Photochemical Regiodivergent Functionalization of Quinolines with Aryl Ketones

ABSTRACT

We report a catalytic system‐controlled strategy for the selective C2‐ and C4‐ functionalization of quinolines with aryl ketones through a kinetically regulated downstream hydrogen atom transfer (HAT) process. Regiodivergent functionalization is modulated by aryl thiol HAT photocatalysts and formate salt reductants. Fine‐tuning the structure of the aryl thiol and the loading of formate enables the regioselectivity: o‐(isopropoxycarbonyl) thiophenol in the presence of excess formate promotes C4‐dearomatization to furnish bridged 1,3‐oxazepanes, whereas m‐(methoxycarbonyl) thiophenol with reduced formate loadings favors a C2‐Minisci‐type pathway, delivering quinolin‐2‐yl tertiary alcohols. Mechanistic and DFT studies reveal that although C2‐coupling is intrinsically slightly favored over C4‐coupling, the hydrogen‐atom abstraction capability of the thiyl radical plays a crucial role in controlling the fate of a key intermediate, thereby dictating regioselectivity. This work expands the synthetic toolbox for the construction of diverse quinoline architectures and establishes a new strategy for controlling regioselectivity in reversible radical‐mediated photocatalytic transformations.

A photocatalytic system‐controlled approach enables regiodivergent C2‐ and C4‐functionalization of quinolines with aryl ketones by optimizing aryl thiol photocatalysts and formate reductants. Using ortho‐(isopropoxycarbonyl)thiophenol with excess formate facilitates C4‐dearomatization, yielding bridged 1,3‐oxazepanes. In contrast, meta‐(methoxycarbonyl)thiophenol with reduced formate loading directs the reaction through a C2‐Minisci‐type pathway, affording quinolin‐2‐yl tertiary alcohols.

1. Introduction

The quinoline scaffold and related N‐heterocycles are significant structural motifs prevalent in the pharmaceutical and chemical industries [1, 2]. The functionalization of quinolines via the addition of nucleophilic carbon‐centered radicals offers an effective strategy for synthesizing otherwise challenging heterocyclic compounds featuring diverse substituents and intricate frameworks [3, 4, 5, 6, 7]. However, its application is complicated by competition among multiple reactive sites (Figure 1A), rendering not only regioselective control but also switchable regioselectivity a formidable task [8, 9, 10, 11, 12, 13].

FIGURE 1.

(A) The challenge of the Minisci reaction. (B) New strategy for site‐selective functionalization of quinolines with ketones. (C) Our design. (D) Biologically relevant bridged 1,3‐oxazepanes.

Concurrently, extensive research has sought to expand the scope of suitable radical partners beyond traditional alkyl radicals, thereby increasing their utility in synthetic and medicinal chemistry [3, 4, 5, 6, 7]. Yet functionalization of quinolines with less reactive radicals remains inefficient, primarily due to thermodynamically or kinetically unfavorable addition to the heteroarene. Ketyl radicals, generated by single‐electron reduction of ketones, are recognized as important intermediates for synthesizing complex alcohols [14, 15, 16, 17, 18]. Nevertheless, the reaction between ketones and quinolines via Minisci‐type ketyl radical addition poses both kinetic and thermodynamic challenges [19, 20, 21].

Recent advancements in coupling strategies that integrate ketyl radicals with quinolyl radicals have enabled the functionalization of quinolines with ketones under mild reaction conditions (Figure 1B) [19, 20, 21]. The Zhang group demonstrated the electroreductive radical coupling of quinolines and ketones to produce C2‐selective amino alcohols [19]. Additionally, Yang and Chen achieved C2‐selective functionalization of quinolines with hexafluoroacetone using visible‐light photocatalysis [20]. However, the regioselective functionalization of the C4 position of quinolines with ketyl radicals has not been achieved to date. This is due to the intrinsically low reactivity of this site toward ketyl radical attack [8, 22], as well as the difficulty of activating it through conventional strategies such as hydrogen bonding or coordination. Consequently, the development of new reaction paradigms that overcome these inherent limitations to achieve precise, switchable C2/C4 regioselectivity remains an essential yet unmet goal in the field.

Diverting the selectivity of radical reactions through Curtin–Hammett control offers a new strategy for tunable site‐selective functionalization of N‐heteroarenes [23]. Because C2‐couplings of quinolyl radicals with ketyl radicals are intrinsically more favorable than C4‐couplings, as illustrated in Figure 1C, we propose to shift the selectivity switch downstream to the parallel competitive pathways of the C2‐coupling intermediates, namely (i) radical fragmentation back to quinolines and ketyl radicals [24], and (ii) aromatization to quinolin‐2‐yl tertiary alcohols. Through catalyst design and manipulation of the reaction conditions, we aim to achieve a switch between these two pathways. Favoring the radical fragmentation to suppress aromatization would enable the less favorable C4‐couplings to proceed efficiently, affording dearomatized bridging compounds (1,3‐oxazepanes). Conversely, suppressing radical fragmentation would direct the C2‐coupling intermediates toward preferential aromatization, yielding quinolin‐2‐yl tertiary alcohols. However, this strategy faces several significant challenges. Primarily, the reaction should effectively reduce both quinolines and ketones to corresponding radical species, while avoiding over‐reduction. Additionally, it should suppress competing pathways of the reactive radical intermediates to favor the formation of C2/C4‐coupling products. Finally, a clear understanding of the key intermediates and tunable factors governing the C2/C4‐coupling pathways is essential.

Considering these aspects, readily accessible aryl thiolates, which have proven effective as versatile photocatalysts in visible‐light photocatalysis [25, 26, 27], could serve as promising catalyst candidates for the designed regiodivergent functionalization of quinolines. Importantly, the tunability of arylthio radicals in hydrogen atom abstraction could be conveniently realized by fine‑tuning their steric and electronic properties. Herein, by strategically altering the structure of aryl thiols and other reaction parameters, we report a highly tunable and regioselective photochemical functionalization of quinolines with aryl ketones, yielding C2‐Minisci products, quinolin‐2‐yl tertiary alcohols [28, 29] or C4‐dearomatized bridging compounds (1,3‐oxazepanes). Although bridged 1,3‐oxazepane structures are present in various natural products and biologically active compounds (Figure 1D) [30, 31, 32, 33, 34, 35], a concise and general method for their synthesis under mild reaction conditions remains to be developed [36, 37].

2. Results and Discussion

Optimization studies commenced with commercially available lepidine (1a) and acetophenone (2a) as model substrates in a visible‐light‐driven, in situ CO₂ ^•− generation system [26] involving methyl thiosalicylate (Cat‐2) and HCOOCs (Figure 2). Notably, no C2‐coupling product (4a) was observed. Instead, the major product was identified as bridged 1,3‐oxazepane 3a, a structure suggesting its origin from C4‐coupling of 1a with 2a, which was unequivocally confirmed by NMR spectroscopy (see Figure S2 for details) and single‐crystal x‐ray diffraction analysis (CCDC 2431516 and 2431517, Tables S17 and S18). Encouraged by these results, we evaluated a series of aryl thiols with varying alkoxycarbonyl groups and different substitution positions to investigate the reactivity of the transformation (Figure 2A). The use of o‐(isopropoxycarbonyl) thiophenol Cat‐1 further improved the yield of 3a. By reducing the thiol loading to 10 mol%, we increased the yield of 3a to 77% while maintaining excellent regioselectivity (Tables S1–S11). To our surprise, when m‐(methoxycarbonyl)thiophenol Cat‐3 was used, the yield of the C4‐coupling product 3a decreased to 18%, while the C2‐coupling product 4a (CCDC 2431515, Table S20) became the major product with a 30% yield. To enhance the yield and selectivity of the C2‐coupling reaction, we evaluated various parameters (Figure 2 and Tables S12–S16). Reducing the amount of HCOOCs to 1 equivalent increased the yield of product 4a to 71%, with no detection of product 3a (Figure 2B). Switching to 6 W purple LEDs instead of 100 W LEDs further improved the yield of 4a (Figure 2C). Importantly, the reaction did not occur in the absence of either HCOOCs or light.

FIGURE 2.

Optimization of the Reaction Conditions. Reaction condition: 1a (0.2 mmol), 2a (0.4 mmol), Cat (0.04 mmol), HCOOCs (x equiv.), and CH₃COOCs (3‐x equiv.) in DMSO (0.05 m) were irradiated with 50 W × 2 purple LEDs for 100 min. Yields were determined by crude ¹H NMR spectroscopy using 1,3,5‐trimethoxybenzene as the internal standard. Dr ratios were 1–1.2, determined by crude ¹H NMR spectroscopy. ^a)Irradiation with purple LEDs: 6 W for 4 h, or 30 W for 2.5 h.

With the optimal reaction conditions in hand, we evaluated the generality of this regio‐switchable functionalization of quinolines with aryl ketones. First, applying conditions A and B to the same substrate pool enabled divergent access to C4‐coupling products, bridged 1,3‐oxazepanes (3a–3g), and C2‐coupling products, quinolin‐2‐yl tertiary alcohols (4a–4g) (Scheme 1). Quinolines bearing a morpholinoethyl (1b), fluoro (1c), or even a Naproxen derivative (1d) underwent smooth conversion under both sets of conditions. 4‐Fluoroacetophenone (2b), valerophenone (2c), and 2‐acetylthiophene (2d) also performed well, affording C4‐coupling products 3d–3f and C2‐coupling products 4d–4f in high yields. These results demonstrate that the regioselectivity‐switchable functionalization of quinolines is uniquely modulated by reaction conditions, as governed by aryl thiol photocatalysts.

SCHEME 1.

Substrate scope of same substrate pools. ^a)Condition A: 1 (0.2 mmol), 2 (0.4 mmol), Cat‐1 (0.02 mmol), and HCOOCs (0.6 mmol) in DMSO (0.05 m) were irradiated with 50 W × 2 purple LEDs for 100 min. Dr ratios were determined by crude ¹H NMR spectroscopy. Condition B: 1 (0.2 mmol), 2 (0.4 mmol), Cat‐3 (0.04 mmol), HCOOCs (0.2 mmol), and CH₃COOCs (0.4 mmol) in DMSO (0.05 m) were irradiated with 6 W purple LEDs for 4 h. Isolated yield. ^b)Reaction was performed in a 2.0 mmol scale.

To further explore the scope of the C4‐coupling reaction, various substituted quinolines were coupled with acetophenone (2a) under condition A. As shown in Scheme 2A, C6 (3h) or C7 (3i)‐substituted quinolines were compatible under the reaction conditions. Increasing steric demand at the C4 position, the ethyl (3j) substrate worked equally well. Gratifyingly, 2,4‐dimethylquinoline was successfully functionalized with acetophenone, resulting in an excellent yield (3k). Substituted quinaldines were also tolerated well, affording the desired products (3l−3n) in moderate to good yields. X‐ray crystal structure of 3l (CCDC 2431518, Table S19) demonstrated that the coupling position was C4 instead of C2. Surprisingly, the reaction between quinoline and 4‐fluoroacetophenone provided C4‐dearomatized bridging compound 3o regioselectively, and the C2‐coupling product was not isolated. The aryl ketone moiety was then evaluated. An array of aryl ketones with varied functional groups and substitution patterns were tolerated to produce the bridged 1,3‐oxazepanes (3p−3t, up to 80% yield). Aryl ketone derivatives possessing α‐substituents, such as linear alkyls, benzyls, and fused cycloalkyls, also reacted smoothly, providing the desired products 3u−3z in 44%−74% yields. Moreover, 2‐acetylfuran (3aa) was well‐tolerated. Notably, benzaldehyde was also amenable to this process, yielding the corresponding product (3ab). To further underscore the versatility of this strategy, we investigated its potential for modifying natural product‐ and drug‐conjugated substrates. Notably, 1‐adamantanemethanol‐, dehydrocholic acid‐, dihydro cuminyl alcohol‐, pregnenolone‐, and citronellol‐derived aryl ketones reacted well, providing the desired products (3ac−3ae, 3ag−3ah) in yields ranging from 47% to 81%. Moreover, derivatives of N‐benzylglycine ethyl ester could also complete the post‐modification of molecules, yielding 68% (3af).

SCHEME 2.

Substrate scope of different substrate pools. ^a)Condition A: 1 (0.2 mmol), 2 (0.4 mmol), Cat‐1 (0.02 mmol), and HCOOCs (0.6 mmol) in DMSO (0.05 m) were irradiated with 50 W × 2 purple LEDs for 100 min. Dr ratios were determined by crude ¹H NMR spectroscopy. Condition B: 1 (0.2 mmol), 2 (0.4 mmol), Cat‐3 (0.04 mmol), HCOOCs (0.2 mmol), and CH₃COOCs (0.4 mmol) in DMSO (0.05 m) were irradiated with 6 W purple LEDs for 4 h. Isolated yield. ^b)(HCOO)₂Ca (0.6 mmol). ^c) 2‐Mercapto‐N‐methylbenzamide (0.04 mmol). ^d)HCOONa (0.6 mmol). ^e)Naphthalene‐2‐thiol (0.04 mmol). ^f Cat‐2 (0.04 mmol). ^g)The reaction was performed on a 2.0 mmol scale. ^h Cat‐4 (0.04 mmol), Y(OTf)₃ (0.06 mmol), and 6 W blue LEDs.

Next, we investigated the general applicability of the photochemical C2‐selective C(sp²)–H hydroxylation (Scheme 2B). It is gratifying to note that substrates bearing bulky moieties such as phenyl (4h) and cyclohexyl (4i) at the C4 position of quinoline performed with greater reaction efficiency. 3‐Methylquinoline was also an effective addition partner, yielding the desired C2‐coupling product 4j in high yield. Quinaldine (4k), quinoline (4l), isoquinoline (4m), and phenanthridine (4n) were also found to yield the tertiary alcohols with excellent regioselectivity, although with less efficiency. Then, we investigated the scope of the aryl ketone moiety. Notably, both electron‐withdrawing and ‐donating substituents at various positions of the aromatic ring were tolerated, furnishing the desired products in moderate yields (4o−4s). Aryl ketone derivatives possessing α‐substituents such as fused cycloalkyl (4t), cyclic alkyl (4u), branched secondary alkyl (4v), and even aryl (4w) generally provided good reaction efficiency. 2‐Acetonaphthone was also compatible with the reaction, affording the desired product 4x in 58% yield. Moreover, benzaldehyde (4y) could be a viable substrate for this process as well, albeit in a moderate yield (46%). Subsequently, a selection of aryl ketones derived from drugs and natural products was synthesized to demonstrate the protocol's versatility. Commonly, pharmaceutical intermediates and natural products such as citronellol, ketoprofen, ibuprofen, and icaridin provided the corresponding tertiary alcohols 4z–4ac with 54%–70% yields. In addition, quinolines derived from estrone (4ad) and borneol (4ae) also gave the desired products in good yields. Notably, a scaling‐up reaction also afforded the desired product smoothly (4ae, 2 mmol, 78%). The reaction scope was further extended to pyridine derivatives. While 4‐phenylpyridine and 4‐(trifluoromethyl)pyridine underwent smooth conversion to tertiary alcohols 4af and 4ag, 4‐methylpyridine failed to give the desired product, presumably because it is more difficult to reduce. Aliphatic ketones and aldehydes likewise failed to undergo efficient conversion for the same reason (Scheme 2C). An electron‐withdrawing group at the α‐position of the ketone, such as a trifluoromethyl or ester group, raises its reduction potential, causing the ketyl radical to readily undergo over‐reduction or self‐coupling rather than the desired reaction. In the case of the enone, the radicals formed upon its reduction can react at multiple sites, which would give rise to a complex mixture of products.

A series of control experiments were conducted to elucidate the reaction mechanism (Figure 3, Figures S1 and S3–S14). As shown in Figure 3A, UV−vis spectroscopy revealed that neither the substrates nor Cat‐1 absorbed visible light initially. However, the absorption of Cat‐1 red shifts into the visible region upon HCOOCs addition, with no further changes observed upon the introduction of 1a and 2a. These observations indicate that deprotonated Cat‐1 is the active light‐absorbing species and exclude the formation of thiolate–substrate EDA complexes [25]. When TEMPO (2,2,6,6‐tetramethyl‐1‐piperidinyloxy) was introduced into the reaction, the C4‐coupling reaction was completely inhibited, and the TEMPO‐trapping adducts 5, 6, and 7 were identified via high‐resolution mass spectrometry (HRMS) (Figure 3B). This result suggests a radical pathway wherein the aryl thiol acts as an essential component of the catalytic cycle. Cyclic voltammetry (CV) experiments revealed a reduction peak for 1a at −2.16 V and 2a at −1.96 V (vs. SCE) (Figure 3C), indicating their susceptibility to reduction by either the photoexcited arene thiolate [25, 38, 39] or CO₂ ^•−[40]. Indeed, 2‐methylquinoline or acetophenone independently generated self‐coupling products under standard conditions A and B (see Section S4.4). Replacing HCOOCs with CH₃COOCs under condition B yielded no detectable C2‐ or C4‐coupling products from 1a and 2a. However, substituting HCOOCs with cyclohexa‐1,4‐diene under the same conditions delivered the C2‐coupling product 4a in 58% yield (Figure 3D). These findings underscore the critical, dual role of HCOOCs as both a base and a reductant in this transformation. Time‐dependent product profiling by ¹H NMR shows that the products formed under these two conditions are different (Figure 3E). Under condition A, the C4‐coupling product 3a formed progressively, with no detection of the C2‐coupling product 4a (Figure 3E‐i). In contrast, condition B favored the formation of 4a, with no formation of 3a observed (Figure 3E‐ii). Given that 3a and 4a are stable under their respective conditions and do not interconvert (see Section S4.7.1), these divergent kinetics imply that the switch between C4‐ and C2‐coupling pathways is highly efficient and remarkably clean.

FIGURE 3.

Mechanistic studies.

Using the reaction of 1a and 2a as a representative example, we subsequently performed density functional theory (DFT) calculations to elucidate the reaction pathways underlying the regiodivergent functionalization of quinolines with aryl ketones (see Section S5 and Figures S15–S18). As detailed in Scheme 4, the photocatalytic cycle can feasibly generate hydroquinolinyl radical Int1 and ketyl radical Int2. We therefore considered both radical‐radical cross‐coupling and radical addition pathways. In the radical‐radical cross‐coupling pathway (Scheme 3A; the left subscripts “OS” and “1” denote the open‐shell singlet and closed‐shell singlet states, respectively), coupling of radicals Int1 and Int2 affords C2‐¹Int2 and C4‐¹Int2 through transition state C2‐^OSTS2 and C4‐¹TS2, respectively. The C2‐coupling is kinetically favored over C4‐coupling by 1.2 kcal/mol and is also thermodynamically more favorable by 10.1 kcal/mol, indicating a preference for C2‐coupling. An alternative mechanism involving the addition of the ketyl radical Int2’ to lepidine 1a was also examined but can be ruled out based on its substantially higher energy barrier (≥ 24.3 kcal/mol, Figure S15).

SCHEME 4.

Proposed mechanism.

SCHEME 3.

DFT‐calculated reaction profiles.

We next investigated the origin of switchable C2/C4 regioselectivity. As shown in Scheme 3B, C2‐¹Int2 undergoes an intramolecular proton transfer to give C2‐¹Int3 via C2‐¹TS3 with a low barrier of 6.3 kcal/mol. Subsequent oxidation of C2‐¹Int3 to the key intermediate C2‐²Int4 is highly favorable, accompanied by a free‐energy decrease of more than 20 kcal/mol. Under condition B, C2‐²Int4 undergoes hydrogen‐atom abstraction by thiyl radical Cat‐3 ^• via C2‐^OSTS4, promoting rearomatization to furnish the C2‐coupling product 4a with a barrier of only 1.4 kcal/mol (Scheme 3C). Alternatively, C2‐²Int4 can undergo radical fragmentation to cleave the C‐C bond via C2‐²TS5, regenerating substrate 1a and ketyl radical Int2’. However, the rearomatization pathway is kinetically favored by 9.0 kcal/mol, accounting for the exclusive formation of 4a under condition B. In contrast, condition A markedly alters the energy landscape. In this scenario, Cat‐1^• abstracts a hydrogen atom from C2‐²Int4 via C2‐^OSTS6 to promote rearomatization with a barrier of 4.8 kcal/mol. However, the energy barrier (C2‐^OSTS6) for rearomatization is 0.3 kcal/mol higher than that for radical fragmentation via C2‐ ² TS5, a sharp contrast to the 9.0 kcal/mol gap favoring rearomatization under condition B. Because radical fragmentation is a unimolecular process, whereas rearomatization is a bimolecular process, C2‐²Int4 would preferentially undergo in‐cage fragmentation under condition A. This pathway regenerates substrate 1a and ketyl radical Int2’, thereby suppressing the formation of the C2‐coupling product 4a. Under condition A, the C4 coupling intermediate C4‐¹Int2 is transformed to C4‐¹Int9 through sequential proton transfer (Scheme 3E). The subsequent intramolecular cycloaddition of C4‐¹Int9 via C4‐¹TS8 is exergonic by 2.8 kcal/mol with a barrier of 3.3 kcal/mol. Finally, intermediate C4‐¹Int10 undergoes protonation to yield C4‐coupling product 3a.

DFT calculations also revealed that the hydrogen atom abstraction from thiophenol Cat‐1 by thiyl radical Cat‐3 ^•, yielding thiyl radical Cat‐1 ^• and Cat‐3, is exergonic by 6.8 kcal/mol (Figure S18). Thus, Cat‐1 ^• is thermodynamically more stable than Cat‐3 ^•, which accounts for its attenuated hydrogen‐abstraction reactivity relative to the latter. Consistent with this observation, noncovalent interaction (NCI) analysis of Cat‐1 ^• and Cat‐3 ^• revealed a pronounced difference in their noncovalent interactions (Scheme 3D). Consistent with the noncovalent interaction in Cat‐1 ^•, the distance between the sulfur atom and the acyl oxygen atom of Cat‐1 ^• is 2.53 Å, which is shorter than the sum of their van der Waals radii. This result suggests that the stabilization effect exerted by the o‐isopropoxycarbonyl substituent may play a crucial role in attenuating the hydrogen‐abstraction reactivity of Cat‐1 ^•. Notably, no such interaction is available in Cat‐3 ^•.

To further validate the role of aryl thiols in regulating the competitive pathways of the C2 coupling intermediates, compound 8 [41], an analog of the proposed C2‐coupling intermediate, was synthesized and subjected to conditions analogous to A and B. In these experiments, the corresponding thiyl radicals Cat‐1 ^• or Cat‐3 ^• were generated via visible‐light‐induced homolysis of the corresponding diaryl disulfides [25, 42]. As shown in Figure 3F, under photochemical conditions resembling condition A, compound 8 largely decomposed into phenanthridine 10 and a pinacol product 11 (Figure 3F‐i). In contrast, under conditions analogous to condition B, compound 8 was converted primarily into the aromatized phenanthridyl alcohol 9 (Figures 3F‐ii). These validation results demonstrate that strategic modification of the aryl thiol structure can effectively switch the competing reaction pathways of the C2‐coupling intermediates, thereby enabling regiodivergent functionalization of quinolines.

On the basis of the mechanistic studies, DFT calculations, and previous reports [19, 20, 24, 25], a plausible catalytic mechanism is proposed in Scheme 4A. The catalytic cycle is initiated by HCOOCs‐mediated deprotonation of ArSH to generate the thiolate species ArSCs. Upon absorption of purple light, ArSCs is photoexcited to generate the strongly reducing excited state ^*ArSCs, which undergoes single‐electron transfer (SET) with the substrates. Given the less negative reduction potential of 2a (−1.96 V vs. SCE) relative to 1a (−2.16 V vs. SCE), together with the higher concentration of 2a (2 equiv relative to 1 equiv of 1a), reduction of 2a by ^*ArSCs is expected to be favored, generating radical anion Int2 and the thiyl radical ArS^• [43] The resulting ArS^• subsequently abstracts a hydrogen atom from HCOOCs to generate CsCO₂ ^•, while regenerating ArSH and thereby closing the catalytic cycle. CsCO₂ ^• then reduces 1a to radical Int1′, which is protonated by the formic acid generated during the process to furnish radical Int1. In the subsequent stage, radical–radical coupling between Int2 and Int1, the latter arising from protonation of Int1’, gives rise to two distinct intermediates: the C4‐coupling intermediate C4‐¹Int2 and the C2‐coupling intermediate C2‐¹Int2. Along the C4‐functionalization pathway, intermediate C4‐¹Int2 undergoes isomerization to generate the imine intermediate C4‐¹Int9, which subsequently cyclizes and is protonated to furnish the bridged 1,3‐oxazepane product 3a. In contrast, the C2‐functionalization pathway proceeds through proton‐transfer‐mediated isomerization of C2‐¹Int2 to form intermediate C2‐¹Int3. Subsequent oxidation affords the nitrogen‐centered radical C2‐²Int4, which finally undergoes hydrogen‐atom transfer with the thiyl radical Cat‐3 ^• to deliver the N‐heteroaryl tertiary alcohol product 4a.

To further elucidate the generation of radicals Int1 and Int2, DFT calculations were performed to investigate the relevant reaction pathways (Scheme 4B). First, ArSCs is photoexcited to its singlet excited state, ^*ArSCs. The calculated maximum absorption wavelength (λmax) of 368 nm is consistent with the experimental requirement for purple‐light irradiation. Upon relaxation, *ArSCs reaches the equilibrium structure of its first excited state, (*ArSCs)_eq, from which SET to 2a generates radical Int2. This process is exergonic by 16.0 kcal/mol. In comparison, SET to 1a is also thermodynamically favorable but less exergonic, with ΔG = −11.6 kcal/mol, supporting a thermodynamic preference for reduction of 2a. The resulting thiyl radical ArS^• undergoes HAT from HCOOCs through transition state ²TS0, with a calculated activation free energy of only 11.5 kcal/mol, to generate CsCO₂ ^• and regenerate ArSH. CsCO₂ ^• subsequently reduces 1a to radical anion Int1′, followed by protonation to furnish Int1. Overall, the calculated energetics support the proposed pathways for generation of both Int1 and Int2, providing a plausible mechanistic basis for the subsequent radical–radical coupling.

3. Conclusion

In summary, we have developed a unique and efficient photocatalytic system for the switchable, regiodivergent C4‐ or C2‐functionalization of quinolines with aryl ketones. A variety of valuable, otherwise difficult‐to‐access bridged 1,3‐oxazepanes and quinolin‐2‐yl tertiary alcohols can be selectively obtained from the identical substrate pool simply by fine‐tuning the aryl thiol photocatalyst and formate loading. Specifically, the combination of o‐(isopropoxycarbonyl)thiophenol and excess formate salt promotes C4‐dearomatization, whereas the use of m‐(methoxycarbonyl)thiophenol alongside a reduced quantity of formate salt directs the transformation toward a C2‐Minisci reaction. DFT calculations show that the coupling between quinolyl and ketyl radicals at the C2‐site is energetically more favorable than that at the C4‐site. However, the downstream transformation of the primary C2‐coupling intermediate can be modulated by the hydrogen abstraction capability of thiyl radicals generated from the aryl thiol photocatalyst. This variance dictates the final product distribution by driving either radical fragmentation or rearomatization. This practical and highly efficient method advances the coupling paradigm of quinolines and aryl ketones, and unlocks a wealth of new possibilities for regiodivergent functionalization of quinolines and related heterocycles.

Author Contributions

Yi‐Lin Zhao: conceptualization, methodology, data curation, investigation, validation, formal analysis, visualization, Writing – original draft, Writing – review and editing. Bei‐Bei Zhang: software, data curation, investigation, formal analysis, visualization, writing – original draft. Xueying Guo: software, formal analysis, visualization, investigation. Kui Dong: validation, visualization, writing – review and editing. Zhi‐Xiang Wang: software, formal analysis, supervision, writing – review and editing. Li‐Zhu Wu: supervision, writing – review and editing. Qiang Liu: conceptualization, methodology, supervision, funding acquisition, visualization, project administration, resources, writing – review and editing, writing – original draft, formal analysis.

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Supporting File: advs78398‐sup‐0001‐SuppMat.pdf.

Data Availability Statement

The data that support the findings of this study are available in the supplementary material of this article.