Palladium-catalysed selective oxidative amination of olefins with Lewis basic amines

Amines are prominent in natural products, pharmaceutical agents and agrochemicals. Moreover, they are synthetically valuable building blocks for the construction of complex organic molecules and functional materials. However, amines, especially aliphatic and aromatic amines with free N–H bonds, tend to coordinate with transition metals and deactivate the catalyst, posing a tremendous challenge to applying Lewis basic amines in the amination of olefins. Here we present an example of oxidative amination of simple olefins with various Lewis basic amines. The combination of a palladium catalyst, 2,6-dimethyl-1,4-benzoquinone and a phosphorous ligand leads to the efficient synthesis of alkyl and aryl allylamines. A series of allylamines were obtained with good yields and excellent regio- and stereoselectivities. Intramolecular amination to synthesize tetrahydropyrrole and piperidine derivatives was also realized. Mechanistic investigations reveal that the reaction undergoes allylic C(sp3)–H activation and subsequent functionalization. Given the importance of amine compounds, methods for their synthesis continue to be in high demand. Now, a palladium-catalysed strategy has been developed for the selective oxidative amination of unactivated olefins with Lewis basic amines, via C(sp3)–H activation, forming architecturally versatile and functionally diverse allylamines in a single step.

T he physiological properties of amines and their antibacterial and anticancer activities have rendered them highly efficient pharmaceutical agents 1,2 . Approximately 80% of the top 200 small-molecule pharmaceuticals by retail sales in 2018 contain an amine moiety. The allylamine derivatives represent an outstanding regimen of prescription medicines 3 , including among others afatinib, pentazocine and naftifine (Fig. 1a). Naftifine is one of the most widely used commercial antifungal drugs and exhibits biological activities against extensive pathogenic fungi both in vivo and in vitro 4 . Remarkably, the functional tertiary allylamine group appears to be responsible for naftifine's antifungal activity. In traditional approaches to the synthesis of linear allylic amines, allylic alcohol has been regarded as a leaving group prior to being substituted by the nitrogen group, and its presence was essential 5,6 . Current strategies also require a multistep approach and an activated reactant before the construction of a C-N bond 7,8 .
Palladium-catalysed oxidative amination of olefins is an atom-economic pathway to construct ubiquitous C-N bonds [9][10][11][12][13][14] (Fig. 1b). Despite the simplicity and high efficiency of this reaction, some factors limit the development of a comprehensive strategy for the oxidative amination of simple olefins and Lewis basic amines. Aliphatic and aromatic amines with free N-H often coordinate to palladium salts more strongly than olefins, leading to the formation of a bis(amine)-palladium complex 15,16 , thus remarkably decreasing the electrophilicity of palladium and the reactivity of the catalyst (Fig. 1c, upper). Previously, several protocols have been successfully utilized to avoid inactivation of the palladium catalyst. The primary approach is the employment of non-basic nitrogen nucleophiles to reduce the electron density of amines, which can undergo efficient oxidative coupling with alkenes 12,[17][18][19][20] . Notably, the coordination of amine and palladium salts can be avoided by increasing the steric hindrance around the N-H bond 21,22 . In addition, our group discovered that the addition of halide ion salts could promote the dissociation of the nitrogen-palladium complex to reinstate the reactivity of the catalyst 23 . Moreover, we recently reported an application of electron-rich olefins with aromatic amines for the synthesis of α-amino acid esters, which reveals that reinforcing the electron density of alkenes could also be workable 24 . Nevertheless, the electrostatic repulsion between the olefin π-system and the nitrogen lone pair and the Markovnikov and anti-Markovnikov selectivity presents a major challenge to the amination of olefins. Although significant research has been conducted in recent years, the feasible palladium-catalysed oxidative amination of olefins with aliphatic amines has not yet been realized. We note that during the submission of this paper, a related study using amine-BF 3 complexes as nitrogen nucleophiles was reported by White and co-workers 25 .
In our efforts to achieve direct oxidative amination of simple olefins with aliphatic and aromatic amines, we proposed an unconventional catalytic system to avoid the deactivation of the palladium catalyst poisoned by Lewis basic amines or polymerization to palladium black (Fig. 1c, lower). The utilization of bidentate phosphine ligands could prevent the active palladium complex from being poisoned by amine coordination, accelerate classic C-H bond activation and improve the electrophilicity of π-allylpalladium complexes 26 . The oxidative quinone oxidants acted as an electron transfer reagent for Pd(0)/Pd(II) reoxidation 27,28 . Meanwhile, the coordination between the amine and metal catalyst could be hindered through intermolecular hydrogen bonding between the quinone and amines. The formed palladium-ligand complex could be stabilized by the π-π stacking effect between quinone and ligand. Therefore, this unique method could preserve the high catalytic activity of the palladium catalyst by stabilizing the Pd(II)/Pd(0)/ Pd(II) catalytic cycle. Here we report palladium-catalysed selective oxidative amination of readily available olefins with dialkyl and aromatic amines via C(sp 3 )-H activation to build architecturally versatile and functionally diverse allylamines in a single step. This protocol exhibits broad functional-group tolerance and enriches the synthetic route of amine drugs, such as naftifine and flunarizine. Under modified reaction conditions, the intramolecular amination reactions of N-(hex-5-en-1-yl)anilines and N-(hept-6-en-1-yl)anilines were also realized, delivering the expected tetrahydropyrrole and piperidine derivatives with satisfactory yields and excellent regioselectivities.

results and discussion
Optimization of the reaction conditions. Our initial exploration focused on accomplishing the selective oxidative amination of allylbenzene with dipropylamine, as seen in Table 1 and further elaborated in Supplementary Tables 1-4. When employing Pd(OAc) 2 as a catalyst, benzoquinone (BQ) as an oxidant, PPh 3 as a ligand, and toluene as a solvent, the expected amination product was afforded in 56% yield (Table 1, entry 1). We were surprised to find that no transformation was observed in the presence of PdCl 2 , Pd(PhCN) 2 Cl 2 or other chloride-containing palladium catalysts (Table 1, entries 2  and 3). The blank control experiments indicate that the combination of phosphine ligands and oxidative quinone oxidants dominated this unconventional amination process (Table 1, entries 5 and 6). Considering the indispensable effect of phosphine ligands and quinone oxidants, the optimal association of ligands and oxidants was inspected cautiously, as shown in entries 7-13. Several quinone oxidants, including BQ and 2,6-dimethyl-1,4-benzoquinone (2,6-DMBQ), were combined with different monodentate and bidentate phosphine ligands (Supplementary Table 3). The results reveal that the combination of 2,6-DMBQ and 1,2-bis(diphenylphosphino) ethane (DPPE) achieved the effective construction of allylamines in 90% yield (Table 1, entry 12). We suspected that in air, the phosphorus ligands would be slowly oxidized, and some of the olefins would be isomerized into internal olefins. Examination of the atmosphere and the loading of catalyst and ligand suggests that the combination of 5 mol% Pd(OAc) 2 and 10 mol% 1,2-bis(diphenylphosphino)propane (DPPP) was the optimal choice under a nitrogen atmosphere, allowing the exclusive formation of 3 in 94% yield, and reducing the olefin usage to 1.2 equiv. (Table 1, entry 14).
In addition to the great applicability of aliphatic amine substrates, a variety of simple olefins were compatible with this process ( Table  2). Various substituted allylbenzenes were tolerated, delivering high yields and sole stereoselectivities (39-53). Substituents at the ortho, meta or para positions on the allylbenzenes could be accommodated. The effect of steric hindrance on the reaction was almost negligible. It was noteworthy that substituting the phenyl ring with electron-withdrawing groups generally led to higher yields and efficiency compared with electron-rich substrates. For instance, a 95% yield of the product could be obtained for p-CF 3 -substituted allylbenzene, whereas allylbenzene (44) with a p-OCH 3 substituent was formed only in 84% yield. Moreover, heterocyclic olefins were also suitable for the reaction, and the desired allylic amine (54) was formed in 83% yield. By slightly adjusting the reaction conditions, the feedstock alkenes, such as 1-octene, and several functionalized alkenes could participate satisfactorily in this oxidative amination, generating the corresponding products (55-60).

Synthetic applications.
The widespread presence of alkylamines in small-molecule drugs and natural products suggests that the efficient synthesis and late-stage functionalization of such compounds could indeed demonstrate the utility of this oxidative amination. Under standard conditions, a series of pharmaceutical agents and bioactive molecules, such as cytisine, amoxapine (CCDC 2038362), desloratadine, dehydroabietylamine, sitagliptin, atomoxetine, desipramine, estrone and (+)-allylated-δ-tocopherol, effectively underwent oxidative amination to afford the corresponding allylamine derivatives (61-70) in satisfactory yields (35-79%) with excellent regio-and stereoselectivities (Table 2). Moreover, the synthesis of bioactive molecules was also feasible. Naftifine (71) 29 could be obtained directly from the amination of allylbenzene and 1-methyl-aminomethyl naphthalene in a single step in 85% yield (Fig. 2a). Cinnarizine (72) and flunarizine (73) were accessed in a two-step sequence proceeding in 92% and 86% yields, respectively (Fig. 2b). In addition, gram-scale (5 mmol) experiments were successfully conducted to deliver naftifine and cinnarizine with 70% and 72% isolated yields, respectively. Product 74, which was formulated from benzyl-protected tryptamine and allylbenzene via this oxidative amination reaction, was an adenylyl cyclase type I (AC1) inhibitor 30 (AC1 belongs to the family of adenylyl cyclases, which are associated with neuropathic and inflammatory pain) (Fig. 2c). Additionally, an abamineSG derivative 31 75, an effective inhibitor of the biosynthesis of abscisic acid, was assembled in a 79% yield in two steps (Fig. 2d).
Scope of aromatic amines and olefins. We next investigated the extension of the palladium-catalysed oxidative amination of olefins with aromatic amines (Supplementary Table 9). The basicity and affinity of aromatic amines are comparatively weaker than those of aliphatic amines, leading to a decrease in their coordination to the transition metal. The employment of monodentate phosphorus ligand could maintain the activity of the catalyst, and O 2 could be used as the terminal oxidant to complete the catalytic cycle.

Mechanistic investigations.
To gain more insight into the palladium-catalysed oxidative amination of olefins, inter-and intramolecular competitive kinetic isotopic effect (KIE) studies were explored (Fig. 3a). The results exhibited high KIE (k H /k D ) values of 4.6 and 2.8 for inter-and intramolecular competition, respectively, implying that allylic C-H cleavage contributes to the rate-determining step. A preliminary Hammett study was performed to investigate the electronic effect on substituents appended to the olefins (Fig. 3b). A ρ value of 0.5355 was obtained for a series of substituted allylbenzenes, indicating that electron-withdrawing groups produced an increase in the amination reaction rate. This is consistent with the mechanism of C-H activation and the finding that electron-poor olefins could promote ligand exchange with the palladium complex 32 . Alternatively, the electron-withdrawing substituents would increase the activity of the allylpalladium intermediate, accelerating subsequent functionalization.
Kinetic analysis experiments for allylbenzene were conducted under optimal reaction conditions. The rate data indicated a first-order dependence on the concentration of the palladium catalyst, DPPP and allylbenzene (Fig. 3c), which reveals that the formation of a π-allylpalladium complex through the C-H activation of olefins should be the rate-determining step. However, dipropylamine, which could easily coordinate with the palladium catalyst and suppress the formation of the π-allylpalladium complex, indicated a zero order in this reaction. This result suggests that the coordination of DPPP with the palladium catalyst was much stronger than that of dipropylamine, and the toxic effect of the amine could be ignored. To clarify the role of 2,6-DMBQ and the ligand in the amination process,   the π-allylpalladium dimer was synthesized and treated with dipropylamine. Target product 3 was obtained in 99% yield in the presence of 2.0 equiv. DPPP. Regardless of the presence/absence of 2,6-DMBQ, the yield of 3 was approximately 35% in the absence of ligand, which indicates that 2,6-DMBQ was not involved in the nucleophilic attack step as a π-acceptor ligand. However, the catalytic amination reaction could not occur without the addition of 2,6-DMBQ, and the amount could affect the yield of 3, implying that 2,6-DMBQ was critical in the C-H activation step (Fig. 3d). Compared with 2,6-DMBQ, the DPPP ligand played a key role in C(sp 3 )-H activation and the nucleophilic attack step. The addition of 20 mol% or 2.0 equiv. of DPPP to the reaction markedly increased the yield of 3 from 36% to 50% or 99%, respectively, suggesting that the ligand increased the electrophilicity of π-allylpalladium complexes and accelerated the fast nucleophilic attack by the amines. No product was detected in the catalytic amination reaction without DPPP (Fig. 3e).
Enlightened by the experimental observations, density functional theory (DFT) studies were performed to further understand the amination reactions. The intriguing suppression of the toxic effect of the amine on the palladium system can be well explained by the coordination competition between DPPP and dipropylamine to the Pd(II) centre, as well as arylamine in comparison. As shown in Fig. 4a, the arylamine-coordinated intermediate IM ′′ 1 is less stable than the DPPP-based complex IM 1 by 18.7 kcal mol −1 , indicating that the phosphine ligand can prevent the poisoning of the metal centre by arylamine substrates. In comparison, the DPPP-based complex IM 1 is more stable than the dipropylamine coordinated complex IM ′ 1 by only 2.3 kcal mol −1 , indicating that the difference in coordination ability of DPPP and dipropylamine is subtle. When the amount of dipropylamine is greater than that of DPPP at the initial stage of the reaction, which might still lead to a toxic effect of dipropylamine in the system, this is consistent with the experimental phenomenon that the product could not be observed when 2,6-DMBQ was not added. Interestingly, with 2,6-DMBQ as the additive (Fig. 4b), the interaction between 2,6-DMBQ and DPPP has an inconsequential influence on the coordination of DPPP with Pd(II), as shown in IMBQ 1 (−1.7 kcal mol −1 ). However, 2,6-DMBQ influences the coordination of dipropylamine to the Pd(II) centre to some extent, leading to a less stable aliphatic-amine-coordinated intermediate IMBQ ′ 1 , with a calculated free energy of 6.1 kcal mol −1 compared to IM ′ 1 . This effect can be attributed to the electron-deficient characteristic of the 2,6-DMBQ, which impedes the dative property of aliphatic amine and perturbs its intramolecular hydrogen bonding with the ligated OAc − . Therefore, 2,6-DMBQ can serve not only as an oxidant but also as an electron-deficient additive that prevents aliphatic amine poisoning of the metal centre. This can perfectly explain the observed experimental phenomenon that the addition of 2,6-DMBQ initiated the reaction of dipropylamine with allylbenzene. DFT results suggest that 2,6-DMBQ could also decrease the dative interaction between aryl amine and the metal centre to the same extent as the case of dipropylamine (IMBQ ′′ 1 ). Because arylamine is not a strongly poisoning substrate due to its weaker affinity compared with dipropylamine, the addition of 2,6-DMBQ has no effect on the initiation of the reaction of arylamine. The second 2,6-DMBQ molecule was also considered in the coordination of DPPP and amine with a metal centre. However, the calculation results show that compared to the first 2,6-DMBQ molecule, the second 2,6-DMBQ molecule could not form π-π interactions with the DPPP and palladium centres due to steric hindrance. Moreover, it could prevent the aliphatic amine poisoning the palladium centre to a greater extent but has no obvious effect on the coordination of arylamine (Supplementary Figs. 14 and 15). The mechanism of the oxidative amination employing allylbenzene and dipropylamine as the representative reactants was investigated by DFT calculations, as displayed in the free energy profile shown in Fig. 4c. First, the DPPP ligand coordinates to the palladium centre of Pd(OAc) 2 , leading to thermodynamically stable intermediate IM 1 (−143.8 kcal mol −1 ) as the active species for the C-H activation of allylbenzene. Although the ligand exchange between one OAc − ligand and allylbenzene is endergonic by 10.9 kcal mol −1 , the resulting intermediate IM 2 (−132.9 kcal mol −1 ) is highly active for C-H activation. The free energy of the C-H activation transition state TS 1 is only 10.3 kcal mol −1 higher than that of IM 2 . In TS 1 , OAc − abstracts the allylic hydrogen atom from the C 3 atom of allylbenzene, resulting in an allylic anion that is nucleophilically bound to the Pd(II) centre with the C 1 atom. The formation of C 1 −Pd (2.268 Å) and C 2 =C 3 (1.429 Å) bonds can be seen in TS 1 , in which the square planar structure of the DPPP bidentate Pd(II) complex is well maintained. The overall free energy barrier of this stage is calculated to be 21.2 kcal mol −1 (IM 1 → TS 1 ). Another scenario in which the ligated OAc − ligand directly abstracts the allylic hydrogen atom is less plausible due to the relatively higher free energy of the corresponding    Fig. 4 | DFt study on the coordination competition between substrates and ligand and the operative catalytic cycle. a, The coordination competition between ligand and substrates. The coordination ability of DPPP is stronger than that of aromatic amines but similar to that of dipropylamine, which indicates that the addition of phosphine ligands alone can effectively prevent the catalyst from being poisoned by aromatic amines, but the poisoning of alkylamines is inevitable. b, The role of 2,6-DMbQ in the coordination competition between ligand and substrates. 2,6-DMbQ can weaken the coordination of dipropylamine and arylamine but has an inconsequential influence on the coordination of DPPP. c, The Gibbs free energy profiles of the oxidative amination for dipropylamine with schematic structures. The catalytic cycle involves the coordination between the catalyst and alkene (iM 2 ), allylic C(sp 3 )-H activation (ts 1 ), nucleophilic attack of dipropylamine (ts 2 ) and product liberation. The rate-determining step is C-H activation with a free energy barrier of 21.2 kcal mol −1 , which is consistent with experimental observations. bond distances are in Å, and key geometric parameters in transition states are highlighted in blue for visual clarity. transition state TS ′ 1 (−15.2 kcal mol −1 ). This may be attributed to the decreased basicity of the coordinated OAc − ligand compared with the dissociated OAc − anion. We also considered the possible participation of 2,6-DMBQ in the first step and excluded this possibility by calculations ( Supplementary Figs. 16 and 17). The formation of the η 3 allylic complex IM 4 has to dissociate one arm of the bidentate DPPP from IM 3. Intermediate IM 4 (−132.6 kcal mol −1 ) initiates C-N bond formation via a nucleophilic attack through a proton-shuttle-assisted transition state TS 2 , in which the nitrogen atom of dipropylamine nucleophilically attacks the allylic C 1 atom (N-C 1 = 2.099 Å), breaking the Pd-C 1 bond (2.630 Å). In contrast, the nucleophilic attack of the allylic C 3 atom by the amine is unfavourable due to steric effects, as reflected by the relatively high free energy of TS ′ 2 (−117.7 kcal mol −1 ). The free energy barrier of C-N bond formation is calculated to be 20.8 kcal mol −1 (IM 3 → TS 2 ). After the formation of the product, Pd(II) will be regenerated from Pd(0) by the oxidant. The reductive elimination mechanism is suggested to be less plausible for C-N bond formation (IM5′ and IM5″ in Fig. 4c). The amino allylic intermediates IM5′ and IM5″ are calculated to be higher than the nucleophilic attack transition state TS 2 by 18.8 and 24.0 kcal mol −1 in free energy, respectively, indicating that the coordination of the bulky dipropylamine ligand is suppressed by the crowded catalytic centre. This nucleophilic attack mechanism interestingly reveals the potential of our developed catalytic system for preventing poisoning of the catalyst centre due to amino coordination, which is key to the successful oxidative amination of olefins with challenging aliphatic amines.

Conclusion
This unconventional catalytic system enables the coupling of a series of Lewis basic amines and olefins to afford corresponding alkyl and aryl allylamines. We anticipate that this approach will address a long-standing unsolved problem posed by oxidative amination chemistry and will greatly facilitate the postmodification of existing chemical drugs and the discovery of new small-molecule pharmaceuticals.

Online content
Any methods, additional references, Nature Research reporting summaries, source data, extended data, supplementary information, acknowledgements, peer review information; details of author contributions and competing interests; and statements of data and code availability are available at https://doi.org/10.1038/ s41557-022-01023-x.