The selective oxidation of alcohols is widely recognized as one of the most fundamental transformations in both laboratory and industrial synthetic chemistry because the corresponding carbonyl compounds serve as important and versatile intermediates for the synthesis of fine chemicals ^{[1, 2, 3, 4]}. Traditionally, non-catalytic methods with stoichiometric, toxic, corrosive and expensive oxidants such as hypochlorite, dichromate, permanganate, and peroxy acids have been widely used for selective oxidation of alcohols ^[5]. From both economic and environmental viewpoints, the development of effective alcohol oxidation processes that use clean, inexpensive and atom-efficient oxidants, such as molecular oxygen (O₂) ^{[6, 7, 8]}, to convert alcohols to carbonyl compounds under mild conditions (25 °C, 1 atm O₂) is an attractive prospect. In the past decade, extensive research to develop active catalysts for O₂ activation has produced a number of catalytic systems, most of which contain metals including Ru ^{[9, 10, 11]}, Pd ^{[12, 13, 14, 15, 16, 17, 18]}, Cu ^{[19, 20]}, and Ti ^{[21, 22, 23]}. Recently, increasing attention has been paid to the development of metal-free catalysts for oxidation processes, mainly to minimize environmental issues.

Carbon nitrides are fascinating materials that have attracted worldwide attention ^{[24, 25, 26, 27, 28, 29]}. Mesoporous graphitic carbon nitride (mpg-C₃N₄) is especially attractive as a heterogeneous catalyst because of its accessible porous framework with a large surface area ^{[30, 31, 32, 33, 34]}. The most active system previously identified in mpg-C₃N₄ is in fact presumed to be a defective N-bridged “poly(tri-s-triazine)” that forms π-conjugated planar layers. These π-bonded planar C-N-C-layers with uncondensed amino groups possess a well-developed semiconductor band structure and can participate in H-N or H-π interactions, either as a donor (valence band with the π*-orbital) or acceptor (conduction band with the π-orbital). Such interactions are stronger than dissipative interactions and are expected to be able to anchor substrates or products and increase the degree of selectivity via orientation and pattern interaction, which makes mpg-C₃N₄ suitable as a catalyst or catalyst support with active bonding sites ^[35]. In addition, because of its appropriate band gap of 2.7 eV, mpg-C₃N₄ can activate O₂ by a one-electron transfer reaction under visible-light irradiation ^{[36, 37, 38]}. For example, under illumination, mpg-C₃N₄ exhibited catalytic activity towards the oxidation of benzyl alcohol, albeit under relatively harsh conditions (100 °C, 8 bar O₂ and high-intensity visible light) ^[39]. Therefore, the active surface structure together with the appropriate band gap make mpg-C₃N₄ a promising candidate to serve as a visible-light one-electron transfer starter for oxygenation. This inspired us to consider that a combination of mpg-C₃N₄ and organocatalyst may provide an effective catalytic system for alcohol oxidation under mild conditions.

In recent years, N-hydroxyphthalimide (NHPI) has been used as a valuable catalyst for the efficient aerobic oxidation of various organic compounds in the presence of co-catalysts, most of which were metallic compounds ^{[40, 41, 42, 43, 44, 45, 46, 47, 48]}. More recently, we developed a photocatalytic system consisting of bulk carbon nitride (g-C₃N₄) and NHPI for the allylic oxidation of C‒H bonds ^[49]. Compared with g-C₃N₄ (S_BET = 8 m²/g), mpg-C₃N₄ (S_BET = 190 m²/g) can in principle markedly enhance the photochemical properties of such a catalyst system because of its increased surface area and multiple scattering effects compared with g-C₃N₄. Here, we report a mpg-C₃N₄/NHPI system that behaves as a non-metal photocatalytic system with excellent activity towards the oxidation of alcohol under mild conditions (25 °C, 1 atm O₂).

Unless otherwise stated, all solvents and chemicals were of analytical grade and used without further treatment. NHPI (99%) was provided by Zhejiang Nhu Co., Ltd, China. Benzoyl peroxide (99%), anthraquinone (99%) and cyclohexanol (>99%) were purchased from Sinopharm Chemistry. Co., Ltd. Benzyl alcohol (BA; >99%), benzaldehyde (>99%) and acetonitrile (>99.5%) were purchased from Shanghai Lingfeng Chemical Reagent Co., Ltd. 4-Methylbenzyl alcohol (98%), 4-methoxybenzyl alcohol (98%), 4-nitrobenzyl alcohol (98%), benzhydrol (99%) and 4,4′-dichlorobenzhydrol (98%) were purchased from Aladdin Chemistry Co., Ltd, China. We synthesized mpg-C₃N₄ as reported previously ^[50]. All gas chromatography (GC) experiments were carried out and recorded using a Shimadzu GC-2010 chromatograph with a flame ionization detector. The structures of products and by-products were identified using an HP6890 GC/MS spectrometer by comparing retention times and fragmentation patterns with those of authentic samples. A household w-filament bulb (250 W, cost of ~$1) with a 420-nm cut-off filter was used as a visible-light source to irradiate the reaction system.

In a typical oxidation experiment, substrate (10 mmol), catalyst (as described in Table 1) and acetonitrile (20 mL) were added to a 25-mL three-neck round bottom glass reactor, which was fitted with a magnetic stirrer and O₂ inlet tube. The reaction was performed at 25 °C in a water bath with magnetic stirring. O₂ was introduced into the reactor at a constant flow rate (5 mL/min). A household w-filament bulb (250 W) and 420-nm cut-off filter were placed ~10 cm from the reactor as a visible-light source. To carry out the reaction in the dark, the glass reactor was covered with aluminum foil. After completion of the reaction, toluene was added to the mixture as an internal standard. Then, mpg-C₃N₄ was removed by filtration and the reaction mixture was injected into the GC for analysis.

Table 1 Results for the oxidation of BA with mpg-C3N4 and NHPI. Entry Catalyst a Conversion (%) Selectivity (%) C3N4 >N-OH -CHO -COOH 1 CN 100 mg NHPI 10% 99 7 92 2 CN 100 mg — 1 98 — 3 — NHPI 10% 2 98 — 4 g-CN 100 mg NHPI 10% 20 96 — 5 b CN 100 mg NHPI 10% 99 8 90 6 c CN 100 mg NHPI 10% 98 6 86 7 d CN 100 mg NHPI 10% 99 10 87 8 CN 100 mg NHPI 5% 44 98 — 9 CN 200 mg NHPI 10% 85 82 13 Reaction conditions:BA (10 mmol),acetonitrile (20 mL),1 atm O2,25℃,W-bulb (250 W) with a cut-off filter>420 nm as the visible-light source,28 h.a CN:mpg-C3N4,g-CN:g-C3N4.b Second run to test the reusability of CN.c Third run to test the reusability of catalyst.d Fourth run to test the reusability of catalyst.

Table 1 Results for the oxidation of BA with mpg-C3N4 and NHPI.

We tested the catalytic activity of the photooxidation system consisting of mpg-C₃N₄ and NHPI using BA as a probe molecule. All reactions were performed at room temperature and ambient pressure using a household W-bulb (250 W, with a cut-off filter > 420 nm) as the visible-light source. Air or O₂ was introduced to the reaction systems as the ultimate oxidant.

The conversion of BA reached 99%, while the selectivity toward benzoic acid was 92% for the mpg-C₃N₄/NHPI system under visible-light illumination (Table 1, entry 1). In contrast, only very low conversion of BA was observed in the absence of either mpg-C₃N₄ or NHPI (Table 1, entries 2 and 3) under the same reaction conditions. These results indicate that mpg-C₃N₄ and NHPI are indeed coupled to form an effective catalyst system for the aerobic oxidation of BA. Notably, the use of mpg-C₃N₄ leads to considerably enhanced conversion compared with that of the g-C₃N₄/NHPI system (Table 1, entry 4). The enhanced catalytic activity of the mpg-C₃N₄/NHPI system compared with that of g-C₃N₄/NHPI could be attributed to the better interaction between NHPI and carbon nitride materials promoting the formation of phthalimide N-oxyl (PINO) radicals. The oxidation afforded only 14% conversion of BA when it was carried out in the dark. These results reveal that visible light was important for the activity of this catalytic system, particularly for mpg-C₃N₄. As a semiconductor material, mpg-C₃N₄ is considered to act an electron-transfer mediator under visible-light illumination in the current system.

As a heterogeneous catalyst, mpg-C₃N₄ can be easily recovered by filtration, washed and then dried at 100 °C. The conversion and selectivity of mpg-C₃N₄/NHPI containing mpg-C₃N₄ used in three consecutive runs were virtually identical to those in the initial experiment with fresh mpg-C₃N₄, which is a prerequisite for practical application (Table 1, entries 5-7). In previous publications, several non-metal co-catalysts have been combined with NHPI as active systems for oxidation ^[51]. Compared with these systems ^{[52, 53, 54]}, the mpg-C₃N₄/NHPI system exhibits promising catalytic activity (Table 2).

Table 2 Results for the oxidation of BA with different catalysts. Entry Co-catalyst Conversion (%) Selectivity (%) -CHO -COOH 1 AQ 1% 7 93 5 2 BPO 1% 6 100 — 3 CN 100 mg 99 7 92 Reaction conditions: BA (10 mmol), NHPI (10%), acetonitrile (20 mL), 1 atm O2, 25 °C, W-bulb (250 W) and cut-off filter >420 nm as the visible-light source, 28 h. CN: mpg-C3N4, AQ: anthraquinone, BPO: benzoyl peroxide.

Table 2 Results for the oxidation of BA with different catalysts.

To clarify the roles of mpg-C₃N₄ and NHPI in this system, we performed some comparative experiments. When the molar amount of NHPI was halved (5 mol%), less BA was oxidized (44%) and benzaldehyde was formed with 98% selectivity (Table 1, entry 8), which indicates that benzaldehyde is a resolvable intermediate of the oxidation process. Adding twice the amount of mpg-C₃N₄ (200 mg) also gave benzaldehyde with a selectivity of 82% instead of BA as main product, but the conversion decreased to 85% (Table 1, entry 9). These results suggest that mpg-C₃N₄ plays an important role in the selectivity for benzaldehyde. Therefore, we analyzed the effect of mpg-C₃N₄ on the selectivity for benzaldehyde more systematically.

As shown in Fig. 1, increasing the amount of mpg-C₃N₄ leads to higher selectivity for benzaldehyde. Under harsh conditions, benzaldehyde is a reactive intermediate and is easily oxidized further to BA. However, our present catalytic system works at room temperature, and thereby has a wide operation window for selective oxidation. The preferred selectivity for aldehydes should also be related to the different abilities of products to adsorb on mpg-C₃N₄. Fig. 2 reveals that the amount of BA adsorbed by mpg-C₃N₄ at equilibrium is about three times that of benzaldehyde, possibly because of the formation of O-H...N or O-H...π interactions, as described in a previous report ^[35]. It seems reasonable that the adsorbed BA is easily oxidized to benzaldehyde by the active species on the surface of mpg-C₃N₄. The produced benzaldehyde with a weaker H-bridge donor than that of BA would then be replaced by a more strongly binding BA molecule, avoiding over-oxidation to some degree.

Fig. 1. Plot of the selectivity for benzaldehyde against the conversion of BA for the oxidation of BA catalyzed by different ratios of mpg-C3N4 (mg) and NHPI (mmol).

Fig. 2. Effect of contact time on the adsorption of BA (1) and benzaldehyde (2) on mpg-C3N4. Experiment conditions: mpg-C3N4 (30 mg), BA or benzaldehyde (0.1 mmol) solution in CH3CN (10 mL), 20 °C.

Scheme. 1. A possible reaction mechanism for the catalytic cycle of the mpg-C3N4/NHPI system.

To further understand the reaction cycle of the mpg-C₃N₄/NHPI system, we investigated its catalytic mechanism (Scheme 1). Under visible-light illumination, irradiated electrons from the conduction band of mpg-C₃N₄ reduced O₂ to the ^•O₂⁻ radical, because the signals for the DMPO-^•O₂⁻ adduct were detected in electron spin resonance (ESR) spectra (Fig. 3, L1). Under the conditions used, the produced ^•O₂⁻ radical cannot induce alcohol oxidation by H abstraction because only trace product (1%) was observed for BA oxidation catalyzed by mpg-C₃N₄ alone ^[39]. The ESR signals for the DMPO-^•O₂⁻ adduct showed only minor change after adding BA to the mpg-C₃N₄ system (Fig. 3, L2). However, the intensity of these signals clearly decreased after adding NHPI to mpg-C₃N₄, which was attributed to the reaction between the ^•O₂⁻ radical and NHPI (Fig. 3, L3). Based on these results, it can be concluded that the ^•O₂⁻ radical can abstract the O-H hydrogen from NHPI to generate the PINO radical.

Fig. 3. Changes in the DMPO spin-trapping ESR spectra for mpg-C3N4 (4 mg), mpg-C3N4 (4 mg)/BA (10 μL) and mpg-C3N4 (4 mg)/NHPI (10 mg) in CH3CN (1 mL). Each mixture was stirred for 10 min with O2 bubbling and visible-light irradiation and then put in a quartz tube for ESR measurement.

Subsequently, the PINO radical smoothly oxidized BA to benzaldehyde or benzoic acid with O₂ as confirmed previously [46, 55, 56]. A careful Fourier transform infrared (FT-IR) investigation of the mpg-C₃N₄/NHPI system in the presence of O₂ under visible-light irradiation revealed a broad peak around 3250 cm⁻¹ that was not observed in the mpg-C₃N₄ and NHPI alone (Fig. 4). This new peak is attributed to the stretching vibration of -OH in >O-H·····^•O-N<. Because of the high dipole moment of the nitroxyl functional group (6.7 D for PINO), PINO readily forms O-H·····^•O-N intermolecular hydrogen bonds with proton donors ^[57]. In fact, similar hydrogen bonds between 2,2,6,6-tetramethylpiperidine-N-oxyl (TEMPO) and phenol have been observed with a broad peak at ca. 3280 cm⁻¹ in FT-IR spectra ^[58]. As time elapsed, the intensity of the broad peak increased. These FT-IR results further confirm the generation of PINO radicals, which was promoted by visible-light irradiation of mpg-C₃N₄.

Fig. 4. FT-IR spectra of mpg-C3N4/NHPI in CH3CN at 25 °C in the presence of O2 under visible-light irradiation (> 420 nm).

To investigate the scope of this catalytic system, its ability to catalyze oxidation of various aromatic alcohols was investigated (Table 3). Three representative para-substituted aromatic alcohols were transformed to their corresponding aldehydes with good selectivity. Interestingly, the oxidation of p-methylbenzyl alcohol yielded the synthetically valuable p-methylbenzaldehyde (Table 3, entry 1). After 17 h at room temperature, the coupled system afforded a high yield in the oxidation of cyclohexanol (Table 3, entry 4). The oxidation of benzhydrol also proceeded smoothly, giving 100% conversion in 15 h (Table 3, entry 5). Relatively good results were achieved using a benzhydrol analog as a substrate (Table 3, entry 6). Thus, it is obviously possible to extend the scope of the mpg-C₃N₄/NHPI system to the selective oxidation of other alcohols.

Table 3 Oxidation of various alcohols with mpg-C3N4 and NHPI. Entry Substrate Product Time(h) Conv.(%) Sel.(%) 1 22 89 94 2 28 77 90 3 28 42 97 4 17 100 97 5 (Ph)2CHOH (Ph)2CHO 15 100 98 6 (4-Cl-Ph)2CHOH (4-Cl-Ph)2CHO 15 96 96 Reaction conditions: Alcohol substrate (10 mmol), mpg-C3N4 (200 mg), NHPI (1 mmol), acetonitrile (20 mL), 1 atm O2, 25 °C, visible light >420 nm.

Table 3 Oxidation of various alcohols with mpg-C3N4 and NHPI.

A visible-light-activated non-metal system that catalyzes selective oxidation of alcohols with good conversion at room temperature was developed. Compared with the g-C₃N₄/NHPI system, the current mpg-C₃N₄/NHPI system exhibited higher catalytic activity. Each of the three components of mpg-C₃N₄, NHPI and visible light make a unique contribution to the system performance. On the one hand, mpg-C₃N₄ not only induces a one-electron transfer under visible-light irradiation, but also acts as a host material with a specific 2D surface structure to control the reaction selectivity to some extent. On the other hand, the PINO radical as a redox center demonstrates powerful catalytic oxidation activity. During the mechanistic investigation, ^•O₂⁻ and PINO radicals were identified as the active species in the catalytic process by ESR and FT-IR measurements. We believe that many visible-light-responsive materials could be coupled with organocatalysts in a similar fashion to that described here to provide fascinating catalytic systems.