1. Introduction

Constructing molecular complexity with high efficiency under mild and environmentally friendly conditions is a central pursuit in synthetic chemistry^[1,2]. Transition-metal-catalyzed oxidative C–H functionalization stands as a direct route for this purpose^[3,4], yet it often depends on stoichiometric oxidants, such as Ag(I), Mn(III), oxone, or organic peroxides, to regenerate the active catalyst (Figure 1a). Replacing these oxidants with molecular oxygen represents a major step toward green synthesis^[5-7]. Taking this further, the use of air is even more attractive due to its low cost, safety, and ready availability^[8-10]. However, catalytic asymmetric reactions using air^[11,12], particularly those proceeding efficiently at room temperature, are exceedingly rare, with only isolated reports for Ir^[13], Ru^[14,15], and Cu^[16,17] systems. Therefore, the development of general catalytic methodologies that operate with air under ambient conditions remains a formidable and standing challenge.

Display Full Size

Figure 1. Design plan: cobalt-catalyzed oxidative enantioselective formal (4+2) cycloaddition.

Transition-metal-catalyzed asymmetric C–H functionalization has emerged as a direct and powerful strategy for constructing chiral molecules^[18-20]. Among the various metals explored, earth-abundant and low-toxicity cobalt has demonstrated unique reactivity in C–H activation^[21-24]. Since Yoshikai’s seminal 2014 report on the cobalt-catalyzed enantioselective C(sp²)–H intramolecular hydroacylation^[25], significant advances have been witnessed in enantioselective cobalt catalysis. They are enabled by diverse chiral ligand systems, including phosphines^[26,27], carboxylic acids^[28-30], cyclopentadienyl ligands^[31,32], and N-heterocyclic carbenes^[33]. Notably, salicyloxazoline (Salox)-based ligands, originally developed by Bolm, have recently emerged as a privileged scaffold in this field (Figure 1b)^[34]. While these systems have achieved substantial progress in enantioselective C(sp²)–H functionalization^[35-40], they typically rely on stoichiometric oxidants and often require elevated temperatures. A critical step toward sustainability was taken by Niu in 2022^[41], who successfully utilized oxygen gas to re-oxidize Co(I) to Co(III) in an asymmetric C–H activation, albeit under heating. More recently, photoirradiation strategies by Shi^[42,43], Ackermann^[44,45], Sundararaju^[46], and Baidya^[47] have demonstrated that light can activate O₂ for this oxidation, enabling transformations like indole dearomatization. More recently, Sundararaju^[48] reported an elegant room-temperature, air-activated strategy for the construction of distal 1,3-C−N diaxes using indole-containing active alkynes. Inspired by these works, we questioned whether the catalytic cycle in asymmetric cobalt-catalyzed C–H activation could be driven by air at room temperature using non-activated alkynes. We hypothesized that coordinating an electron-rich ligand could lower the oxidation potential of the low-valent cobalt intermediate, facilitating its re-oxidation by air without external energy input (Figure 1c). Herein, building on our longstanding interest in photoinduced asymmetric cobalt catalysis^[49-53], we report an efficient, cobalt-catalyzed aerobic oxidative enantioselective annulation of C(sp²)–H bonds using air as the sole oxidant at ambient temperature, which is compatible with a variety of non-deactivated alkynes. Indeed, we found that the catalytic cycle can be augmented by photoinduced oxygen activation for less reactive alkynes and benzamides.

2. Experimental/Methods

Under air, Co(NO₃)₂·6H₂O (0.02 mmol, 5.82 mg), L4 (0.015 mmol, 8.08 mg), and 1,1,1,3,3,3-hexafluoro-2-propanol (HFIP, 2 mL) were added to a 10 mL vial. After stirring at room temperature (rt) for 2 h, corresponding substrate 1 (0.2 mmol), alkyne 2 (0.3 mmol), PC5 (1.64 mg, 0.004 mmol), and PivONa (49.9 mg, 0.4 mmol) were added to the resulting mixture. The mixture was stirred at rt under 5 W blue light-emitting diodes (LEDs) for approximately 48 h. The concentrated reaction residue was purified by flash column chromatography on silica gel with petroleum ether/ethyl acetate (3:1-1.5:1) as eluent to give the corresponding product 3.

3. Results and Discussion

Initially, we investigated the reaction of 4-methyl-N-(quinolin-8-yl)benzamide 1a with diphenylacetylene 2a as the model substrates (Table 1). Preliminary screening revealed that the reaction conducted in 1,1,1,3,3,3-hexafluoro-2-propanol (HFIP) with Co(NO₃)₂·6H₂O as the catalyst provided the product 3aa in 46% yield and 90% ee (Tables S1 and S2 in Supporting Information). Notably, most alcoholic solvents exhibit relatively high yields and enantioselectivity, likely attributable to noncovalent interactions with the cobalt catalyst or salox ligand^[48]. We then identified the optimal photocatalyst (PC) (Table 1, entries 1-5). Upon replacing the photocatalyst with PC5 and irradiating with a 5 W blue LED, the yield increased significantly to 86%, while the enantioselectivity slightly decreased to 89% ee. Moreover, we found that atmospheric oxygen can also act as the terminal oxidant under otherwise identical conditions, resulting in almost consistent outcomes (Table 1, entry 6). Next, a series of chiral salox ligands was tested (Table 1, entries 7-11). A benzyl (L2) or indene skeleton (L3) on the oxazoline moiety dramatically reduced the enantioselectivity. The salox ligand with a methoxy substituent at the 5-position of the phenolic hydroxyl (L4) gave encouraging enantioselectivities of up to 95% ee and yields of 94%. Electron-withdrawing 5-CF₃-substituted (L5) and electron-donating 6-OMe-substituted (L6) salox ligands may reduce the stability of the monoanionic bidentate coordinating pattern. Surprisingly, the same product yield was achieved under control conditions: without PC5 under purple LEDs (Entry 12) and neither light nor PC5 (Entry 13). We then performed a preliminary substrate scope investigation. Substrates bearing different substituents (3ba, 3ga, 3ka in Table 1) showed markedly different reactivities under the three condition sets: (1) no light and no PC5, (2) purple LEDs without PC5, and (3) blue LEDs with PC5. These results clearly demonstrate that the necessity of photochemical conditions for this transformation depends critically on the electronic nature and the position of the substituents in the amide substrates 1.

Table 1. Optimization of reaction conditions.^a

Display Full Size

Entry	PC	Light	Oxidant	L	Yield (%)	ee (%)
1	PC1b	5 W white LEDs	O2	L1	46	90
2	PC2b	5 W white LEDs	O2	L1	51	89
3	PC3	5 W white LEDs	O2	L1	14	85
4	PC4	5 W blue LEDs	O2	L1	30	85
5	PC5	5 W blue LEDs	O2	L1	86	89
6	PC5	5 W blue LEDs	Air	L1	85	89
7	PC5	5 W blue LEDs	Air	L2	95	61
8	PC5	5 W blue LEDs	Air	L3	85	67
9	PC5	5 W blue LEDs	Air	L4	94	95
10	PC5	5 W blue LEDs	Air	L5	87	82
11	PC5	5 W blue LEDs	Air	L6	67	75
12	-	5 W purple LEDs	Air	L4	94	96
13	-	No light	Air	L4	93	96

^a: 1a (0.1 mmol), 2a (0.15 1mmol), PC (2 mol%), Co(NO₃)₂·6H₂O (10 mol%), L (15 mol%), PivONa (2 equiv) in HFIP (1 mL) under O₂ at rt for 40 h. Yields were determined using ¹H NMR spectroscopy with 1,3,5-trimethoxybenzene as an internal standard. All the ee values in this work were determined by chiral HPLC analysis of purified products; ^b: Using PC (10 mol%); ^c: Without light and PC5; ^d: Under 5 W purple LEDs instead of 5 W blue LEDs, without PC5; ^e: Using 5 W blue LEDs and PC5. LED: light-emitting diode; HFIP: 1,1,1,3,3,3-hexafluoro-2-propanol; PC: photocatalyst; NMR: nuclear magnetic resonance.

Subsequently, a series of mechanistic experiments was carried out to rationalize the way photocatalysis works and to understand the explicit role of the light (Figure 2). First, H/D exchange experiments were performed with 1b/D₂O in the absence of both alkynes, PC5 and light, and it was found that the recovered 1b was hardly deuterated at the ortho-positions of the phenyl ring (Figure 2a). This indicates that the cleavage of the C–H bond is irreversible. Second, the parallel kinetic isotope effect (KIE, k_H/k_D = 1.2) was then evaluated, suggesting that C–H bond cleavage may not be involved in the rate-determining step (Figure 2b). Third, employing the cobaltacycle intermediate [Co]-1 in a stoichiometric reaction with alkynes can give the desired cycloadduct 3ga in 93% yield with 99% ee (Figure 2c). Under catalytic conditions, [Co]-1 delivered 3ga in 96% yield with 89% ee (Figure 2d). Both experiments indicate that the chiral ligand chelated Co(III) species [Co]-1 might be involved in the catalytic cycle. The structure of [Co]-1 was confirmed by X-ray diffraction (CCDC: 2489953).

Display Full Size

Figure 2. Mechanistic experiments. [Co], Co(NO₃)₂·H₂O. DMA, 9,10-dimethylanthracene.

To gain a more mechanistic understanding, we performed electron paramagnetic resonance (EPR) studies with 2,2,6,6-tetramethylpiperidin-4-one (4-Oxo-TEMP) as a singlet oxygen (¹O₂) trapping agent (Figure 2e)^[54,55]. When 4-Oxo-TEMP was added under condition B, a triple peak signal was observed after 10 min of 5 W purple LEDs irradiation, with g = 2.0056 and A_N = 1.58 mT (purple trace). Such a spectrum was in good agreement with the simulated EPR spectrum of piperidine N-oxygen radical 4-Oxo-TEMPO (blue trace). This result provides evidence for the existence of singlet oxygen (¹O₂). Next, the three sets of unsubstituted (1b), 4-Br-substituted (1g), and 2-Br-substituted (1k) benzamide derivatives were selected as models to explore their optical and electrical properties in this system. First, addition of 2 equiv of DMA decreased the yield from 51% to 29% under 5 W purple LEDs. In contrast, no significant change was observed under conditions of no light or 5 W blue LEDs and PC5, supporting that ¹O₂ is generated under purple LEDs (Figure 2f). Then, to elucidate the role of light in promoting the oxidation of low-valent cobalt by oxygen, we performed a series of control experiments using the 4-Br-substituted benzamide derivative 1g as the model substrate (Figure 2g). In the absence of both the photocatalyst PC5 and light, the reaction afforded only 25% yield after 15 h. While Mn(OAc)₃·2H₂O (2 equiv) gave a 78% yield, light irradiation increased it to 54%, and the extending the time (40 h) further increased to 95%, supporting our hypothesis that light facilitates aerobic oxidation, albeit less effectively than the stoichiometric oxidant. Next, we measured the UV/Vis spectra for each component of the system and the different mixtures using 2-Br-substituted benzamide derivatives (1k) as substrate (Figure 2h). The solutions containing the mixture of [Co]+L4, [Co]+1k, and [Co]+L4+1k had a distinct absorption in the visible light region. Subsequent fluorescence quenching experiments showed that Co+L4+1k had a more significant quenching effect on the PC5 as compared to Co+1k and Co+L4 (Figure 2i). In addition, detailed cyclic voltammetry tests were conducted to investigate the effects of ligand L4 and benzamide derivatives (Figure 2j). A comparison of the oxidation potentials of [Co] (E_onest([Co]) = 2.09 V vs. SCE in HFIP) and [Co]+L4 (E_onest([Co]+L4) = 1.19 V vs. SCE in HFIP) revealed that the addition of ligand could reduce the oxidation potentials of [Co]. Whereas the introduction of benzamide derivatives further altered the oxidation potential of the in situ generated [Co]+L4+1 complexes. Comparison of [Co]+L4+1b (E_onest([Co]+L4+1b) = 0.90 V vs. SCE in HFIP) and [Co]+L4+1g (E_onest([Co]+L4+1g) = 1.01 V vs. SCE in HFIP) showed that electron-withdrawing substitution of the benzamide derivative leads to an increase in the oxidation potential of the [Co]+L4+1 complex. The further increase in the oxidation potential of [Co]+L4+1k (E_onest([Co]+L4+1k) = 1.21 V vs. SCE in HFIP) indicated that the ortho-substituted substrate weakens the coordination of the benzamide derivative to the cobalt center due to steric hindrance, resulting in a further increase in the oxidation potential of the [Co]+L4+1 complex.

Combined with the three sets of conditions and detailed experimental results required in the reaction, three distinct mechanistic pathways were proposed in the process of oxidative regeneration of cobalt(III) Int-2 (Figure 3). In the conditions without both PC and light, the reaction undergoes path a: the cobalt(II) intermediate Int-1 was oxidized directly by oxygen to the cobalt(III) intermediate Int-2, releasing the superoxide anion O₂^•-. In the conditions of 5 W purple LEDs without PC, the reaction undergoes path b: the ground state oxygen (³O₂) was excited by purple light to afford the singlet oxygen (¹O₂), and then the cobalt(II) intermediate Int-1 was oxidized by the singlet oxygen (¹O₂) to the cobalt(III) intermediate Int-2, accompanied by the release of superoxide anion O₂^•−. In the conditions of 5 W blue LEDs and PC, the reaction undergoes path c: the ground state PC reaches an excited state (PC*) under irradiation with blue LED, which is subsequently quenched by cobalt (II) Int-1 in a reductive process via a SET pathway and forms PC^•-. The oxidation of PC^•- by oxygen delivers the ground state PC and the superoxide anion O₂^•-.

Display Full Size

Figure 3. Possible Mechanism.

Based on the mechanistic studies and related literature, a plausible mechanism was proposed (Figure 3). Firstly, cobalt(II) salt coordinates with L4 and substrate 1 to generate the cobalt(II) intermediate Int-1, which is oxidized to the cobalt(III) Int-2 by oxygen in the air. Notably, this process may need to be facilitated by direct photoexcitation and photoredox catalysis, considering the different electrical properties and steric hindrances of the substrates. Subsequently, carboxylate-assisted C–H activation led to the five-membered cobaltacycle intermediate Int-3, which undergoes ligand exchange with alkynes 2 to give the intermediate Int-4. Then, the migratory insertion of C–Co(III) bond of alkynes 2 affords the seven-membered cobaltacycle intermediate Int-5. Finally, the reductive elimination of Int-5 leads to chiral product 3, releasing the cobalt(I) intermediate Int-6, which could be easily re-oxidized to the cobalt(II) intermediate Int-1 under aerobic conditions.

After determining the optimal reaction conditions and plausible mechanism, we began to investigate the substrate scope of this asymmetric oxidative [4+2] cycloaddition under three different reaction conditions (Figure 4). Electron-donating substituted diarylacetylenes (3ab, 3ac) gave the desired products with reduced yields under the conditions without both PC5 and light, and the irradiation with a 5 W purple LEDs was necessary in order to ensure a good yield. Electron-withdrawing substituted diarylacetylenes (3ad-3ag) and diheteroarylacetylene (3ah) had low yields without both PC5 and light. However, irradiation with a 5 W purple LEDs improved the yield, while a 5 W blue LEDs and PC5 gave better results. Unfortunately, 4-Br-substituted diarylacetylene (3af) afforded lower yields under all three sets of conditions, likely due to the poor solubility of 2f in HFIP. Nevertheless, light still produced a measurable yield enhancement. Alkyl alkynes (3ai, 3aj) with low site resistance also gave good yields under the conditions without both PC5 and light, which can be attributed to the fact that the alkynes act as weak electron-rich ligand to facilitate the oxidative regeneration of low-valent cobalt intermediates^[56]. Asymmetric di-substituted alkynes (3ak, 3al) showed moderate yields under conditions without both PC5 and light, although prop-1-yn-1-ylbenzene (3ak) was poorly regionally selective. When mono-substituted aryl (3am), heteroaryl (3an), cyclohexenyl (3ao), and diphenylphosphinyl (3ap, 3aq) acetylenes were used as substrates, irradiation with a 5 W purple LEDs always enhanced the yield, and the yield was further improved under the condition of a 5 W blue LEDs and PC5. Remarkably, 2-thiophene acetylene (3an) only maintained a moderate yield. This low reactivity could be partially explained by sulfur-induced inactivation of the cobalt catalyst^[57].

Display Full Size

Figure 4. The scope of alkynes. ^a: 1 (0.2 mmol), 2 (0.3 mmol), Co(NO₃)₂•H₂O (10 mol%), L4 (15 mol%), PivONa (2 equiv) in HFIP (2 mL) under air at rt for 48 h, isolated yields; ^b: Under 5 W purple LED; ^c: Using PC5 (2 mol%), 5 W blue LED; ^d: Using Co(NO₃)₂·H₂O (30 mol%), L4 (45 mol%). LED: light-emitting diode; HFIP: 1,1,1,3,3,3-hexafluoro-2-propanol.

We next turned our attention to determining the scope of arylamides that are amenable to asymmetric cyclization reaction under the three sets of conditions (Figure 5). A range of amide groups containing electron-neutral or -rich substituents at the para-position, such as Me, H, OMe, could be employed as suitable substrates, and the corresponding products (3aa-3ca) were all produced in good to excellent yields (82-98%) with high enantioselectivity (96-97% ee) under the three sets of conditions. However, the analogous reaction with electron-deficient substituents at the para-position, such as CO₂Me (3da), SO₂Me (3ea), CN (3fa), Br (3ga), and I (3ia), was less effective under the conditions without both PC5 and light, and irradiation with a 5 W purple LEDs was needed to ensure good yields. Ortho-substituted Br, Cl, 1-naphthyl, and 3,5-/2,4-di-substituted (3ja-3na) benzamide derivatives did not react or had only trace yields without both PC5 and light. Again, the yield can be increased by irradiation with a 5 W purple LED, and the replacement of a 5 W blue LEDs and PC5 led to better results. Remarkably, 2-thienylamide (3oa) and the introduction of an electron-rich group (Me, 3ia) at the 2-position of the benzamide are well tolerated under the three sets of conditions. Heterocyclic amides, including 3-thienyl (3pa), 2-furanyl (3qa), 3-furanyl (3ra), and ortho-substituted pyridine (3sa) were subsequently reacted, with inferior results obtained without both PC5 and light, but irradiation with a 5 W purple LEDs often resulted in significantly enhanced yields. Benzamides bearing electron-deficient (F, Cl, Br, I), or electron-rich (Me, OMe) substituents in the meta-position reacted efficiently with 2a to give the corresponding product 3ta−3ya in good to excellent yields (78-96% yield) and high enantioselectivity (90-96% ee) under the three sets of conditions. The regioselectivity (C2 or C6) was entirely determined by the size of the steric hindrance of the substituent group. The steric hindrance may weaken the coordination between N-atom and the cobalt center and reduce the electron density of the low-valent cobalt intermediate. Additionally, the 2-position (3za) or 7-position(3aaa) methyl substitution on quinoline also leads to a decrease in enantioselectivity. The absolute stereochemistry of 3aa was established by X-ray diffraction (CCDC: 2416014).

Display Full Size

Figure 5. The scope of arylamides. ^a: 1 (0.2 mmol), 2 (0.3 mmol), Co(NO₃)₂·H₂O (10 mol%), L4 (15 mol%), PivONa (2 equiv) in HFIP (2 mL) under air at rt for 48 h, isolated yields; ^b: under 5 W purple LED; ^c: Using PC5 (2 mol%), 5 W blue LED; ^d: 36 h. LED: light-emitting diode; HFIP: 1,1,1,3,3,3-hexafluoro-2-propanol.

Product 3aq-major could be smoothly transformed to C–N axially chiral monophosphine compound 4 with high enantioselectivity by using HSiCl₃ (Figure 6a). Notably, the compound 4 could act as a potential chiral ligand for the palladium-catalyzed asymmetric Tsuji−Trost reaction, providing the desired product 7 in 78% yield with 45% ee (Figure 6b).

Display Full Size

Figure 6. Synthetic application.

4. Conclusion

In summary, we have developed a Co-catalyzed asymmetric aerobic oxidative C(sp²)–H bond activation/[4+2] cycloaddition. The reactions proceed under room temperature with air as an oxidant, and provided an efficient and highly enantioselective method for the assembly of axially chiral isoquinolinones. Light irradiation can increase the yield of lowly active substrates by activating oxygen. Preliminary experiments indicate that this axially chiral skeleton is a promising chiral ligand for Pd-catalyzed asymmetric reactions. Substrate studies and mechanistic experiments reveal a correlation between substrate activity and its substituent electrical properties and steric hindrance. We believe that this work provides an unprecedentedly mild and sustainable new strategy for transition metal catalyzed asymmetric aerobic oxidation.

Supplementary materials

The supplementary material for this article is available at: Supplementary materials.

Acknowledgement

We thank Prof. Jia-Rong Chen, Prof. Zhihan Zhang, and many other group members for helpful discussions.

Authors contributions

Wang LN: Investigation, writing-original draft.

Fu YLT: Writing-review & editing.

Liao CY: Investigation, Resources.

Xiao WJ: Methodology, writing-review & editing.

Lu LQ: Conceptualization, methodology, writing-review & editing.

Conflicts of interest

The authors declare that they have no conflicts of interest.

Ethical approval

Not applicable.

Consent to participate

Not applicable.

Consent to publication

Not applicable.

Data availability statement

The data supporting this article have been included as part of the Supplementary Information. Crystallographic data for (3aa, [Co]-1) has been deposited at under (2416014, 2489953) and can be obtained from [URL of data record, format https://doi.org/DOI].

Funding

This work was supported by the National Natural Science Foundation of China (Grant Nos. 92256301, 22471089, and 22271113), and the National Key R & D Program of China (Grant Nos. 2023YFA1507203 and 2024YFA1509700).

Copyright

© The Author(s) 2026.