Introduction

Benzylamines are an important sub-class of amines and are frequently encountered in both natural products and synthetic bioactive molecules1. The development of general and sustainable catalytic systems for the synthesis of benzylamines is of great importance due to their broad distributions in pharmaceuticals that can be used to treat a variety of diseases such as depression, dementia, cardiovascular diseases, and bacterial infections (Fig. 1a)2,3,4. Compared with commonly used strategies such as the Gabriel synthesis and reductive amination5,6, the direct arylation of α-amino C(sp3)–H bonds is a straightforward way to construct benzylamine motifs through C–C bond formation, obviating the tedious substrate pre-functionalizaiton steps and thereby substantially streamlining the synthetic efforts7,8. Despite notable recent achievements9,10, the development of novel arylation strategies leading to the site-selective C–H bond differentiation and conversion among a series of α-amino C(sp3)–H bonds with very similar characteristics, is of great importance for the synthesis of valuable benzylamines, yet still a long-standing challenge facing synthetic chemists.

Fig. 1: Site-selective α-amino C(sp3)–H arylation for the construction of benzylamines.
figure 1

a Ubiquitous benzylamine structural motifs in pharmaceuticals. b Site-selective α-amino C(sp3)–H arylation reactions: state of the art and challenges to be achieved. c Our new C(sp3)–H arylation strategy: radical aryl migration through hydrogen atom transfer catalysis. d This work: radical aryl migration of ureas through photoredox-mediated HAT catalysis for site-selective α-amino C(sp3)–H arylation and benzylamine synthesis.

Full size image

A literature survey reveals that the known methods for the synthesis of benzylamines through intermolecular α-amino C(sp3)–H arylation of acyclic dialkylamides tend to perform selectively at the sterically less-hindered positions11, among which the N-methyl C–H bonds with the least steric hindrance are most likely to be functionalized (Fig. 1b)12. For example, the well-known transition metal catalysis in combination with directing groups selectively functionalizes N-methyl C–H bonds of dialkylthioamides13,14. In addition, free-radical involved approaches using the strategy of metallaphotoredox catalysis or dual radical coupling also exhibit similar site-selectivity because the C–C bond formation steps in these transformations are kinetically favorable for sterically less-hindered positions15,16,17,18,19,20,21. In particular, the activated methylene C(sp3)–H bonds in benzylamine motifs tend to be arylated due to the relatively weak benzylic C–H bonds22,23,24,25,26. However, to the best of our knowledge, the general strategy that allows the C–H arylation reactions perform predominantly on sterically more-hindered secondary or tertiary C(sp3)–H bonds, rather than the functionalization of a coexisting N-methyl group, has yet to be achieved. Given the fact that the N-methyl moiety is ubiquitous in life science molecules and plays an essential role to modulate the biological and pharmaceutical activities of lead compounds27,28,29, the development of novel C–H arylation strategy to navigate this question and synthesize valuable methylbenzylamines is of great importance for drug discovery30,31,32.

Given the inherent propensity for C–H site-selectivity in intermolecular C–C couplings, we envisioned that the intramolecular aryl migration from a heteroatom to a carbon center via radical ipso substitution—a process termed radical Truce-Smiles rearrangement—would be a promising strategy to enable a complementary approach for α-amino C(sp3)–H arylation that tends to perform at sterically more-hindered positions33,34,35,36,37,38,39. Nevertheless, the general methodologies for inert C(sp3)–H arylation through radical aryl migration remain rare. The elegant approaches separately pioneered by Studer and Zhu use extraneous directing groups as radical handles for geometry-controlled intramolecular HAT, thereby allowing distal C(sp3)–H (hetero)arylation40,41,42,43. However, to the best of our knowledge, the catalytic approach for C(sp3)–H arylation via radical aryl migration, especially without the use of extraneous functional groups for intramolecular radical translocation, has remained elusive to date. Within this context, we considered that the strategy of hydrogen atom transfer (HAT) catalysis, in which a catalytic amount of mediator is used for intermolecular hydrogen atom abstraction, thereby functionalizing C(sp3)–H bonds through free radical pathways, should have potential to achieve inert C(sp3)–H arylation through radical aryl migration (Fig. 1c)44,45,46,47. Given that the hydridic and relatively strong α-amino C(sp3)–H bonds can be homolytically cleaved by a range of HAT reagents (α-amino C–H bonds BDE ≈ 92 kcal mol)48, we speculated that the α-amino C(sp3)–H arylation through radical aryl migration would be achieved by HAT catalysis using the photoredox condition to regulate the single electron transfer processes49.

Inspired by Clayden’s elegant achievements in the field of anionic 1,4-aryl migration of ureas50,51,52,53, we assumed that the tetrasubstituted ureas derived from dialkylamines could be employed in our site-selective C–H arylation protocol (Fig. 1d). It should be noted that the intrinsic limitation in terms of the enthalpically disfavored deprotonation step substantially restricts the direct functionalization of inert C(sp3)–H bonds via anionic Truce-Smiles rearrangement. By contrast, the α-amino C–H homolytic cleavage step governed by polar effects and C–H bond dissociation energies (BDEs) is kinetically favorable through cooperation with electrophilic HAT reagents54,55. Therefore, we assumed that the initial intermolecular H atom abstraction could be regulated as the rate-determining step by delicately tuning the electrophilic properties and steric hindrance of HAT catalysts, thereby leading to the site-selective C–H differentiation and conversion through radical aryl migration56,57. Herein, we report the use of photoredox-mediated HAT catalysis to achieve the site-selective α-amino C(sp3)–H arylation of secondary amine-derived ureas via intramolecular radical aryl migration. This transformation exhibits specific C–H site-selectivity, in which the secondary and tertiary C–H bonds with higher steric hindrance are preferentially arylated and the alternative conversion of coexisting N-methyl moiety is effectively circumvented. In addition, the use of an extraneous functional group (e.g., sulfonyl group) to facilitate the aryl migration by following an irreversible chemical step (e.g., SO2 extrusion) is circumvented in this protocol, thus offering a new perspective for the discovery of radical aryl migration reactions58,59,60,61,62,63. Due to the exceptional site-selectivity in combination with the easily-handling operations for deprotection, we believe that this protocol would provide an expedient approach for the construction of a diverse array of benzylamines that are highly valuable but inaccessible from known methods.

Results

Reaction condition optimization

We commenced our studies with a systematically evaluation of various HAT catalysts, photocatalysts, solvents, and bases in the presence of methylethylamine-derived urea 1 as the model substrate. We found that the use of blue light (from a 40 W blue LED light cylinder) in the presence of 4-DPAIPN (PC-I), triisopropylsilanethiol (iPr3Si–SH), and Na2CO3 enabled the C(sp3)–H arylation in excellent yield (99%) to furnish the desired product 2 (Table 1, entry 1). To our delight, this C–H arylation reaction performed exclusively on methylene position without the detection of any N-methyl C–H functionalization product. The necessity of each reaction components—photocatalyst, thiol, base, and light—was demonstrated through a series of control experiments (Table 1, entries 2–5). In addition, the inert reaction atmosphere is required for optimal reaction performance (Table 1, entry 6). To our surprise, the reaction efficiency dramatically decreased when a series of cyanoarene-based donor-acceptor photocatalysts were used to replace 4-DPAIPN (Table 1, entries 7–9). We speculated that the strong reducing ability of reduced-state photocatalyst 4-DPAIPN•− (E1/2(P/P−) = −1.52 V vs SCE in MeCN) could effectively facilitate the product formation step through SET reduction64. A range of thiols including silanethiols, aryl thiols and alkyl thiols were ineffective to initiate the aryl migration (Table 1, entries 10), presumably ascribed to their lower S–H bond dissociation energies (BDEs) compared with iPr3Si–SH, thus resulting the HAT process thermodynamically unfavorable65,66. Notably, using quinuclidine in lieu of iPr3Si–SH was incapable for reaction initiation (Table 1, entries 11), although the H atom abstraction by quinuclidine radical cation is polarity-matched according to the previous reports16.

Table 1 Condition optimization and control experimentsa
Full size table

Substrate scope of acyclic amine-derived ureas

With the optimal conditions in hand, we next sought to examine the scope of this site-selective C(sp3)–H arylation protocol using a variety of ureas as substrates. As shown in Fig. 2, the α-amino C(sp3)–H bonds function exclusively on methylene positions and the methyl C–H arylation products are not detected. For example, the acyclic alkyl groups with different length or bearing terminal aliphatic rings provided the desired C–H arylation products in good to excellent yields (3–8, 54–98% yield). An array of additional functional groups, such as phenyl, terminal alkene, ether, amide, cyano, ester, thiophene and pyridine, were all well tolerated (9–16, 40–83% yield). Notably, the site-selectivity was not interfered by varying the electronic property and steric hindrance of the migrating aryl groups, providing the methylene C-H arylation products exclusively (17–20, 54–96% yield). Not surprisingly, diethyl amine-derived ureas engaged in this transformation with ease (21 and 22, 98 and 62% yield, respectively), which suggested us to further evaluate the variants of urea substitution patterns in this C–H arylation protocol. It is noteworthy that the secondary amide substrates containing N–H bonds are unable to perform this transformation, presumably due to both the high BDEs of these C–H bonds and the preferential conformation of secondary amide moiety where the alkyl group tends to be proximal to the carbonyl oxygen, which is unfavorable for aryl migration (for experimental details, see Supplementary Fig. 20).

Fig. 2: Scope of acyclic amine-derived ureas in site-selective C(sp3)–H arylation.
figure 2

Reaction conditions: urea (0.2 mmol), 4-DPAIPN (2 mol%), iPr3SiSH (30 mol%), and Na2CO3 (2.5 equiv.) in CH3CN (0.1 M) at room temperature for 24 h under 450 nm blue LED light irradiation. Site-selectivity were recorded by 1H NMR spectroscopy analysis. Isolated yields. If not specified, the products are obtained with >99:1 site-selective ratios.

Full size image

We next explored the site-selectivity between tertiary and primary C–H bonds in this transformation. We found that the isopropylmethylamine-derived urea gave moderate site-selectivity, which appears a lack of practicability (for more details, see Supplementary Fig. 20). To our delight, the tertiary C(sp3)–H bonds on cyclobutyls were arylated exclusively, even in the presence of a competing N-methyl or N-ethyl group (23–25, 83–98% yield). We believe that this outcome provides a reliable approach to synthesize various cyclobutyl-tethered benzylamines that can be employed as isosteres in drug discovery67,68. The tertiary C–H selectivity was also observed on cyclopentyl (26, 62% yield), cyclopentenyl (28, 56% yield) and 2-indane (29, 66% yield), highlighting the potence of this C(sp3)–H arylation protocol. Notably, the site-selectivity slightly diminished with the engagement of 3-gem-difluorocyclopentyl (27, 81% yield, 3°:1° = 12:1). We speculate that the presence of electron-withdrawing gem-difluoromethyl group increases the thermodynamical hydricity of the tertiary α-amino C–H bond, thereby retarding the polar effects-controlled HAT process69.

Next, we sought to examine the site-selective α-amino C(sp3)–H arylation of N,N’-diaryl ureas for the preparation of N-benzylaniline derivatives. To our delight, the radical aryl migration performed predominately at the methylene positions rather than the alternative N-methyl moiety on the other side of urea scaffolds (30, 70% yield). The acquirement of excellent site-selectivity was unaffected by varying the electronic density of the migrating aryl groups (31 and 32, 43 and 41% yield, respectively). Remarkably, the tertiary C–H bonds in isopropyl, cyclobutyl and cyclopentyl were preferentially functionalized with the coexistence of a competing N-methyl moiety on the other side of the urea scaffolds. We believe that these results further demonstrate the value of this site-selective C(sp3)–H arylation strategy.

Substrate scope of cyclic amine-derived ureas

We next examined the generality of this C–H arylation protocol using various cyclic amines that are prevalent in drug molecules and natural products (Fig. 3)70,71,72. To our delight, the aryl group migrated smoothly to produce 2-arylated piperidine derivative 36 in satisfactory yield (88% yield). The reaction efficiency was not influenced with the expansion of ring size (37 and 38, 87 and 97% yield, respectively). In addition, valuable functional groups introduced to the piperidine ring, such as carbonyl and gem-difluoromethyl, were well tolerated in this C–H arylation protocol (39 and 40, 79 and 72% yield, respectively). Furthermore, the variants of piperidine such as morpholine and piperazine provided the desired C–H arylated heterocycles, albeit with the decrease of reaction efficiency (41 and 42, 36 and 30% yield, respectively). Remarkably, the 2- or 4-substituted piperidines also engaged in this radical Truce-Smiles rearrangement, providing C–H arylation products with high diastereomeric ratios. For example, a range of functional groups on 4-position of piperidine, including methyl, phenyl, silyloxymethyl or ester, were well-tolerated to give trans−2,4-disubstituted piperidine derivatives (43–47, 77–95% yield). In addition, the urea substrate derived from 4-silyloxy piperidine provided the C–H arylation product with good diastereoselectivity (48, 87% yield, d.r. = 5:1). Notably, 2-substituted piperidines performed C–H arylation reactions predominantly at the sterically less-hindered methylene positions, affording 2,6-disubstituted piperidine derivatives with exclusive cis diastereoselective control (49 and 50, 47 and 62% yield, respectively). Moreover, the piperidine substituted with 3-ester participated in the C–H arylation protocol with 3.4:1 site-selectivity (51, 60% yield), and the good diastereomeric ratios were obtained for both regioisomers (51′, 18% yield). It should be noted that the urea derived from cis−3,5-dimethyl piperidine worked quite well to deliver the 2,3,5-trisubstituted piperidine derivative in excellent yield and trans selectivity (52, 95% yield).

Fig. 3: Scope of cyclic amine-derived ureas.
figure 3

Reaction conditions: urea (0.2 mmol), 4-DPAIPN (2 mol%), iPr3SiSH (30 mol%) and Na2CO3 (2.5 equiv.) in CH3CN (0.1 M) at room temperature for 24 h under 450 nm blue LED light irradiation. Site- and diastereo-selectivity were recorded by 1H and 2D NMR spectroscopy analysis. Isolated yields. If not specified, the products are obtained with >99:1 site-selective ratios.

Full size image

To demonstrate the origin of diastereoselective control for the C–H arylation of piperidine-derived ureas, we conducted the time-course experiments by recording the diastereomeric ratios of C–H arylation product 43 at fixed time intervals (see Supplementary Fig. 14 for details). At the start, the C–H arylation reaction tended to deliver the cis−2,4 diastereomeric isomer. Along with the prolong of reaction time, a diastereoselective epimerization of cis−2,4 product, presumably proceeding through the photoredox catalyzed reversible HAT approach, afforded finally the trans−2,4 piperidine 43 as the thermodynamically more stable product73. Notably, the similar epimerization strategy by Ellman produces cis-2,4-disubstituted piperidines, which is opposite to our approach. We speculate that the opposite diastereoselectivity obtained from our epimerization system is attributed to the presence of methylaminoformyl protecting group, which leads to the formation of trans−2,4 piperidine derivatives thermodynamically more favorable.

Scope of aryls in this C–H arylation protocol

We next sought to survey the scope of aryls in this radical C–H arylation protocol. As outlined in Fig. 4, a wide range of aryls function efficiently to provide corresponding 2-aryl piperidine derivatives. For example, both electron-donating and electron-withdrawing groups, containing methoxy, trifluoromethoxy, fluorine, chloride, ester, cyano and trifluoromethyl, were well tolerated to provide the desired products in moderate to excellent yields (53–59, 50%–98% yield). The efficiency for radical aryl ipso-substitution was not disrupted with the engagement of ortho- or meta-substituted aryl groups in this protocol (60–67, 32–93% yield). Notably, 2,6-diisopropyl phenyl and 2-naphthalene performed favorably (68 and 69, 86 and 83% yield, respectively), demonstrating that the aryl migration efficiency is unaffected by the steric hindrance of migrating aryls. To our delight, an array of heteroaromatic systems participated in this transformation with high efficiency, including quinoline, benzothiophene, dibenzofuran, thiophene and pyridine (70–76, 33–98% yield). We believe that this protocol featured with a broad scope of migrating aryl groups and represents a synthetically useful strategy for the preparation of benzylamines using C–H arylation logic.

Fig. 4: Scope of aryls in this C–H arylation protocol.
figure 4

Reaction conditions: urea (0.2 mmol), 4-DPAIPN (2 mol%), iPr3SiSH (30 mol%) and Na2CO3 (2.5 equiv.) in CH3CN (0.1 M) at room temperature for 24 h under 450 nm blue LED light irradiation. Isolated yields.

Full size image

Derivatization of drug molecules

Next, a range of complex drug molecules were engaged to further investigate the potential applications of this C–H arylation strategy in late-stage functionalization of biologically relevant systems (Fig. 5). Several drug molecules containing N-methyl group, such as fluoxetine, maprotiline, atomoxetine and nortriptyline were efficiently functionalized with exclusive methylene C–H selectivity. (77–80, 40–63% yield). We were also pleased to find that the bioactive molecules containing cyclic amine skeletons, such as risperidone core and cytosine, were also engaged in this C–H arylation protocol and the desired products were obtained in good yields and diastereoselectivity (81 and 82, 61 and 42% yield, respectively). Notably, the regioisomers obtained from the paroxetine-derived urea were separable with overall satisfactory yields (83 and 83’, 93% overall yield), in which the minor regioisomer was obtained with excellent diaseteroselectivity (83’, 24% yield, d.r. > 20:1) whereas the equal amounts of diasteromeric isomers were observed from the major regioisomer (83, 69% yield, d.r. = 1:1).

Fig. 5: Derivatization of drug molecules.
figure 5

Site- and diastereo-selectivity were recorded by 1H and 2D NMR spectroscopy analysis. Isolated yields.

Full size image

Deprotection, mechanistic studies, and proposed reaction mechanism

The methylamino formyl group acts as an amino-protecting group of C–H arylation products and can be readily removed through hydrolysis operations. As shown in Fig. 6a, the treatment of C–H arylation product 44 with tert-butyl nitrite followed by hydrolysis with LiOH afforded the free piperidine product in excellent yield and without variation of diastereomeric ratios (84, 90% yield). Moreover, the direct hydrolysis with NaOH is also an expedient approach for protecting group removal (84, 94% yield). The cis−2,6-disubstituted free piperidine 85 was also obtained from C–H arylation product 49 using both hydrolysis methods (90 and 97% yield from methods A and B, respectively)74,75. Both methods to yield free amines exhibit remarkable functional group tolerance such as methoxy, cyano, pyridine, and alkene (see Supplementary Fig. 5).

Fig. 6: Protecting group removal experiments and mechanistic studies.
figure 6

a Protecting group removal experiments. b Stern-Volmer quenching studies. c Radical clock experiments. d Deuterium labeling experiments. e Competitive aryl migration experiments.

Full size image

Next, a series of mechanistic studies were performed to gain insight into the details of the reaction mechanism. Stern–Volmer luminescence quenching experiments of 4-DPAIPN with triisopropylsilanethiol and urea 1 show that the excited-state photocatalyst tends to be quenched by the thiol preferentially (Fig. 6b). The reaction was fully shut down through the addition of TEMPO as a radical scavenger, thus supporting the radical nature of this cascade (see Supplementary Information for details). Notably, the quantum yield of this dual-catalysis system is determined to be 0.54 (Φ = 0.54), demonstrating that the reaction prones to undergo a radical-polar crossover mechanism instead of a radical chain propagation pathway (see Supplementary Information for details). This mechanistic scenario could be further verified by the time-course light on/off experiments, as the reaction suspension was observed in the light-off time intervals (see Supplementary Fig. 10). Moreover, the amidyl radical species is unlikely to be involved in the radical cascade because the ring-opening product was not detected using 2-phenylcyclopropyl amine-derived urea 86 for the radical clock experiment (Fig. 6c). The reactions of urea S-95 with either benzaldehyde or acetone couldn’t give any nucleophilic addition products, suggesting that the α-amido carbanion species may not be generated in our C–H arylation protocol (see Supplementary Fig. 11). This conclusion can be further supported by calculating the reduction potentials of representative secondary and tertiary alkyl radicals, as both are lower than −2.0 V (see Supplementary Fig. 19), which significantly exceed the reducing ability of 4-DPAIPN•− (E1/2(P/P–) = −1.52 V vs SCE and −1.60 V by calculation). In addition, the decarboxylative deuteration experiment attempting to capture the carbanion species with D2O didn’t yield the 2-deuterated piperidine product, further precluding the generation of carbanion species as well as the anionic aryl migration pathway (see Supplementary Fig. 11).

To investigate the kinetic characteristics of this transformation, a series of deuterium-labeling experiments were conducted using 2,6-deuterated piperidine substrate d4−88. Upon irradiation for two hours, the combination of 88 with the same amount of d4−88 in one pot provided the corresponding products 89 and 90 with around 5.7:1 relative ratio (Fig. 6d, above). Moreover, the kinetic isotope effect (KIE) studies using d4−88 as the substrate revealed a large primary KIE value (kH/kD = 3.1) (Fig. 6d, below). These results unambiguously demonstrate that the C−H bond cleavage through H atom abstraction is a rate-determining step, thereby leading to the site-selectivity for radical aryl migration. In addition, the intramolecular competing experiment reveals that the aryl groups with different electronic densities (p-CF3- and p-OMe-C6H4) exhibit comparable migration efficiency (92 and 93, 48% and 32% yield, respectively) and the electron-rich aryl groups migrate slightly faster than the electron-deficient ones. (Fig. 6e).

To further understand the excellent site-selectivity of this intramolecular C–H arylation reaction, we also calculated the free energy profile for the N-methyl C(sp3)–H arylation of 1 (Fig. 7). We found that the initial H atom abstraction towards a methyl C–H bond shows a lower activation free energy compared with methylene C–H abstraction pathway (15.8 vs. 16.9 kcal•mol-1). However, the ipso radical addition of the primary carbon radical (IM1’) to the distal aryl group exposes a higher activation free energy than the secondary carbon radical species IM1 (14.7 vs. 11.2 kcal•mol-1). These results indicate that the rate-determining step of N-methyl C–H arylation pathway should be the ipso aryl substitution whereas the methylene C–H abstraction step by the thiol radical predominantly determines the reaction rate of secondary C–H arylation. It is important to note that the entire activation free energy for N-methyl C–H arylation is higher than that of the methylene position (18.7 vs. 16.9 kcal•mol-1), suggesting that the methylene C–H arylation should be kinetically more feasible. In addition, the tertiary C(sp3)–H arylation of urea 23 is energetically more favorable than methyl C–H bond according to the calculation results, which further supports the site-selectivity in this radical C–H arylation protocol (entire activation free energy for tertiary and primary C–H bonds: 18.2 vs. 19.6 kcal•mol-1, see Supplementary Fig. 17 for details).

Fig. 7: Theoretical calculation of reaction mechanism (ethyl pathway, in black) with comparison with methyl pathway (in blue).
figure 7

DFT calculations were carried out at the M06-2X-D3/ma-def2-TZVPP//M06-2X-D3/def-TZVP level of theory with continuum solvent model for CH3CN. Free energies referring to the change of free energy in the corresponding reactions are given in kcal mol−1.

Full size image

According to the results of mechanistic studies and DFT calculations (Fig. 8), a detailed mechanism for HAT catalysis-promoted radical Truce-Smiles rearrangement is depicted in Fig. 8. Upon irradiation with blue light, the donor-acceptor fluorophore 4-DPAIPN is known to undergo a photo-excitation to generate the long-lived excited state 4-DPAIPN* (lifetime τ = 84 μs)76, which can function as an oxidant (E1/2(P*/P−) = +1.1 V vs SCE in MeCN) and can be quenched by the thiol catalyst iPr3Si–SH (E 1/2ox = +0.28 V vs SCE in MeCN) to produce 4-DPAIPN•– 64,77, as well as the thiol radical (iPr3Si–S•). The H atom abstraction on methylene position by TIPS–S• regenerates the thiol catalyst and produces simultaneously the secondary alkyl radical species IM1 via transition state TS1. The activation free energy for this HAT step is calculated as 16.9 kcal• mol-1 (Fig. 7). Next, the ipso radical addition of IM1 to a distal aryl group affords spiral radical intermediate IM2 with the activation free energy of 11.2 kcal•mol-1, which can further be reduced by 4-DPAIPN•‒ to produce a delocalized anion species IM3 and to close the photoredox cycle by the regeneration of the fluorophore. With aromatization as the driving force, the carbon anion IM3 is prone to undergo a β-scission process to finish the 1,4-aryl migration process and to produce the nitrogen anion species IM4 with free energy releasing of 17.9 kcal•mol-1. Finally, the rapid protonation of IM4 affords the desired C-H arylation product 2. An alternativeradical β-scission pathway of IM2 is ruled out because the activation free energy for this radical C–C bond cleavage step is calculated to be 17.8 kcal mol-1, which is kinetically unfavorable compared with the ionic pathway (see Supplementary Fig. 15–18 for more calculation details).

Fig. 8: Proposed mechanism.
figure 8

A radical aryl migration mechanism by HAT catalysis is proposed to elucidate the site-selective α-C(sp3)–H arylation of secondary amines.

Full size image

Discussion

In conclusion, we have achieved the site-selective α-amino C(sp3)–H arylation of secondary amines through HAT catalysis-enabled radical aryl migration. The ureas that can be easily prepared from amines are engaged smoothly in this protocol and this transformation exhibits unique site-selectivity. For acyclic dialkylamine-derived ureas, the secondary and tertiary C–H bonds are arylated exclusively without functionalizing a coexsiting N-methyl group. The similar site-selectivity is also observed for aniline-derived N,N’-diaryl ureas. Moreover, this radical C–H arylation strategy provides an avenue for the preparation of a range of multi-substituted piperidine derivatives with excellent diastereoselectivity. Mechanistic studies and DFT calculations demonstrate the presence of a radical-polar crossover cascade and also reveal the rationale for the generation of exceptional site-selectivity. The pursue of enantioseletive C–H arylation using this strategy in our laboratory is on the way.

Methods

General procedure of photoredox-mediated HAT catalysis for α-amino C(sp3)–H arylation of ureas via radical aryl migration

A flame-dried Schlenk-tube equipped with a magnetic stir bar was charged with 1 (0.2 mmol, 41.2 mg, 1.0 equiv), 4-DPAIPN (0.004 mmol, 3.2 mg, 0.02 equiv), and Na2CO3 (0.5 mmol, 53.0 mg, 2.5 equiv). The tube was then evacuated and backfilled with N2 for 3 times. Afterward, iPr3SiSH (0.06 mmol, 13.0 μL, 0.3 equiv) and dry MeCN (0.1 M, 2.0 mL) were added in sequence by syringe. The tightly sealed tube was then irradiated by a blue LED light cylinder (40 W) and kept at room temperature by two fans. After 24 hours, the reaction was quenched by H2O and the mixture was then transferred into a 100 mL separating funnel. After extraction with EA (15 mL each) for three times, the organic layers were combined and dried over Na2SO4. After filtration, the filtrate was concentrated under reduced pressure to give the crude product. Further purification by flash chromatography on silica gel gave the desired C–H arylation product 2 as a yellow oil (40.0 mg, 98% yield).