Introduction

Chiral N-oxide compounds, characterized by N-stereogenic centers, represent a distinctive category prevalent in both natural products and synthetic bioactive species1. Noteworthy examples include acetylcholinesterase inhibitors like 4β-Hydroxyisodo-lichantoside2 and Ganstigmine CHF-28193, adenosine uptake-inhibitors exemplified by Diltiazem derivatives4, antitumor drugs such as Sophoridine derivatives5, NF-kB inhibitor like Tylophoridicine F6, Parkinson’s disease drug, Selegiline N-oxide7,8 which is a derivative of MAO-B inhibitor (Fig. 1a). Beyond their biological significance, these N-oxides exhibit excellent ligand properties9,10,11,12,13, establishing them as pivotal scaffolds in organometallic chemistry (Fig. 1b). Recognizing the multifaceted roles of N-oxides in diverse applications spanning chemistry, biology, and pharmacy, researchers have increasingly directed their attention toward unraveling their inherent complexities and unlocking their potential for advancements in these disciplines14,15,16,17. The direct oxidation of tertiary amines stands out as the most straightforward route to obtain N-oxides18,19,20,21,22. However, a significant challenge arises from the rapid flipping of the lone pair on the nitrogen atom, accompanied by a low energy barrier, leading to the racemization of the oxidation outcomes. Achieving optimal enantioselectivity in the dynamic kinetic resolution (DKR) of tertiary amines is contingent upon the unequal reactivity of the isomers with oxidants23,24,25. Researchers have undertaken various endeavors to address this issue, with some exploring bioinspired catalysts. These catalysts draw inspiration from enzyme cores, simplifying unnecessary or impractical scaffolds and thereby emulating enzyme functionalities to facilitate biochemical reactions with enhanced precision. Colonna and colleagues initially discovered the catalytic potential of bovine serum albumin (BSA) in mediating asymmetric sulfoxidation26. Subsequently, the Gaggero group advanced the BSA-catalyzed oxidation system, extending its application to N-oxidation27. However, the efficacy of this activated protein was confined to a relatively narrow temperature range, predominantly around ambient conditions, posing limitations on the enantioselectivity achievable in N-oxidation. Flavin-containing monooxygenases (FMOs) have demonstrated their ability to catalyze the in vivo metabolic processes of amines28,29. Ottolina and co-workers isolated cyclohexanone monooxygenase (CYMO) and harnessed co-factor NADPH and oxygen to achieve enantioselective amine oxidation28. Despite modest results, the potential of the FMOs family to induce chirality became evident. In response, Tang’s group introduced a biomimetic catalyst named chiral alloxan, elevating the enantioselectivity of N-oxidation to an impressive 93%30. Designed based on the functional cores of FMOs, chiral alloxan circumvented the steric hindrance posed by the protein cleft, relying instead on the chiral auxiliary of the catalyst and bulky groups (such as adamantly) to achieve heightened enantioselectivity (Fig. 1c). An alternative approach involves the utilization of metal and peroxo-groups for oxygen transfer to the target. In 2016, Yamamoto and co-workers devised a bimetallic-centered catalyst with a chiral ligand scaffold31. In their proposed mechanism, one titanium center was oxidized by tert-butyl hydrogen peroxide (TBHP), forming a peroxo-group to facilitate the oxygen transfer to the amine. Simultaneously, another titanium center played a crucial role in capturing the hydroxyl group on the amine substrate, acting as a vital directing group (Fig. 1d). Collectively, these examples emphasize the challenging nature of expanding the amine scope and achieving enantioselectivity, requiring either potent inducing effects or substantial steric hindrance to counteract the swift racemization of the lone pair on tertiary amines.

Fig. 1: Background and catalyst development.
Fig. 1: Background and catalyst development.
Full size image

a Chiral N-oxides as medicinal molecules. b Chiral N-oxides as ligands in organocatalysis. c BSA analogs and Flavin-containing monooxygenases and mimics applied in asymmetric N-oxidation. d Organometallics and heteropolymetalates applied in asymmetric N-oxidation. X-ray crystallographic structure of [BG]2+ [(μ-SO4)2Mo2O2(μ-O2)2(O2)2]2- and its utilizing in N-oxidation of alky substituted amines and aryl substituted amines.

Ion-pair catalysts comprising a chiral cationic component and an inorganic anionic salt, have demonstrated remarkable efficacy in catalyzing various reactions with high stereoselectivity32,33. Our previous work showcased the proficiency of bisguanidinium permanganate in mediating enantioselective dihydroxylation and oxohydroxylation34 of α,β-unsaturated esters. Later, we accomplished a remarkably proficient and enantioselective oxidative sulfoxidation of heterocyclic and alkyl aryl sulfides, catalyzed by bisguanidinium, using aqueous hydrogen peroxide. The active anionic species were identified as diphosphatobisperoxotungstate35 and dinuclear oxodiperoxomolybdosulfate36, respectively. Moreover, bisguanidinium tetraperoxyditungstate, [BG]2+ [W2O2(µ-O)(O2)4]2- was isolated and determined as an active catalyst in asymmetric epoxidation of allylic and homoallylic amides37. Motivated by these achievements, we aim to broaden the scope of ion-pair catalysis involving anionic metal oxides to encompass the asymmetric oxidation of challenging tertiary amine substrates (Fig. 1d). In this report, we unveil a highly enantioselective oxidation of cyclic and acyclic tertiary amines, employing ion-pair catalysts BG5 whose structure has been elucidated through X-ray crystallography.

Results

In our previous research, we detailed a highly enantioselective sulfoxidation catalyzed by bisguanidinium dinuclear oxodiperoxomolybdosulfate. Seeking to broaden the application of bisguanidinium-catalyzed oxidations, we delved into the challenging realm of N-oxidation of tertiary amines. This specific transformation has eluded efficient strategies for an extended period. In this work, we explored both cyclic and acyclic amines, and the results exhibited remarkable enantioselectivity in the majority of the cases eliminating the need for bulky substituents. Initially, we systematically assessed various bisguanidinium-catalyzed oxidative systems to gauge their suitability for N-oxidation. Unfortunately, potent metalated oxidants, such as permanganate and tungstate, led to undesired amide formation by oxidizing the methylene group on the phenyl moiety. Notably, molybdate emerged as the singular candidate with the capacity to oxidize the amine functionality in high chemoselectivity.

Consequently, we subjected 2-(benzyl(methyl)amino)ethan-1-ol 1a, to the reported standard sulfoxidation conditions, leading to the isolation of N-benzyl-2-hydroxy-N-methylethan-1-amine oxide 2a with low enantioselectivity (Table 1, entry 1). Given the solubility of sodium molybdate in the aqueous phase and the anticipated fast ion exchange between the organic and aqueous phases, we hypothesized that the formation of heteropoly complexes was hindered. Transitioning to silver as the counter ion for molybdate resulted in a marginal increase in enantioselectivity (entry 2). Further, substituting sodium bisulfate with potassium bisulfate, notably elevated the yield to 99% (entry 3). To enhance the reaction conditions further, our focus shifted to catalyst modification. Reflecting on our prior models, we recognized that the substrates in the earlier studies were slightly bulkier than those involved in N-oxidation. For instance, sulfoxidations featured sulfides with two aromatic systems and epoxidations employed tosyl as a directing group on allylic amines. Consequently, we explored the strategy of augmenting the catalyst’s bulkiness to fortify the control exerted by the chiral cation over the substrates. The replacement of the tert-butyl group on the side chain of BG1 with a trimethylsilyl (TMS) group proved successful, with the bulkier BG2 exhibiting enantioselectivity (entry 4). Evidently, reducing the loading of KHSO4 and diluting the concentration of hydrogen peroxide reduced the reaction rate, allowing for improved chiral induction by the catalyst (entry 5). Encouraged by the success with TMS substitutions, we investigated triethylsilyl (TES) group as a substituent to confine the substrate within a narrower cage. Unfortunately, both yield and enantiomeric excess (ee) declined sharply (entry 6). The setback with TES encouraged us to continue adjusting the size of the side chain in pursuit of an optimal balance. To refine the catalyst further, we retained two methyl groups on the silyl moiety, and the introduction of dimethylethylsilyl (DMES) group in BG4 yielded almost identical enantioselectivity as achieved with BG2, albeit with a slight reduction in yield (entry 7). The culmination of our modifications involved strategically placing both the TMS and DMES groups within the orthometric space of the four side chains on BG. Remarkably, this strategic incorporation in BG5 produced 98% yield and 87% ee (entries 8, 9).

Table 1 Optimization study of asymmetric N-oxidation
Full size table

Armed with the optimized reaction conditions, we proceeded to explore the substrate scope of alkylamines for asymmetric N-oxidation (Fig. 2). Within the N-benzyl group, both electron-withdrawing substituents (e.g., trifluoromethyl 2b, halogenated 2c-2e, trifluoromethoxy 2f, nitryl 2g, ester group 2 h, acyl 2i) and electron-donating substituents (e.g., phenyl 2j, alkyl 2k-2m, methoxy 2n) were well tolerated at the para position of the aromatic ring. The resulting chiral N-oxides were obtained in high yields (80–98%) with high enantiomeric ratios (er) (90:10 to 96.5:3.5). Noticeably, for substrates bearing potent electron-withdrawing substituents (2b, 2c, 2f, 2g, 2r), lesser amounts of BG5 and Ag2MoO4 sufficed to achieve higher er. We speculated a competition between protonation and oxidation for amine substrates under acidic conditions. Strong electron-withdrawing groups at the para-position inhibited protonation, allowing for oxidation through a gentle and slow process, enabling specific recognition with reduced catalyst and oxometallate loading. Extending our exploration to meta and ortho positions, diverse substituents were well tolerated, yielding N-oxides 2o-2w with excellent chemoselectivity (85–98% yields) and enantioselectivity (89.5:10.5 to 95.5:4.5 er). Compatibility was also observed with other aromatic systems 2x-2z. Due to steric effects, 1-naphthyl 2 y was expected to pose a greater hindrance to the chiral discrimination compared to 2-naphthyl 2x. To investigate the role of β-hydroxyl on the substrate as a directing group, we introduced hydroxyl on the phenyl 1aa and γ-hydroxyl 1ab. While 2aa was obtained with 92:8 er, indicating unaffected chiral recognition, the introduction of hydroxyl at the γ-position in 2ab resulted in only racemized N-oxide with low yield, suggesting sensitivity to the distance between N-center and hydroxyl. However, without the hydroxy group, the yield and ee dropped largely when hydroxy was replaced by methyl (2ae). It meant hydroxy might be a good directing group. Further substrate modifications revealed that the asymmetric oxidation was impeded when the methyl substituent on the amine was replaced by ethyl 2ac (in comparison to 2a). Unfortunately, the amine could not be oxidized when increasing the length of the carbon chain between phenyl and nitrogen atom 2ad.

Fig. 2: Asymmetric N-oxidation of acyclic tertiary amines.
Fig. 2: Asymmetric N-oxidation of acyclic tertiary amines.
Full size image

aReaction conditions: 1 (0.05 mmol), BG5 (8 mol%), Ag2MoO4 (4 mol%), KHSO4(0.3 equiv.) and 25% aqueous H2O2 (1.2 equiv.) in 1.0 mL iPr2O at 0 °C for 36-48 h, The isolated yields were given, The er values were determined by HPLC analysis on a chiral stationary phase. bBG5 (2 mol%) and Ag2MoO4 (2 mol%) were used.

To the best of our knowledge, the direct asymmetric oxidation of arylamines has been scarcely reported up to date. Unlike alkylamines, the lone pair electrons of nitrogen in arylamines pose a greater challenge for oxidant capture due to the p-π conjugation effect. In our pursuit to further broaden the substrate scope of this method, our focus shifted towards studying the N-oxidation of arylamines (Fig. 3). As a representative model substrate, 3a was chosen, considering the importance of tetrahydroisoquinoline skeleton in pharmacy. Regrettably, 3a exhibited limited reactivity under the optimized conditions established for alkylamines. Drawing from prior knowledge, we hypothesized that the cyclic amine’s structural bulkiness might impede its entry into the cavity of BG5. After screening various smaller substituents on the catalyst’s side chain, we found that BG1 yielded more favorable results. Later, a systematic optimization of reaction conditions for 3 was conducted, and 3a served as the model substrate. The resulting chiral N-oxide 4a was obtained in a 70% yield with 96:4 er (Supplementary Tables 9 and 10). Encouragingly, other N-naphthyl tetrahydroisoquinolines exhibited satisfactory reactivity under the new condition, furnishing 4b–4g in moderate yields with excellent enantioselectivity. Strikingly, steric hindrance around the N-center significantly contributed to chiral induction. Phenanthryl (4h) was furnished in 99% ee, while 2-naphathyl (4i) and phenyl (4j) showed a slight decrease in enantioselectivity, aligning with our expectations.

Fig. 3: Asymmetric N-oxidation of cyclic tertiary amines.
Fig. 3: Asymmetric N-oxidation of cyclic tertiary amines.
Full size image

aReaction conditions: 3 (0.05 mmol), BG1 (8 mol%), Ag2MoO4 (8 mol%), KHSO4 (1.0 equiv.) and 35% aqueous H2O2 (4.0 equiv.) in 0.5 mL iPr2O at 0 °C for 48–72 h, The isolated yields were given, The er values were determined by HPLC analysis on a chiral stationary phase. bBG1 (4 mol%), Ag2MoO4 (2 mol%) and H2O2 (2.0 equiv.) were used.

To demonstrate the utility of this asymmetric oxidation method, we initially conducted a gram-scale reaction, achieving both high yield and good enantioselectivity. In the presence of the terminal hydroxyl group on the alkyl chain of chiral N-oxides, the facile introduction of certain drug molecules was achieved through a straightforward esterification process (Fig. 4). Notable examples include COX1/2 inhibitors indomethacin and diclofenac, NF-kB inhibitor sulindac, nonsteroidal anti-inflammatory drugs (NSAIDs) felbinac and isoxepac, calcium channel blocker nicardipine, and PPAR agonist fenofibric acid. Each of these drug molecules effectively condensed with the β-hydroxyl N-oxides, resulting in the formation of their corresponding derivatives (5–11). Importantly, during these further transformations, there was no noticeable loss of enantiomeric purity, showcasing the robustness of the chiral N-oxide structure. Furthermore, the stability of the N-oxide structure and its stereo configuration was not only maintained under ambient conditions but also in more challenging reaction environments. Remarkably, the direct synthesis of the antihistamine chiral trimetazidine analog 13 was achieved using the asymmetric oxidation method, highlighting the versatility and applicability of this approach in accessing diverse pharmaceutical analogs.

Fig. 4: Further transformation of N-oxides.
Fig. 4: Further transformation of N-oxides.
Full size image

a Condensation conditions: N-oxide (0.05 mmol, 1.0 equiv.), drug (1.2 equiv.), DMAP (5 mol%), EDCI (1.2 equiv.) in DCM at RT for 4–6 h. For products 6 and 9, racemic drugs were used. The isolated yields were given, The er values were determined by HPLC analysis on a chiral stationary phase. b Reaction conditions: amine (0.05 mmol, 1.0 equiv.), BG5 (8 mol%), Ag2MoO4 (10 mol%), KHSO4(0.3 equiv.), and 25% aqueous H2O2 (1.2 equiv.) in 1.0 mL iPr2O at 0 °C for 48 h, The isolated yield was given, The er value was determined by HPLC analysis on a chiral stationary phase.

Discussion

In pursuit of the actual catalytic oxidation species, BG6 was synthesized by replicating the reaction condition without the amine substrate. In a solution comprising BG5 (0.04 mmol), Ag2MoO4 (0.5 equiv.) and KHSO4 (4.0 equiv.) in iPr2O (1 mL), 25% H2O2 (10.0 equiv.) was added. The resulting solution was stirred for 2 h in the dark, with the color changing from gray to yellow. Following solvent removal under reduced pressure, the residue was reconstituted in DCM. The resultant suspension was filtered, and BG6 was obtained as a pale yellow solid. A single crystal suitable for X-ray diffraction was grown from DMF at room temperature. The X-ray diffraction analysis of BG6 clearly revealed that a symmetrical heteropoly salt [(μ-SO4)Mo2O2(μ-O2)2(O2)2]2- was embedded in the chiral cavity of the bisguanidinium cation, akin to our previous reported BG7 (Fig. 5a, k). Intriguingly, 1a could be directly oxidized by BG6 without additional oxidant H2O2. When BG6 complex was loaded in 0.1 and 0.25 equivalent, 2a was obtained in 8% and 24% yield, respectively (Fig. 5b). This indicated that only one of the side-on peroxo groups on the metalloanion functioned as an oxygen transfer agent between chiral catalyst and the substrate. Compared with the standard in situ conditions, it is noteworthy that with sufficient KHSO4, using either BG6 or BG7 in catalytic amount yielded similar results (Fig. 5b). This strongly supported the notion that an ion-pair of bisguanidinium cation and metalloanion constituted the genuine active catalyst in this asymmetric oxidation. Subsequent kinetic experiments aimed at further mechanistic insights involved observing the variation in the reaction induction period under standard conditions during the initial 6 h. The yield of N-oxides accumulated slowly, while the enantioselectivity increased linearly during the induction period and stabilized after entering the resting phase (Fig. 5d). Conversely when BG6 was employed as the catalyst under altered conditions, the reaction rate exhibited first-order kinetics over time (Fig. 5e, f). No lag phase was observed during the induction period, and the yield increased linearly. Furthermore, the yield was directly proportional to the amount of catalyst loading within a limit of 4 mol%. Surprisingly, the ee value of 2a gradually increased until the middle stage of the reaction in all kinetic experiments, implying the involvement of different mechanisms throughout the reaction process. The final enantioselectivity was affected by initial rate of reaction, increased loading of BG6 resulted in a higher initial rate, leading to higher enantioselectivity. Notably, the rate showed marginal improvement when the loading of BG6 increased from 3 mol% to 4 mol%, with little change in the ee value. In addition, the final ee value exhibited further improvement with the addition of an extra 0.1 equivalent of KHSO4, while the reaction rate displayed a slight increase (Fig. 5g). This suggested that the oxygen on N-oxides could re-oxidize the substrate, leading to the racemization. A series of verification experiments were conducted to substantiate the racemization process (Fig. 5c). In the solution of 2a in iPrOH, no loss of enantiomeric purity was detected after 5 days. However, when 1.0 equivalent of 1a was added to the solution of 2a, the ee value decreased from 87% to 78%, the ee had a smaller decrease when additional HCl was added.

Fig. 5: Mechanistic investigations.
Fig. 5: Mechanistic investigations.
Full size image

a Preparation of active catalysts BG6 and BG7. b Asymmetric oxidation using BG6 and BG7 compared with in situ conditions. c Racemic reaction between 1a and 2a. d Determination the yield and ee of 2a at different times under in situ conditions. e Determination of the yield and ee of 2a at different times using 4 mol% BG6 as catalyst with KHSO4. f Initial reaction rate plot with different BG6 loadings. g Initial reaction rate constant k and final ee of 2a at different BG6 loadings. h Linear relationship between ee values of BG6 and 2a. i Hammett plot for the competitive oxidation of different para-substituted 1. j Chemical shift on 19F NMR of 2c, 1c, and 1c (1.0 equiv.) with 10 mol% BG6 or BG7. k Insight ion-pair structures of BG6 and BG7 compared with in situ conditions.

Besides, ee values of BG6 and 2a exhibited a linear relationship, implying that active catalyst BG6 largely dominated the enantioselectivity of the reaction (Fig. 5h). Hammett studies were also conducted for the oxidation of β-amino alcohols with para-substituted phenyl groups (Fig. 5i). The Hammett plot of (log(kX/kH) versus σ) yielded a straight line with good linear correlation except for the para-fluorine substituent, implying that the oxidation occurs through a single mechanistic pathway. The negative slope (ρ = − 0.5116) implicated the accumulation of positive charge in the rate-determining transition state. For the deviation with the para-fluorine substituent, we postulated that the fluorine substituent was influenced by other factors. To explore this, the 19F NMR chemical shifts of 2c, 1c, and those components with catalyst were assessed. In both the solution of 10 mol% BG6 with 1c and the solution of 10 mol% BG7 with 1c, the major peak was observed at δ -115.6 ppm, identical to the 19F NMR of 1c (Fig. 5j). This constant chemical shift excluded the interaction between fluoro group on 1c and silyl group on the side chain of catalyst. However, in the mixed solution of prepared catalyst and substrate in a 1:10 ratio, the minor peak likely represented 2c. Comparing the spectra of 2c and the solution of 1c with the catalysts, the − 1.2 ppm (19F NMR of 2c versus 19F NMR of 1c and BG6) and the − 1.4 ppm (19F NMR of 2c versus 19F NMR of 1c and BG7) indicated that Si-F interactions did not occur. Instead, an interaction between the fluoro group and oxo group on the functional anion in the reductive state likely took place.

Transition state structures for the oxidation of amine 1a with catalyst BG6 leading to R- or S-configuration amine oxide were calculated and modeled with density functional theory. TSC_S is revealed to be the most stable (Fig. 6), with a barrier of ΔG‡sol = 19.0 kcal/mol relative to BG6 and 1a, followed by TSA_R with a relative barrier of ΔG‡sol = 20.6 kcal/mol. The DFT model predicts (S)-amine oxide would be expected to be a major enantiomer, and the theoretical e.r. is calculated to be 95 by comparing TSA_R and TSC_S, ΔΔG‡sol = 1.6 kcal/mol. These DFT results fit excellently with experimental e.r. of 94 for 2a. Activation strain model analyses of the transition states further provided a rationale for the energetic preference38. From the energies, TSC_S balanced both the geometry distortion, measured by ΔE‡distortion, and interaction energy between catalyst and amine, measured by ΔE‡interaction, resulted in the lowest activation energy, ΔE‡activation (Fig. 6). TSA_R on the other hand, having favorable ΔE‡interaction, had greater strain and distortion in the transition state, ΔE‡distortion, putting it higher than TSC_S based on ΔE‡activation. The ΔE‡activation ranking follows the ΔG‡sol.

Fig. 6: Computational analysis of transition state.
Fig. 6: Computational analysis of transition state.
Full size image

3D representation of transition state structures calculated at ωB97M-V/def2-TZVP/SMD//oniom(M06/6-31 G(d,p)+SDD:pm6) level of theory. The ΔG‡sol values are solution-free energy barriers relative to starting BG6 and 1a in kcal/mol. The activation strain model analyses provided ΔE‡distortion, ΔE‡interaction, and ΔE‡activation in kcal/mol. Hydrogen was omitted for visual clarity. Atoms are color-coded: C (gray), O (red), N (blue), and Mo (green-blue).

Based on the mechanistic insights gained from the preceding experiments, we propose a plausible catalytic cycle (Fig. 7). Initially, an active anionic species A is generated in the aqueous phase through the oxidation by H2O2. Subsequently, A is effectively captured by the chiral cation BG2+ and transferred into the organic phase, leading to the formation of the oxidizing ion-pair B. In the next step, the nitrogen of 1a engages with one of the side-on peroxo groups of B, initiating an intermolecular oxygen transfer process that yields the chiral N-oxide (S)-2a. B reverts to its reductive state to afford C. Simultaneously, the oxygen attached to the nitrogen of (S)-2a exhibits oxidative capacity, oxidizing the amine substrate 1a at a relatively slow rate to produce rac-2a. This process can be hindered by the protonation of 1a or 2a under acidic conditions. Lastly, the anion component of C transitions back to the aqueous phase regenerating species A through oxidation by H2O2, thereby reactivating the catalytic cycle.

Fig. 7
Fig. 7
Full size image

Working model.

In summary, we have successfully developed an ion-pair strategy for achieving high enantioselectivity in the N-oxidation of two distinct types of tertiary amines, 2-(aryl(methyl)amino)ethan-1-ols and N-aryl tetrahydroisoquinolines catalyzed by BG. The newly developed BG5 incorporates the fundamental structure of the bisguanidinium cation. The true catalytic ion-pair species was unequivocally identified through X-ray crystallography, as [(µ-SO4)2Mo2O2(µ-O2)2(O2)2]2-. Employing environmentally benign aqueous hydrogen peroxide as the terminal oxidant in stoichiometric amounts, our catalytic oxidation process demonstrated outstanding yields (98%) and enantioselectivities (99% ee) for alkyl and aryl amines. Mechanistic investigations have yielded detailed insights into the DKR process, providing valuable information for the design of catalysts for a broader range of potential substrates. The mild operation conditions and tolerance towards various functional groups accentuate the versatility of this method. Furthermore, the N-oxides were successfully transformed into a series of medicinal derivatives while maintaining high enantiomeric purity, offering potential applications in the modification of existing drugs within the field of medicinal chemistry.

Methods

General procedure for asymmetric oxidation of alkyl amines

To a tube was added alkyl amine (0.05 mmol, 1.0 equiv.), iPr2O (1.0 mL), BG5 (2−8 mol%), KHSO4 (0.3 equiv.) and Ag2MoO4 (2−10 mol%) at RT, then the mixture was allowed to cool to 0 °C. After 15 min, 25% H2O2 (1.2 equiv.) was added, and the reaction was allowed to stir for 36−48 h. After completion, TEA (100 μL) was added to quench the excess H2O2 and basify the reaction. Then the mixture was directly loaded onto a silica gel column, followed by gradient eluent (EA/MeOH = 9/1 to EA/MeOH =  4/1) to afford the chiral N-oxide.

General procedure for asymmetric oxidation of alkyl amines

To a tube was added aryl amine (0.05 mmol, 1.0 equiv.), iPr2O (0.5 mL), BG1 (4−8 mol%), KHSO4 (1.0 equiv.) and Ag2MoO4 (4−8 mol%) at RT, then the mixture was allowed to cool to 0 °C. After 15 min, 35% H2O2 (1.0−2.0 equiv.) was added. About 24−36 h later, another 35% H2O2 (1.0−2.0 equiv.) was added, and the reaction was allowed to stir for an additional 24−36 h. After completion, TEA (100 μL) was added to quench the excess H2O2 and basify the reaction. Then the mixture was directly loaded onto a silica gel column, followed by gradient eluent (EA/MeOH = 20/1 to EA/MeOH = 4/1) to afford the chiral N-oxide.