Green photocatalytic N-debenzylations with molecular oxygen catalyzed by recyclable metal-free tubular carbon nitride

Abstract

Debenzylations are widely used in fields such as pharmaceuticals and fine chemicals. In this study, metal-free tubular carbon nitride (CN-T) was prepared and used in photocatalytic debenzylations with molecular oxygen as the sole oxidant, affording 44 kinds of debenzylated amines in yields up to 93%. The average yield of 88% and catalyst recovery rate of 97% were obtained in the 6 cycle recovery and amplification experiments. Both experiments and DFT calculations were conducted to reveal the detailed reaction mechanism. The in situ generated Cl· radicals from HCl play an important role in the HAT process for debenzylations of tertiary amines. ¹O₂ could promote the debenzylations of primary/secondary amines by simultaneously extracting hydrogen atoms from N–H and benzylic C–H groups to form imines, which readily undergo hydrolysis. The utilization of CN-T offers a convenient pathway for catalyst recovery and reuse.

---

1 Introduction

The benzyl group, which exhibits exceptional chemical stability and good tolerance to acidic and basic conditions, is an essential organic synthetic protective group and widely used in the synthesis of drugs and fine chemicals.^1–5 The removal of the benzyl group, namely debenzylation, requires harsh reduction/oxidation processes. Traditional reduction and oxidation systems, such as H₂/Pd-C,^6,7 ceric ammonium nitrate (CAN),⁸ oxone,⁹ and dimethyldioxirane,¹⁰ give rise to multiple concerns including catalyst deactivation,¹¹ safety hazards,¹² and environmental implications. Lately, scientists have been diligently engaged in developing a sustainable methodology that utilizes molecular oxygen as an environmentally benign oxidant for the effective removal of the benzyl group.

As early as 1983, Gigg and co-workers¹³ prepared acetylated amines via debenzylations in the presence of KO^tBu and O₂. In general, DMSO, known for its elevated boiling point and viscosity, served as the co-reductant and solution for this reaction system^13,14 (Fig. 1a), always leading to solvent residual concerns.¹⁵ Photocatalysis is a green technology capable of transforming abundant photonic energy into valuable chemical energy, offering a sustainable and eco-friendly solution.^16–18 The photocatalytic strategy provides a convenient way to drive debenzylations under mild conditions due to the spontaneous electron transfer processes between N-benzylamines and electronically-excited photosensitizers with strong oxidizing properties. Both König¹⁹ and Cho²⁰ groups had successfully removed benzyl groups from N-benzylated amines in the absence of bases/acids utilizing flavin and [Ru(II)(bpy)₃]²⁺ as photocatalysts (Fig. 1b). Nevertheless, this method fell short in effective debenzylations of tertiary amines. For a long time, it has been a great challenge to remove the benzyl group efficiently from general N-benzyl tertiary amines while preserving other groups by using molecular oxygen as the sole oxidant. Anandhan and his co-workers²¹ had removed the benzyl group from tertiary amines by utilizing rose bengal as the photocatalyst, nevertheless the alkyl groups connected to the same nitrogen atoms were converted into the corresponding carbonyl groups. Moreover, the two N-benzyl groups were all converted to benzoyl groups in the presence of [Acr⁺-Mes]BF₄⁻ and acetic acid²² (Fig. 1c). Fortunately, our group had succeeded in removing the benzyl group from N-benzyl tertiary amines by photocatalytic N-debenzylations with molecular oxygen in the presence of 4CzIPN, HCl and acetone (Fig. 1c), forming products with other groups unchanged and obtained yields of up to 88%.²³ HCl and acetone were indispensable for the reaction system. It is to be noted that the applicability within the chemical industry would be constrained by challenges related to poor selectivity, difficulty in recycling homogeneous catalysts and limited substrate scope. Therefore, the continued exploration for the development of efficient heterogeneous catalytic protocols for the aerobic oxidative debenzylation is of paramount importance.

Fig. 1 Oxidations of N-benzylated amines with molecular oxygen as the sole oxidant. (a) Debenzylations mediated by t-BuOK; (b) photocatalytic debenzylations of primary and secondary amines; (c) photocatalytic oxidations of tertiary amines; (d) this work.

Graphitic carbon nitride (g-C₃N₄) is an appealing metal-free material due to its thermal and chemical stability, nontoxicity, low cost, and visible light optical response (the band absorption edge is about 460 nm).^24,25 O₂ tends to be adsorbed on the surface of g-C₃N₄, thus promoting the electron transfer process to produce O₂˙⁻ radicals to participate in the photocatalytic redox reactions.²⁶ The unique band structure of g-C₃N₄ facilitates the generation of photogenerated electrons and holes through electron migration when illuminated with visible light, making it highly favourable for reactions involving electron transfer processes, such as Sakurai reactions, allylations,²⁷ dehydrogenations,²⁸ and oxidations of alcohols,²⁹ but its application to debenzylations remains unexplored. Compared with other metal oxide semiconductor materials, such as TiO₂,^30,31 CeO₂,^32,33 RuO₂³⁴ and ZnO,³⁵ g-C₃N₄ exhibits excellent resistance to strong acids, thus ensuring the recovery and reuse of the catalyst in reactions mediated by HCl.

In this paper, a series of g-C₃N₄ or metal-doped g-C₃N₄ with different morphologies were prepared and applied to photocatalytic N-debenzylations with molecular oxygen as the sole oxidant (Fig. 1d). The solid-state g-C₃N₄ exhibits excellent resistance to HCl, thereby enabling the recycled catalyst to retain both high activity and a superior recovery rate. Both experiments and DFT calculations were conducted to reveal the detailed debenzylation mechanisms of both tertiary amines and primary/secondary amines. The phenomenon that HCl promotes the debenzylations of tertiary amines, but not primary/secondary amines, was explained from the view of DFT. This approach not only broadens the application scope of g-C₃N₄ in photocatalytic organic reactions, but also holds greater potential for industrial applications in the future.

2 Experiments and calculations

2.1 Preparation of catalysts

CN-M. CN-M was prepared according to Xiao's work.³⁶ Melamine (20 g, 159 mmol) in a ceramic crucible was heated to 550 °C at a rate of 2.3 °C min⁻¹ in a muffle furnace under an air atmosphere. The temperature was maintained at 550 °C for 4 hours and CN-M was obtained after cooling to room temperature.

CN-U. CN-U was prepared according to Zhang's work.³⁷ Urea (20 g, 333 mmol) in a ceramic crucible was heated to 550 °C at a rate of 2.3 °C min⁻¹ in a muffle furnace under an air atmosphere. The temperature was maintained at 550 °C for 4 hours and CN-U was obtained after cooling to room temperature.

CN-T. CN-T was prepared according to Wang's work.³⁸ The mixture of melamine (1.0 g, 8 mmol) and urea (12.0 g, 200 mmol) was fully ground with a mortar and then heated to 550 °C at a rate of 2.3 °C min⁻¹ in a muffle furnace under an air atmosphere. The temperature was maintained at 550 °C for 4 hours and CN-T was obtained after cooling to room temperature.

CN-G. CN-G was prepared according to Jun's work.³⁹ Melamine (0.5 g, 4 mmol) and cyanuric acid (0.51 g, 4 mmol) were dissolved in 20 and 10 mL DMSO, respectively. After sonication for 30 min to dissolve them completely, the two solutions were mixed together to obtain a white precipitate. The precipitate was filtered, washed with water and ethanol, dried at 50 °C, and then calcined at 550 °C for 4 h with a heating rate of 2.3 °C min⁻¹.

CN-F. CN-F was prepared according to Yu's work.⁴⁰ Melamine (0.5 g, 4 mmol), and a mixture of cyanuric acid (0.51 g, 4 mmol) and oxalic acid (0.1 g, 1 mmol) were dissolved in 20 and 10 mL DMSO, respectively. After sonication for 30 min to dissolve them completely, the two solutions were mixed together to obtain a white precipitate. The precipitate was filtered, washed with water and ethanol, dried at 50 °C and then calcined at 550 °C for 4 h with a heating rate of 2.3 °C min⁻¹.

TiO₂-F. TiO₂-F was prepared according to Yu's work.⁴¹ Potassium titanate oxalate (0.8 g, 2.2 mmol) and urea (2.4 g, 40 mmol) were added to a mixture of diethyleneglycol (60 ml) and deionized water (20 ml). The mixture was then sealed and stirred for 2 hours. After stirring, the dispersed mixture was loaded into a 100 mL hydrothermal reactor and reacted at 180 °C for 12 hours. Finally, TiO₂-F was obtained by centrifugation, washed with anhydrous ethanol and deionized water, and vacuum dried overnight at 70 °C.

CN-M-Fe. CN-M-Fe was prepared according to Ahmad's work.⁴² The catalyst was synthesized with the incipient wetness impregnation method using FeCl₃ as a Fe source and CN-M as support. FeCl₃ (20 mg, 0.12 mmol) and CN-M (1.0 g) were dissolved into 20 mL of deionized water. The mixture was stirred vigorously at room temperature for 2 h, followed by stirring and heating at 80 °C until the water almost completely evaporates. The sample was dried at 50 °C and then calcined at 550 °C for 4 h with the heating rate of 2.3 °C min⁻¹.

CN-M-Ni. CN-M-Ni was prepared according to Ahmad's work.⁴² The catalyst was synthesized with the incipient wetness impregnation method using Ni(NO₃)₂·6H₂O as a Ni source and CN-M as support. Ni(NO₃)₂·6H₂O (20 mg, 0.07 mmol) and CN-M (1.0 g) were dissolved into 20 mL of deionized water. The mixture was stirred vigorously at room temperature for 2 h, followed by stirring and heating at 80 °C until the water almost completely evaporates. The sample was dried at 50 °C and then calcined at 550 °C for 4 h with the heating rate of 2.3 °C min⁻¹.

2.2 General photocatalytic N-debenzylation

An oven-dried Schlenk tube (25 mL) equipped with a stirring bar, gas balloon and an inner light resource was charged with 1 (0.2 mmol, 1.0 equiv.), catalyst (5 mg), HCl (37%, 25 μL, 0.3 mmol) and acetone (2 mL). The light source (415 nm, 3 W) was 4 centimeters higher than the liquid level, with an intensity of about 8192 Lx. The reaction mixture in a sealed tube was stirred at 25 °C for about 15 minutes and then reacted under light for 1–72 h. The yield was obtained via column separation, high performance liquid chromatography (HPLC) or gas chromatography-mass spectrometry (GC-MS). The light intensity for different light resources is summarized in Table S1.†

2.3 DFT calculations

All calculations were carried out with Gaussian 16.⁴³ Geometry optimization and vibrational frequencies for all structures of reductants, products, transition states and intermediates were performed using the M06-2x hybrid meta functional^44,45 with a basis set of 6-311 + G(d,p) for all atoms. Acetone was used as a solvent and the self-consistent reaction field (SCRF) polarizable continuum model (PCM) model was applied to calculate the solvation energy corrections. Only one imaginary frequency corresponding to the reaction coordinates was found for the transition states. The more accurate electron energies were obtained from the frequency calculations at a level of M06-2x/def2tzvp.

3 Results and discussion

3.1 Characterization of catalysts

Fig. 2 depicts the SEM images of different catalysts, including CN-M, CN-U, CN-T, CN-G, CN-F and TiO₂-F. CN-M, obtained through calcination of melamine, exhibited a typical block structure with particle sizes above 10 μm (Fig. 2a). CN-U, which was calcined from urea, displayed an irregular and unformed pore-like structure (Fig. 2b). The size of CM-U is significantly smaller than that of the block shaped CN-M. Remarkably, CN-T (Fig. 2c) with a hollow tubular structure, was successfully synthesized from a blend of melamine and urea. These nanotubes boast a diameter of 0.35 μm and a length of 2.1 μm. The formation of this hollow tubular structure is intriguing, as it results from the transformation of urea to cyanuric acid near 400 °C,³⁸ followed by the self-assembly of cyanuric acid with melamine to form a supramolecular structure. Contrastingly, the globular form of CN-G (Fig. 2d) with an average particle size of 3.73 μm was prepared using melamine and cyanuric acid as key materials. Compared to CN-G, CN-F (Fig. 2e) with a wreath-like structure was prepared using the same amount of melamine and cyanuric acid, but mediated by oxalic acid. Additionally, floral-like TiO₂-F was also successfully prepared, exhibiting a diameter of 2.68 μm (Fig. 2f). Further information related to the specific surface area was determined with the multi-point BET method. As a result, the specific surface area decreased in the order of CN-G (122.639 m² g⁻¹) < CN-U (125.839 m² g⁻¹) < CN-F (133.748 m² g⁻¹) < CN-T (136.330 m² g⁻¹).

Fig. 2 SEM images of CN-M (a), CN-U (b), CN-T (c), CN-G (d), CN-F (e), and TiO₂-F (f).

To deeply understand the properties of carbon nitrides with different morphologies, X-ray diffraction (XRD), Fourier transform infrared (FT-IR), UV visible diffuse reflectance (UV-vis DR) and photoluminescence (PL) spectroscopy were employed to analyze the microstructure and optical properties. As depicted in the XRD patterns in Fig. 3a, all five carbon nitrides, despite their differing morphologies, exhibit the same identical characteristic peaks at 2θ angles of 13° and 27°, indicating no damages to the essential structural units of g-C₃N₄. The diffraction peaks observed at 2θ angles of 13° and 27° correspond to the (100) crystal plane created by the s-triazine unit⁴⁶ (JCPDS Card 87-1526) and the interlayer stacked (002) crystal plane of the conjugated aromatic system⁴⁷ (JCPDS Card 87-1526), respectively. The diffraction peaks of CN-T, CN-G, CN-F, and CN-U at a 2θ angle of 13° are much weaker and wider, indicating porous structures. The diffraction peaks at a 2θ angle of 27° for CN-T, CN-G, CN-F, and CN-U are significantly weaker, suggesting a decrease in CN crystallinity and thin layer structures.⁴⁸ The FT-IR spectra in Fig. 3b reveal that the five samples have analogous chemical structures. The characteristic peak at 810 cm⁻¹ corresponds to the respiratory mode of the triazine structure.⁴⁹ The characteristic peaks at 1200–1700 cm⁻¹ belong to typical vibrations of aromatic triazine heterocycles (C–N and C=N),⁵⁰ whereas peaks between 3000–3600 cm⁻¹ correspond to stretching vibrations of the N–H group.⁵¹

Fig. 3 (a) XRD patterns, (b) FT-IR patterns, (c) UV-vis DR spectra, and (d) PL spectra of CN-M, CN-U, CN-T CN-G and CN-F.

Based on the UV-vis DR spectra depicted in Fig. 3c, it is evident that all carbon nitrides, regardless of their various morphologies, exhibit a characteristic semiconductor absorption across a broad wavelength range from UV to visible light. CN-G and CN-F display noticeable red shifts and possess a broadened visible light spectral absorption range extending up to 800 nm, indicating significant improvements in optical absorption capacity.⁵² This could be ascribed to the fact that the increased specific surface area leads to an increased number of active sites, thereby enhancing the optical absorption capacity and hence the catalytic performance of CN-G and CN-F. The hollow and one-dimensional (1D) structures of CN-T facilitate the migration of charge carriers along the 1D length, thereby enhancing effective carrier separation. Unexpectedly, CN-T shows no significant advantages in the spectral absorption range, but only slightly higher absorbance at 250–430 nm. Photoluminescence spectroscopy was performed to evaluate the effect of morphology on the recombination of electron–hole pairs. As shown in Fig. 3d, all of the five carbon nitrides exhibit significant emission signals between 400–600 nm (excitation wavelength of 390 nm). The PL emission intensity decreased significantly in the order of CN-F < CN-G < CN-T ≤ CN-U < CN-M, indicating that significant enhancement in photoelectron separation and transfer rate was realized through the modulation of catalyst morphology. CN-U and CN-T have significantly improved photoelectron separation/transfer rate compared with CN-M, but slightly lower than CN-G and CN-F. Combined with UV-vis DRS analysis, CN-F shows the best performance on both light absorbance and photoelectron separation/transfer rate.

3.2 Optimization of N-debenzylations

N-Benzylcarbazole (1a) was chosen as the model substrate for photocatalytic debenzylations as Fig. 4a shown. 75% isolated yield of 2a was obtained after 1.5 h at 25 °C under 3 W 415 nm LED irradiation in the presence of CN-M (5 mg), HCl (1.5 equiv.) and air (Fig. 4b). The yield is significantly higher than that without adding catalyst (28%). Among the various morphologies of carbon nitrides, CN-U, CN-G, CN-F, and CN-T, with large surface areas (125.839–136.333 m² g⁻¹), exhibited excellent yields ranging from 80–82% (Fig. 4b). CN-T showed a slightly higher activity, achieving a yield of 82%. These findings contrast with the UV-vis DRS and PL analysis, which indicated that CN-F possesses superior performance in both light absorbance and photoelectron separation/transfer. This intriguing phenomenon can be attributed to the tubular structure with a big cavity, which exhibits a higher surface area and provides more active sites. In contrast, bulk CN-M demonstrated a lower yield of only 75%. In addition, the incorporation of metals (Fe or Ni) into CN-M materials significantly impeded the debenzylation of 1a, resulting in lower yields of only 63–67% (Fig. 4b). Flower-like TiO₂-F also gave 2a in a yield of 79%. However, TiO₂ is unstable in acidic environments, limiting its application to a certain extent. Both decreasing and increasing the catalyst dosage resulted in a decrease in yields (Fig. 4b, 75–76%). A 75% yield was achieved by reducing the catalyst dosage to 3 mg, which can be attributed to the decrease in the number of active sites. Conversely, increasing the catalyst dosage (7 mg) restricted light transmittance, ultimately leading to a decrease in yield (76%). The wide spectral absorption range of CN-T enables it to achieve high catalytic activity in the wavelength range of 254–450 nm, resulting in similar yields of 80–82% (Fig. 4c). Instead, only 25–46% yields were obtained in the wavelength range of 550–660 nm. The results were consistent with its UV-vis spectra results in Fig. 3c. Despite testing various acids such as formic acid, HOAc, H₂SO₄ and TsOH, only HCl was found to effectively promote the debenzylations. When 1.5 equiv. of HCl was added, the yield of 2a reached its peak (Fig. 4d, 82%). Both increasing and decreasing the amount of HCl led to a yield decrease (Fig. 4d). Gases with different oxygen contents (21%, 14%, 11%, 7%, 5%, 0%) were prepared by mixing nitrogen and air (V_air : V_N2 = 1 : 0, 2 : 1, 1 : 1, 1 : 2, 1 : 3, 0 : 1). Gratifyingly, a yield of 92% (90% of isolated yield) was achieved within 3 h using a 7% content of oxygen as oxidant (Fig. 4e). The yields in different oxygen contents indicate that high concentration of oxygen is not conducive to debenzylations. Of course, too low concentration of oxygen can also lead to a long reaction time or even no reactions.

Fig. 4 Reaction optimizations. (a) Debenzylation of N-benzylcarbazole (1a). (b) Catalyst screening. (c) Light resource screening; (d) optimization for the usage of HCl; (e) optimization for the content of O₂. General reaction conditions: 1a (0.2 mmol), acid, catalyst in solvent (2 mL) under 3 W 415 nm LEDs at 25 °C for 1.5 h. Yield was obtained by external standard method through high performance liquid chromatography (HPLC).

3.3 Investigation of substrate scope

With the optimal reaction conditions in hand, the substrate scope was explored and the results are shown in Tables 1 and 2. Notably, the types and positions of substituents on the benzene ring of the benzyl group exert a profound influence on debenzylations. The weakly electron-donating methyl group, positioned at various sites (1b, 1c, and 1d), exerted minimal influence on debenzylations, resulting in consistently high yields ranging from 85% to 90%. In contrast, only 34% yield was obtained for 1epara-substituted with a t-Bu group, which was known for their strong electron-donating effect. 1f and 1j, meta-substituted with weak electron withdrawing groups –OMe and –Br, gave moderate yields of 43–62%. Surprisingly, 1g bearing two –OMe groups at the 3 and 5 positions gave 2a in 84% yield, while 1k bearing two –Br groups at the 3 and 5 positions produced 2a with a moderate yield of only 57%. Fortunately, 1i ortho-substituted with a –Br group gave the highest yield of 93%. Interestingly, 1m and 1n, containing a –Cl group at the ortho- and meta-positions, exhibited a distinct trend compared to that containing a –Br group at the same positions (1i and 1j). We also observed that good yields of 80–88% could be obtained for 1d, 1h, 1l and 1o, whose benzene ring were para-substituted with –Me, –OMe, –Br and –Cl groups. In addition, lower yields (40–58%) were obtained when the para- and meta- positions are replaced by strong electron withdrawing groups, such as –OCF₃, –COOCH₃, –CN and –NO₃ (1p–1s). The substituent groups on the carbazole ring have little effect on debenzylations. Good to excellent yields (72–90%) were obtained when the carbazole ring were substituted with –Cl, –Br, –OMe, –Ph and –t-Bu (1t–1ae). N-Benzylacridone (1af) and N,N-diphenylbenzylamine (1ag), both featuring a nitrogen atom directly linked to two phenyl groups, also gave the corresponding debenzylated products in excellent yields of 81–85%, indicating good practicability. 1ah, containing a N–Et group and a phenyl group, also gave 83% yield but with a long reaction time of 72 h. Besides, no debenzylated products were detected for 1ai, 1aj and 1ak, whose N atoms were directly bonded to methyl and benzoyl groups with a typical delocalization effect.

Table 1 Substrate scope of tertiary amines

Reaction conditions: 0.2 mmol of 1 (1.0 equiv), CN-T (5 mg), HCl (1.5 equiv.) in 2 mL of acetone under 3 W 415 nm LEDs in the presence of air : N₂ (v/v = 1 : 2) mixed gas balloon. The reactions were monitored by TLC analysis and the products were obtained in isolated yields.

---

Table 2 Substrate scope of primary and secondary amines

Reaction conditions: 0.2 mmol of 1 (1.0 equiv), CN-T (5 mg) in 2 mL of acetone under 30 W 395 nm LEDs in the presence of air balloon. The yields were obtained by GC-MS using the peak-area normalization method.

---

The addition of HCl led to the formation of primary/ secondary amine salts, thus inhibiting the debenzylations of primary/ secondary amines. Therefore, the primary and secondary amines were explored in the absence of HCl under 395 nm LED irradiation (Table 2). The C–N bonds of primary amines (1al–1as) containing diverse substituents on the benzene ring undergo smooth cleavage, resulting in the formation of corresponding benzaldehydes with yields ranging from 60% to 86%, along with the liberation of NH₃. For secondary amines 1at and 1au containing –Me and –Et groups, 1au gave a higher yield than 1at (85% vs. 70%). To date, a long reaction time was required for debenzylations of primary/secondary amines. Besides, no benzaldehyde but only imines were detected for 1av and 1aw, which containing t-Bu and benzoyl groups with a strong super conjugation effect.

3.4 Catalyst recovery experiments

Catalyst recovery experiments and amplification experiments (five times) were conducted to explore the recyclability and reusability of CN-T (Fig. 5a). The results are shown in Fig. 5b. In the catalyst recovery experiments, 1a (five times, 1 mmol, 257.12 mg) was used as the substrate and undergoes debenzylation in the presence of CN-T (25 mg), HCl (1.5 equiv.), acetone (10 mL) and molecular oxygen (7%) under 30 W LEDs (395 nm) at 25 °C. CN-T was recovered by centrifugation, followed by washing with acetone and subsequent drying in vacuum at 50 °C for 4 h or calcined at 150 °C for 2 h. In the 1st and 2nd cycles, 2a was obtained in yields of only 85–86% after a reaction time of 5 h. In the 3rd cycle, the yields were slightly increased to 88% when the reaction time was prolonged to 5.5 h. The reaction rate for the 3rd cycle was slighty lower than that for the 1st/2nd cycles based on the thin layer chromatography analysis, thus requiring a longer reaction time. To further enhance the activity of the recycled CN-T, the drying temperature was adjusted to 150 °C in the 3rd–6th cycles to avoid the solvent residue problem. The catalyst recovery rate for the 3rd cycle (93%) was obviously lower than that for the 1st/2nd cycles (99% and 95%), while that for the 4th–6th cycles were maintained at the high level (97–98%). In the 4th–6th cycles, 2a was obtained in stabe yields of 89–90%. After six cycles of recovery experiments, the catalyst structure did not change significantly based on FT-IR analysis (Fig. S1†). In short, the recovered catalyst exhibits excellent durability and effectiveness.

Fig. 5 (a) The recycling experimental procedure scheme. (b) The results and important information of each recycling experiment. (c) Alternating light–dark experiment over time. (d) EPR spectrum of the mixture of 1a, CN-T and TEMP in acetone. (e) EPR spectrum of the mixture of 1a, CN-T and DMPO in methanol. (f) Control experiments for 1a and 1al with different scavengers (3 equiv.). (g) The exploration for the role of ¹O₂. (h) The exploration for the role of HCl.

3.5 Reaction mechanisms

To gain insights into the underlying mechanism of photocatalytic debenzylations with molecular oxygen, a series of control experiments and DFT calculations were performed. Fig. 5c illustrates the results of the alternating light–dark experiments, indicating that no reaction proceeds in the dark. It underscores the importance of light as the driving force for debenzylations. Only 28% of 2a was obtained in the absence of a catalyst, while no reaction occurred for 1ah under the same conditions. These results further highlight the the crucial role of catalysts in debenzylations. No product was detected in the absence of molecular oxygen (Fig. 4e), emphasizing the crucial role of molecular oxygen. 2,2,6,6-Tetramethylpiperidine (TEMP) and 5,5-dimethyl-1-pyrroline N-oxide (DMPO) were used as ¹O₂ and O₂˙⁻ quenchers in the EPR experiments. As shown in Fig. 5d and e, the EPR signals of TEMPO (ref. 53) and DMPO–OOH (ref. 54) were recorded in the presence of TEMP and DMPO, respectively. These results indicate that both ¹O₂ and O₂˙⁻ were generated in this photocatalytic system. As depicted in Fig. 5f, the addition of the radical quencher 2,2,6,6-tetramethylpiperidinooxy (TEMPO) and O₂˙⁻ quencher 2,3-dichloro-5,6-dicyano-1,4-benzoquinone (DDQ) inhibited the debenzylations of both 1a and 1al, resulting in a significant decrease in yields (16–50%). This phenomenon indicates that debenzylations proceed via a radical mechanism. The O₂˙⁻ quenching experiment for 1al seems invalid, since the addition of DDQ to the reaction system of 1al, even without light, caused a violent reaction accompanied by the production of many bubbles. Tetraphenylporphyrin (H₂TPP) is widely known to promote the formation of ¹O₂. Only 7% yield was achieved for 1a using H₂TPP as photocatalyst, while 72% yield was obtained for primary amine 1al (Fig. 5g). Besides, Che and co-workers⁵⁵ found that imines were formed from secondary amines via the ¹O₂ mechanism using H₂TPP as the photocatalyst. Therefore, ¹O₂ was unnecessary for the debenzylations of tertiary amines, but can promote that of primary/secondary amines without any doubt, despite that 1,4-diazabicyclo[2.2.2]octane (DABCO), a singlet oxygen quencher, inhibited the debenzylations of primary and tertiary amines (Fig. 5f). In addition, no product was detected for groups adding benzoquinone (BQ) as an electron scavenger, revealing the indispensable role of photoexcited electrons for debenzylations. The crucial role of HCl in the debenzylation of tertiary amines was evident, as no debenzylated product was produced when KCl or alternative acids, including formic acid (HCOOH), acetic acid (HOAc), sulfuric acid (H₂SO₄), and p-toluenesulfonic acid (TsOH) (Fig. 5h), were utilized instead. Based on the above discussion, several possible reaction routes shown in Fig. 6 were proposed.

Fig. 6 Possible reaction routes.

To further reveal the difference between the debenzylations of primary/secondary amines and that of tertiary amines, DFT calculations were performed (Fig. 7). Both primary and secondary amines have NH groups. To distinguish from our previous research²³ that uses 1a to explore the reaction mechanism, 1ah and 1al were selected to investigate the mechanisms of tertiary and primary amines, respectively. All the structures of transition states are depicted in Fig. 8. The formation of radical B-1ah is the key step for the debenzylation of 1ah. Paths I, III, IV and V were considered for the formation of B-1ah (Fig. 7a). In path I, the valence band (VB) obtains an electron from 1ah spontaneously (ΔE = −184.56 kJ mol⁻¹), resulting in the formation of A-1ah. Subsequently, a proton transfer process occurred between A-1ah and acetone viaTS1 (ΔE^‡ = 106.12 kJ mol⁻¹), forming B-1ah. In path III, a continuous increase of energy (ΔE = 101.93 kJ mol⁻¹) is observed for the reactions of 1ah with O₂˙⁻ to form B-1ah, suggesting unfeasible processes. In path IV and V, B-1ah is obtained directly with a one-step HAT process between Cl·/¹O₂ and 1ahviaTS3 (ΔE^‡ = 36.27 kJ mol⁻¹) and TS2 (ΔE^‡ = 49.27 kJ mol⁻¹). Based on the above disscussion, path IV is the most feasible route to form B-1ah with the lowest energy barrier (ΔE^‡ = 36.27 kJ mol⁻¹). The prerequisite for this pathway is the formation of the radical Cl·, which has a high requirement for the electron transfer ability of the catalyst. Subsequently, the imine cation C-1ah is formed from B-1ah by losing an electron in the VB site, simultaneously releasing energies of 344.04 kJ mol⁻¹. C-1ah is ready to be converted to the debenzylated product 2ahvia hydrolysis with an energy barrier of 146.14 kJ mol⁻¹.

Fig. 7 DFT calculations. (a) Energy profiles for debenzylation of 1ah; (b) energy profiles for debenzylation of 1al.

Fig. 8 Structures of transition states TS1–TS9.

For the transformation of 1al, a continuous increase of energy was observed for the HAT process between 1al and O₂˙⁻ (Fig. 7b, path III), as well the deprotonation process mediated by acetone (Fig. 7b, path I). In path II, C-1al is formed smoothly via two single-electron transfer (SET) processes and a deprotonation process, which are all energy decreasing processes. It is to be noted that the deprotonation process should be mediated by the organic alkaline 1al. Surprisingly, ¹O₂ is proved to effectively react with 1alviaTS6 by simultaneously extracting both the benzylic hydrogen atom and the hydrogen atom bonded to the nitrogen atom, ultimately resulting in the formation of H₂O₂ and imine E-1al (Fig. 7b, path VI). This process is highly exothermic, releasing a substantial amount of energy at 242.23 kJ mol⁻¹. Furthermore, it requires a minimal energy barrier of just 12.11 kJ mol⁻¹, suggesting a smooth and efficient reaction. In the presence of organic alkali 1al, the subsequent hydrolysis belongs to alkali-catalyzed hydrolysis. Firstly, HO⁻ attacks the carbon atom of the C–N double bond viaTS7, encountering an energy barrier of 33.45 kJ mol⁻¹, forming the anion F-1al. Subsequently, anion F-1al abstracts a H⁺ from a water molecule viaTS8 with a relatively low energy barrier of only 4.06 kJ mol⁻¹, resulting in the formation of G-1al. Finally, benzaldehyde and debenzylated amine (2al) are produced, accompanied by the breakage of C–N bonds viaTS9 with a kinetic energy barrier of 142.52 kJ mol⁻¹. The whole reaction is exergonic by 220.58 kJ mol⁻¹, suggesting a feasible process from the view of DFT calculations. The result is consistent with our experiment using H₂TPP as the photocatalyst (Fig. 5g). Based on the above discussion, both paths II and VI contributed to debenzylation of 1al. Of course, the addition of HCl will promote the protonation of 1al, thus suppressing the debenzylations.

Based on the above analysis, we can conclude that the debenzylations of N-benzyl tertiary amines proceed through path IV, whereas primary/secondary amines undergo the reactions primarily through paths II and VI. HCl plays an important role in the deprotonation processes of tertiary amines. Besides, molecule oxygen is ultimately converted into H₂O₂.

4 Conclusions

In conclusion, a recyclable metal-free mesoporous tubular carbon nitride (CN-T) was prepared and applied to the photocatalytic debenzylations with molecular oxygen as the sole oxidant. The method is suitable to a broad range of N-benzyl primary/secondary/tertiary amines, resulting in excellent yields up to 93%. In the 6 cycle catalyst recovery experiments, the average yield and catalyst recovery rate are 88% and 97%, respectively. The recovered catalyst after 6 cycles retains its original structural integrity and exhibits high catalytic activity. Both experiments and DFT calculations revealed that the debenzylation mechanisms for primary/secondary amines and tertiary amines are different. The in situ generated radical Cl· plays an important role in the HAT process, thus promoting the debenzylations of tertiary amines. For the debenzylations of primary/secondary amines, ¹O₂ could efficiently simultaneously extract two hydrogen atoms from N–H and the benzylic C–H groups to form imines, which readily undergo alkali-catalyzed hydrolysis. Undeniably, primary/ secondary amines also undergo debenzylations similar to tertiary amines. This study provides new insights for the development of greener and safer debenzylation processes.

Data availability

The data supporting this article have been included as part of the ESI.†

Conflicts of interest

There are no conflicts to declare.

Acknowledgements

This work was supported by the National Natural Science Foundation of China (52304209 and U20A20143). The authors also thank the DUT Instrumental Analysis Center and Network and Information Center of Dalian University of Technology for their support.

References

1. M. Seki, Synthesis, 2014, 46, 3249–3255.
2. S. G. Davies, R. L. Nicholson, P. D. Price, P. M. Roberts, A. J. Russell, E. D. Savory, A. D. Smith and J. E. Thomson, Tetrahedron: Asymmetry, 2009, 20, 758–772.
3. T. Farooq, L. K. Sydnes, K. W. Tornroos and B. E. Haug, Synthesis, 2012, 44, 2070–2078.
4. F. Shang, R. Liu, M. Lv, Y. Ma, J. Liu, P. Zhou, C. Zhang and K.-l. Han, ACS Omega, 2022, 7, 21912–21924.
5. M. Iftikhar, M. Zhou and Y. H. Lu, Mini-Rev. Org. Chem., 2021, 18, 1037–1048.
6. Y. Yamamoto, E. Shimizu, K. Ban, Y. Wada, T. Mizusaki, M. Yoshimura, Y. Takagi, Y. Sawama and H. Sajiki, ACS Omega, 2020, 5, 2699–2709.
7. S. A. Yakukhnov and V. P. Ananikov, Adv. Synth. Catal., 2019, 361, 4781–4789.
8. S. D. Bull, S. G. Davies, A. W. Mulvaney, R. S. Prasad, A. D. Smith and G. Fenton, Chem. Commun., 2000, 337–338.
9. K. Moriyama, Y. Nakamura and H. Togo, Org. Lett., 2014, 16, 3812–3815.
10. R. Csuk and P. Dorr, Tetrahedron, 1994, 50, 9983–9988.
11. X. F. Tang, R. Zhu, T. J. Shi, Y. Wang, X. C. Niu, Y. Zhang, J. C. Zhu, W. Li, W. P. Hu and R. Q. Xu, Materials, 2022, 15, 409.
12. K. J. Liang, X. P. Li, D. L. Wei, C. H. Jin, C. W. Liu and C. F. Xia, Chem, 2023, 9, 511–522.
13. R. Gigg and R. Conant, J. Chem. Soc., Chem. Commun., 1983, 465–466.
14. A. A. Haddach, A. Kelleman and M. V. Deaton-Rewolinski, Tetrahedron Lett., 2002, 43, 399–402.
15. X. L. Wang, J. Y. Tian, X. C. Guo, F. Q. Zhang, L. F. Liang and X. M. Zhang, Chem. – Eur. J., 2021, 3753–3760.
16. C. P. Xu, P. R. Anusuyadevi, C. Aymonier, R. Luque and S. Marre, Chem. Soc. Rev., 2019, 48, 3868–3902.
17. J. G. Yu, T. R. Zhang and N. Q. Wu, Sol. RRL, 2021, 5, 2100037.
18. K. S. Schanze, P. V. Kamat, P. D. Yang and J. Bisquert, ACS Energy Lett., 2020, 5, 2602–2604.
19. R. Lechner and B. König, Synthesis, 2010, 1712–1718.
20. N. Iqbal and E. J. Cho, Adv. Synth. Catal., 2015, 357, 2187–2192.
21. N. Neerathilingam, M. Bhargava Reddy and R. Anandhan, J. Org. Chem., 2021, 86, 15117–15127.
22. N. Neerathilingam and R. Anandhan, RSC Adv., 2022, 12, 8368–8373.
23. Y. Wu, J. Kang, H. Zhu, M. Bi, J. Li, Q. Meng, X. Lyu and Z. Wu, ACS Sustainable Chem. Eng., 2024, 12, 6640–6647.
24. S. M. Ruban, K. Ramadass, G. Singh, S. N. Talapaneni, G. Kamalakar, C. R. Gadipelly, L. K. Mannepalli, Y. Sugi and A. Vinu, Sci. Technol. Adv. Mater., 2023, 24, 2188879.
25. Y. Lei, J.-F. Huang, X.-A. Li, C.-Y. Lv, C.-P. Hou and J.-M. Liu, Chin. J. Catal., 2022, 43, 2249–2258.
26. M. X. Ran, J. R. Li, W. Cui, Y. H. Li, P. D. Li and F. Dong, Catal. Sci. Technol., 2018, 8, 3387–3394.
27. L. Möhlmann and S. Blechert, Adv. Synth. Catal., 2014, 356, 2825–2829.
28. L. Cao, P. Dai, L. Zhu, L. Yan, R. Chen, D. Liu, X. Gu, L. Li, Q. Xue and X. Zhao, Appl. Catal., B, 2020, 262, 118277.
29. S. Verma, R. B. N. Baig, M. N. Nadagouda and R. S. Varma, ACS Sustainable Chem. Eng., 2016, 4, 1094–1098.
30. H. J. Cheng and W. T. Xu, Org. Biomol. Chem., 2019, 17, 9977–9989.
31. Q. Guo, C. Zhou, Z. Ma and X. Yang, Adv. Mater., 2019, 31, 1901997.
32. M. M. Khan, S. N. Matussin and A. Rahman, Bioprocess Biosyst. Eng., 2022, 45, 619–645.
33. M. Yang, G. Shen, Q. Wang, K. Deng, M. Liu, Y. Chen, Y. Gong and Z. Wang, Molecules, 2021, 26, 6363.
34. L. Wu, Q. Liang, J. Zhao, J. Zhu, H. Jia, W. Zhang, P. Cai and W. Luo, Chin. J. Catal., 2023, 55, 182–190.
35. Y. Chen, S. Deng, N. Xu and J. Chen, Nanomaterials, 2021, 11, 2150.
36. Y. Xiao, G. Tian, W. Li, Y. Xie, B. Jiang, C. Tian, D. Zhao and H. Fu, J. Am. Chem. Soc., 2019, 141, 2508–2515.
37. J. Zhang, X. Chen, K. Takanabe, K. Maeda, K. Domen, J. D. Epping, X. Fu, M. Antonietti and X. Wang, Angew. Chem., Int. Ed., 2010, 49, 441–444.
38. X. Wang, C. Zhou, R. Shi, Q. Liu, G. I. N. Waterhouse, L. Wu, C.-H. Tung and T. Zhang, Nano Res., 2019, 12, 2385–2389.
39. Y.-S. Jun, E. Z. Lee, X. Wang, W. H. Hong, G. D. Stucky and A. Thomas, Adv. Funct. Mater., 2013, 23, 3661–3667.
40. W. Yu, T. Zhang and Z. Zhao, Appl. Catal., B, 2020, 278, 119342.
41. X. Yu, Z. Zhao, J. Zhang, W. Guo, L. Li, H. Liu and Z. L. Wang, CrystEngComm, 2017, 19, 129–136.
42. K. N. Ahmad, S. A. Anuar, W. N. R. Wan Isahak, M. I. Rosli and M. A. Yarmo, ACS Appl. Mater. Interfaces, 2020, 12, 7102–7113.
43. M. J. Frisch, G. W. Trucks, H. B. Schlegel, G. E. Scuseria, M. A. Robb, J. R. Cheeseman, G. Scalmani, V. Barone, G. A. Petersson, H. Nakatsuji, X. Li, M. Caricato, A. V. Marenich, J. Bloino, B. G. Janesko, R. Gomperts, B. Mennucci, H. P. Hratchian, J. V. Ortiz, A. F. Izmaylov, J. L. Sonnenberg, D. Williams-Young, F. Ding, F. Lipparini, F. Egidi, J. Goings, B. Peng, A. Petrone, T. Henderson, D. Ranasinghe, V. G. Zakrzewski, N. R. J. Gao, G. Zheng, W. Liang, M. Hada, M. Ehara, K. Toyota, R. Fukuda, J. Hasegawa, M. Ishida, T. Nakajima, Y. Honda, O. Kitao, H. Nakai, T. Vreven, K. Throssell, J. A. Montgomery, J. E. P. Jr, F. Ogliaro, M. J. Bearpark, J. J. Heyd, E. N. Brothers, K. N. Kudin, V. N. Staroverov, T. A. Keith, R. Kobayashi, J. Normand, K. Raghavachari, A. P. Rendell, J. C. Burant, S. S. Iyengar, J. Tomasi, M. Cossi, J. M. Millam, M. Klene, C. Adamo, R. Cammi, J. W. Ochterski, R. L. Martin, K. Morokuma, O. Farkas, J. B. Foresman and D. J. Fox, Gaussian 16, Revision B.01, Gaussian, Inc, Wallingford CT, 2016.
44. Y. Zhao and D. G. Truhlar, Theor. Chem. Acc., 2008, 120, 215–241.
45. A. Castro-Alvarez, H. Cameros, D. Sánchez and J. Vilarrasa, J. Org. Chem., 2015, 80, 11977–11985.
46. S. Das, K. Murugesan, G. J. Villegas Rodríguez, J. Kaur, J. P. Barham, A. Savateev, M. Antonietti and B. König, ACS Catal., 2021, 11, 1593–1603.
47. J. Kroeger, A. Jimenez-Solano, G. Savasci, V. W. h. Lau, V. Duppel, I. Moudrakovski, K. Kuester, T. Scholz, A. Gouder, M.-L. Schreiber, F. Podjaski, C. Ochsenfeld and B. V. Lotsch, Adv. Funct. Mater., 2021, 31, 2102468.
48. H. Zhu, J. Zhao, C. Ma, Z. Yu, J. Li and Q. Meng, ACS Appl. Mater. Interfaces, 2023, 15, 59280–59295.
49. X. Yang, Z. Tian, Y. Chen, H. Huang and J. Hu, Int. J. Hydrogen Energy, 2020, 45, 12889–12902.
50. J. Yang, Y. Liang, K. Li, G. Yang, K. Wang, R. Xu and X. Xie, Catal. Sci. Technol., 2019, 9, 3342–3346.
51. X. Zhou, H. Liu, S. Liu, L. Zhang, T. Wang, C. Wang and D. Su, Fuel, 2023, 332, 126147.
52. M. Tipplook, G. Panomsuwan, T. Sudare and K. Teshima, ACS Appl. Nano Mater., 2022, 5, 3422–3433.
53. G. Nardi, I. Manet, S. Monti, M. A. Miranda and V. Lhiaubet-Vallet, Free Radicals Biol. Med., 2014, 77, 64–70.
54. G. M. Rosen, A. Beselman, P. Tsai, S. Pou, C. Mailer, K. Ichikawa, B. H. Robinson, R. Nielsen, H. J. Halpern and A. D. MacKerell, J. Org. Chem., 2004, 69, 1321–1330.
55. G. Jiang, J. Chen, J.-S. Huang and C.-M. Che, Org. Lett., 2009, 11, 4568–4571.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: https://doi.org/10.1039/d4cy00981a

---