steric demand of the tert-butyl substituent negatively influences initial enolate formation. N-Aryl amides also engage in this cascade,  albeit with lower efficiency, as documented by the preparation of 3ah (26%).
Next, it was investigated whether a diastereoselective aryl migration is feasible by using enolates generated with chiral lithium amides. To this end, p-tolylsulfonamides derived from (S)-phenethylamine and valine methyl ester were reacted under the optimized conditions. The corresponding amides 3ai and 3aj were isolated in high yields (78-80%) as a 1:1 mixture of the two diastereoisomers. Guided by the Myers alkylation 79-81 , p-tolylsulfonamides 1k-n derived from pseudonorephenamine were prepared and subjected to the aryl migration sequence in combination with ketene 2a. However, reaction did not work for the O-unprotected compound (3ak) and for the O-protected congeners, diastereoselectivity could not be controlled (3al-an).
Studies were continued by investigating the substrate scope with respect to the migrating aryl group with 2a as the ketene component. p-Methyl, p-fluoro and p-chloro substituents are tolerated on the migrating aryl moiety and the corresponding amides were isolated in 70% (3ao), 63% (3ap) and 84% (3aq) yield (Fig. 3). The method tolerates both electron-deficient and electron-rich arenes as migrating units and the p-CF 3 -and p-methoxy-phenyl substituted amides 3ar and 3as were isolated in 87% and 69% yield. Lower yields were noted for the p-nitro, p-cyano, and p-amido congeners (see 3at, 3au, and 3av). Substituents in ortho and meta position on the migrating aryl moiety are also tolerated (3aw, 67%; 3ax, 76%; 3ay, 49%). Reaction with the sterically hindered 1-naphthylsulfonamide was less efficient and the product 3az was isolated in 19% yield, but the 2-naphthyl congener 3aaa was obtained with 73% yield. The migration of heteroarenes was found to be lower yielding. Thus, 2-thienyl (3aab), 2-benzothienyl (3aac), and 2-benzofuryl (3aad) amides were obtained in 27-43% yield. The pyridyl and benzothiazolyl migration products 3aae and 3aaf were formed in trace amounts only, as detected by ESI-MS.
Finally, different ketenes were tested and 1a was selected as the sulfonamide component in this series (Fig. 4). Starting with alkyl phenyl ketenes 2b-d, the methyl-, isopropyl-and cyclopentylphenyl amides 3ba-da were obtained in 29-63% yields. A tetrahydronaphthyl derivative could also be accessed (see 3ea, 42%) and the p-iodo-3fa as well as the p-bromophenylamide 3ga were successfully prepared by using ethyl p-iodophenylketene 2f and ethyl p-bromophenylketene 2g as precursors. Moreover, the ibuprofen® derivative 3ha was isolated in good yield (75%). However, diphenyl ketene 2i is not a suitable substrate and the corresponding amide 3ia could be detected in traces only by ESI-MS. The higher stability and the larger steric demand of the α,αdiphenylated α-amide radical likely prevents the aryl migration. While most alkyl aryl ketenes are rather stable compounds that  can be stored in the freezer under argon atmosphere for some time, dialkyl ketenes are reactive intermediates that have to be used in situ directly after their generation. We were pleased to find that dialkyl ketenes could also be successfully applied to the enolate formation aryl migration cascade. For example, the methyl benzyl derivative 3ja was obtained in 49% yield. Better yields were noted for the methyl cyclohexylamide 3ka and the cycloheptane derivative 3la (74-81%).
We made the experience that amide hydrolysis in our case was very challenging. Addressing that issue, amide 3af with the assistance of the hydroxyl groups liberated after acetal hydrolysis was readily hydrolyzed under acidic conditions to the corresponding α-quaternary carboxylic acid 4 in 71% yield (Fig. 5). Thus, with this strategy a variety of α-quaternary carboxylic acids can be prepared, further improving the applicability of the introduced method.
Reaction mechanism. Our suggested mechanism for the polarradical crossover cascade is presented in Fig. 6. First, the sulfonamide 1a is deprotonated with n-BuLi and the resulting Li-amide 1a-Li then adds to the ketene to generate the enolate A. The enolate A is oxidized by the ferrocenium ion (Fe III , E 1/2 (Fc/ Fc + ) = 0.32 V vs. SCE 82 ) to generate the α-amide radical B 83-90 , which attacks the arene at the ipso-position to give the spirocyclic intermediate C. Homolytic cleavage of the C−S bond and extrusion of SO 2 lead to the amidyl radical D 64,67,68 . Since ferrocene (Fe II ) is obtained as a by-product in stoichiometric amounts, the amidyl radical D is not efficiently reduced by ferrocene preventing the Fe-salt to act as a redox catalyst. Therefore, we assume that the amidyl radical D further reacts via hydrogen abstraction 67 from Et 2 O to finally afford product amide 3. This may also explain why ethereal solvents provide the highest yields.
In summary, we presented a simple, efficient and scalable method for the preparation of sterically hindered and synthetically valuable α-quaternary amides via a polar-radical crossover enolate oxidation-aryl migration sequence. Ketenes were introduced as highly valuable precursors for the generation of α-amide radicals. The starting materials, both the ketenes and the arylsulfonamides, are easily prepared from a wide variety of commercially available compounds. More than thirty N-alkyl as well as N-arylamides bearing an all carbon α-quaternary center were readily prepared via this cascade, convincingly documenting the broad applicability of the method and also showing the excellent functional group tolerance. The method convincingly shows the potential of polar-radical crossover processes in organic synthesis by merging valuable anionic with equally important radical steps.

Data availability
Supplementary information and chemical compound information accompany this paper at www.nature.com/ncomms. The data supporting the results of this work are included in this paper or in the Supplementary Information and are also available upon request from the corresponding author.  Fig. 6 Plausible mechanism of the polar-radical crossover cascade. The reaction proceeds through anionic (1a-Li and A) and radical intermediates (B, C and D).