Continuous photocatalytic preparation of hydrogen peroxide with anthraquinone photosensitizers

Abstract

Hydrogen peroxide (H₂O₂) is a versatile oxidant, and its eco-friendly synthesis via visible light and air oxygen is crucial. Anthraquinone compounds were employed as homogeneous photocatalysts for alcohol oxidation, yielding H₂O₂ under visible light. Instead of using traditional catalyst library screening, a model based on density functional theory was proposed to structurally optimize anthraquinone photocatalysts, resulting in the discovery of a novel catalyst significantly boosting H₂O₂ generation (up to 621.2 mM h⁻¹ intermittently). Continuous flow reactor design improvements can enhance chemical transformation efficiency, as demonstrated by continuous H₂O₂ production using 2,6-di-tert-butyl anthraquinone. When isopropanol was used as the hydrogen atom donor, H₂O₂ production rates reached 3950.6 mM h⁻¹ in a 7-minute reaction, with a space–time yield of 7.90 mol (L h)⁻¹, showcasing promising advancements in green H₂O₂ synthesis.

---

Introduction

Hydrogen peroxide, a versatile oxidant with robust oxidative properties spanning the entire pH range,¹ is employed in the oxidative preparation processes of various crucial chemical products, including fatty aldehydes, aromatic ketones, cyclohexanone oxime, and epoxy propane.^2,3 Additionally, its oxidation reactions ultimately yield water, making it one of the cleanest and most environmentally friendly oxidants.⁴ Currently, the primary method for large-scale production of hydrogen peroxide (H₂O₂) is through the AQ (anthraquinone) process: anthraquinone reduction, hydrogenation, H₂O₂ extraction, and anthraquinone solution circulation.⁵ The commonly used anthraquinone catalyst is 2-ethylanthraquinone (EAQ), loaded at 120–140 g L⁻¹ in the working solution. For reduction, H₂ is used with Pd/C catalysts, achieving 9.0–12.0 g L⁻¹ hydrogenation efficiency (HE).⁶ The hydrogenated anthraquinone catalyst, acting as a hydrogen carrier, enters a countercurrent oxidation tower for reduction with air oxygen, yielding H₂O₂. Water is then added for H₂O₂ extraction from the solution, which is recycled for subsequent reaction cycles.

However, the above-mentioned process has several issues. The high loading of 2-ethylanthraquinone catalysts, potential deactivation of Pd/C and, along with the complexity of working fluid composition, high equipment investment, and significant discharge of waste, make this process hazardous, complex, and costly. There is a need to develop an efficient, safe, and environmentally sustainable method for H₂O₂ production.⁵

The production of hydrogen peroxide (H₂O₂) through photocatalysis stands as a sustainable process, leveraging water and oxygen as primary source materials and solar energy as its driving force. Recently, photocatalytic H₂O₂ production has become a significant area of research.⁷ A wide range of photocatalytic strategies has been explored, mainly categorized into heterogeneous and homogeneous systems. In heterogeneous strategies for H₂O₂ production using water or organic sacrificial agents, the focus is on various inorganic semiconductors, such as g-C₃N₄,^8–13 TiO₂,¹⁴ ZnO,¹⁵ BiVO₄ ¹⁶ and other doped semiconductor materials.^17–21 Additionally, organic materials like polymers,^21–25 COFs,^26–29 MOFs³⁰ and supramolecular photocatalyst.^31–33 After modifications and improvements like nanostructure design, co-doping, metal loading, surface modification, and organic photosensitizer introduction, the highest H₂O₂ production yield achieved in heterogeneous photocatalytic strategies is 1.7 mM h⁻¹ (approx. 1.7 mmol g⁻¹ h⁻¹). This was with Co(II)-doped TiO₂ under 420 nm light.

Homogeneous photocatalytic strategies, typified by anthraquinone compounds like EAQ, have showcased superior catalytic efficiency in H₂O₂ production. Ren³⁴ and team initially demonstrated H₂O₂ production using the industrial anthraquinone process's working solution, yielding 43.2 mM h⁻¹. Chen³⁵et al. utilized ethanol as a hydrogen donor, achieving 54.0 mM h⁻¹ under simulated sunlight. Vibbert³⁶et al. used 2-tert-butylanthraquinone (BAQ) and toluene, achieving 63.1 mM h⁻¹ under 395 nm light. In our previous work,³⁷ we achieved 323.2 mM h⁻¹ using EAQ and isopropanol. While EAQ is commonly used, its susceptibility to oxidation in photocatalytic processes calls for novel, stable, and efficient photocatalysts for H₂O₂ photoproduction (Scheme 1).

Scheme 1 Overview detailing the intelligent photocatalytic continuous production of H₂O₂.

To reveal the structure–performance relationships of anthraquinone catalysts and address the issue of low photocatalytic efficiency, a density functional theory (DFT)-based surrogate model was developed to identify novel anthraquinone photocatalysts with high catalytic performance and reduce the synthesis workload. The AQ catalysts were characterized with the electrostatic potential (ESP) on molecular surface and the dipole moment, which were selected as critical descriptors and utilized to establish the surrogate model for predictions of the production of hydrogen peroxide in photocatalysis using the least squares regression method. The use of the surrogate model yielded both qualitative understanding of the catalytic rules and quantitative identification of several potential AQ catalysts. These catalysts were subsequently synthesized and their enhanced performance in producing H₂O₂ was confirmed through experimentally validation.

Given that industrial H₂O₂ applications use aqueous solutions with varied mass fractions, this approach reduces H₂O₂ mass transfer efficiency in organic phases, weakening oxidation efficiency. Introducing the aqueous phase complicates post-processing in organic reactions. This work aims to establish a novel strategy and apparatus for continuous, on-demand high-concentration H₂O₂ organic phase solutions. Tailored to reactants and reaction needs, this approach enables real-time, in situ production of target mass fraction H₂O₂ solutions, ensuring an environmentally friendly, efficient, controllable, and inherently safe process.

Results and discussion

Establishment of catalyst design model

First of all, the DFT-based electrostatic potential of molecular surface was performed on the existing anthraquinone-based catalysts in the laboratory. The molecular simulation results were utilized to analyze and predict the photocatalytic performance of each catalyst in the production of H₂O₂.

Linear regression analysis conducted on five anthraquinone-based catalysts—2-ethylanthraquinone (EAQ), 2-tert-butylanthraquinone (BAQ), 2-bromoanthraquinone (BrAQ), 2-chloroanthraquinone (CAQ), and 9-thioxanthenone (TX)—yielded a high correlation (R² = 0.9955) between the dipole moment (DM) descriptor computed by DFT and the measured rates of H₂O₂ production in photocatalysis^42–46 (eqn (1), Fig. 1B and C). The regression results demonstrated that the dipole moment^47–49 of catalysts within a certain range is closely associated with the photocatalytic performance of anthraquinone-based catalysts for H₂O₂ production. Accordingly, it was hypothesized that designing a novel catalyst and forecasting its photocatalytic performance for H₂O₂ generation can be realized by calculating the dipole moment via DFT calculations and applying eqn (1). Further experimental work confirmed this methodology using 2-methylanthraquinone (MAQ), which possessed a dipole moment of 0.6496 and was predicted to have an H₂O₂ yield of 535.4 mM h⁻¹. This prediction had a relative error of only 1.81% from the experimentally measured yield of 525.7 mM h⁻¹ (Fig. 1D and E), highlighting the superior catalytic performance of MAQ over BAQ. Additionally, the prediction for 2-tert-amylanthraquinone (TAAQ) (a long-chain alkane-substituted derivative) showed an H₂O₂ production of 451.4 mM h⁻¹ predicted by the dipole moment of 0.9365, deviated by only 2.39% from the actual yield of 440.6 mM h⁻¹ (Fig. 1D and E). Considering catalyst deactivation risks due to oxidation and minimizing active C–H bonds, 2,6-di-tert-butylanthraquinone (DBAQ) was designed and synthesized with a dipole moment of 0.0007, predicting an H₂O₂ yield of 717.0 mM h⁻¹ compared to the experimental 621.2 mM h⁻¹ with a deviation of 12.94% (Fig. 1D and E). Despite increased deviation, substituents significantly affect electron cloud distribution. Combining the newly tested three catalysts with the initial five and re-conducting the linear surrogate model, the fitting results yielded an R² value of 0.9828 for eqn (2) (Fig. 1F), confirming the robustness of this catalyst design and predictive modeling strategy. This technique promotes the fine-tuning of catalytic activity, provides guidance for experimental work, lowers research costs, and predicts future research directions. This DFT-based surrogate model is referred to as the “Prediction model of anthraquinone photocatalytic H₂O₂ production based on electrostatic potential and dipole moment (PMAQ-PHP-EPDM)”.

Fig. 1 (A) The structures of five anthraquinone molecules used to establish linear relationships and the corresponding reaction conditions. (B) The dipole moments and generation rates of H₂O₂ table of basic 5 AQs. (C) Fitting curves and linear regression equation (eqn (1)) derived from the dipole moments of 5 AQs and their correlation with photocatalytic H₂O₂ production rates. (D) Three anthraquinone catalysts obtained via “Prediction model of anthraquinone photocatalytic H₂O₂ production based on electrostatic potential and dipole moment” (PMAQ-PHP-EPDM). (E) The dipole moments and generation rates of H₂O₂ table of total 8 AQs. (F) Fitting curves and linear regression equation (eqn (2)) derived from the dipole moments of 8 AQs and their correlation with photocatalytic H₂O₂ production rates.

In addition to quantitative predictions, a qualitative assessment was conducted on a wider range of anthraquinone derivatives with diverse functional groups. Electrostatic potential diagrams revealed crucial features contributing to photocatalytic H₂O₂ production. High-performing compounds like DBAQ, MAQ, BAQ, EAQ, and TAAQ showed a richly electronegative rhombic ring structure characterized by the C=O group and adjacent benzene rings (Fig. 2, red diamonds). Notably, these structures lack an electron-deficient center illustrated in deep blue.

Fig. 2 Qualitative relationship between the catalysts’ electrostatic potential energy surface and photocatalytic H₂O₂ production (all data are experimentally measured).

Anthraquinones exhibiting electron-deficient centers (Alizarin, HAQ, Emodin, DAAQ, Fig. 2, blue circles) showed poor performance. This reduced performance is attributable to the increased electronegativity from hydroxyl or amino substitutions, which increased the dipole moments and hindered excited-state hydrogen atom transfer (HAT) processes via hydrogen bonds.

In summary, we have established a strategy for the rapid design, screening, and performance prediction of anthraquinone photocatalysts. First, a series of anthraquinone compounds were drawn using Gaussian View software. Then, electrostatic potential diagrams were obtained based on DFT calculations, and anthraquinone compounds with more red or yellow regions and no blue regions were selected (indicating that AQs with strong electron-rich properties and no electron-deficient centers have better HAT performance). On this basis, molecular dipole moments were calculated via DFT, and the calculated dipole moments were substituted into the linear equation of the existing predictive model (PMAQ-PHP-EPDM, eqn (2)), thereby obtaining the predicted photocatalytic H₂O₂ production values of these anthraquinone compounds under standard conditions. Catalysts without blue electron-deficient centers have smaller dipole moments and lower molecular polarity, which is conducive to uniform electron distribution and results in better HAT performance.⁵³

The novel anthraquinone catalyst, 2,6-di-tert-butylanthraquinone (DBAQ), features tert-butyl substitutions at the 2,6-positions, enhancing electron cloud density on the ring compared to BAQ. This modification significantly improves HAT and electron transfer (ET) efficiencies under light.^50,51 DBAQ synthesis is simple and doesn't require column chromatography purification. Initially, anthracene ring alkylation adds two tert-butyl groups, followed by oxidation to produce DBAQ (see ESI†).

After selecting DBAQ as the optimal photocatalyst, we initially investigated the light source's impact on the photocatalytic reaction (Scheme 2A). Two light sources, a 300 W Xe lamp and a 40 W LED, were compared. Despite the Xe lamp's higher power, the H₂O₂ production rate under standard conditions with simulated sunlight (AM 1.5G, λ > 320 nm) was only 356.8 mM h⁻¹, the low H₂O₂ production rate can be attributed to the significant decomposition of H₂O₂ caused by the ultraviolet portion of the light. In contrast, using a 427 nm blue LED light yielded a higher production rate. Subsequently, we explored LED light sources of different wavelengths, focusing on energy efficiency and catalytic efficiency. The wavelength of the light source has been screened, and the optimal wavelength is 427 nm, with a H₂O₂ production rate of 621.2 mM h⁻¹. Consequently, a 40 W 427 nm Kessil LED lamp was selected as the optimal light source.

Scheme 2 Screening and investigation of reaction conditions.

The optimal catalyst loading was determined by reducing DBAQ below the existing 0.1 mol% loading (Scheme 2B). However, this resulted in a noticeable drop in the H₂O₂ production rate per unit time. Increasing to 0.2 mol% did not lead to a significant improvement in the production rate (592.8 mM h⁻¹). Higher loadings (>0.5 mol%) caused turbidity and a sharp reduction in photocatalytic efficiency. Consequently, 0.1 mol% (4.16 g L⁻¹) was selected as optimal, achieving a hydrogenation efficiency (HE) of 21.0 g L⁻¹ for intermittent H₂O₂ production. In comparison, the industrial EAQ method uses 140 g L⁻¹ for an HE of 12.0 g L⁻¹. A 75% HE increase was achieved by our system while reducing catalyst usage by 97% (wt%).

To investigate pH's influence, phosphoric acid was used as an acidic modifier. Addition notably enhanced H₂O₂ production. The highest rate, 699.7 mM h⁻¹, was at 0.1 mol% phosphoric acid, H₂O₂ mass fraction 3.03 wt%. Further addition decreased H₂O₂ production. Above 0.5 mol%, rates returned to baseline. In alkaline conditions, 1 mol% KOH yielded 446.8 mM h⁻¹. However, strong alkalinity causes H₂O₂ instability and decomposition to H₂O and O₂ (Scheme 2C and D).

Expansion of hydrogen atom donors

The traditional AQ method for H₂O₂ production involves using hydrogen gas as a hydrogen source and Pd/C as a hydrogenation catalyst under pressure in a fixed bed. Our approach is to activate the catalyst's potential through illumination at very low loadings without requiring H₂. This enables efficient hydrogen atom transfer (HAT) and avoids expensive catalysts and reaction risks.

Under intermittent conditions, we explored a total of 16 hydrogen atom donors (HADs) (Scheme 3), including low-carbon alcohols (C < 5) such as ethanol, isopropanol, n-butanol, 1,3-propanediol, and 1,3-butanediol. Apart from isopropanol, alcohols like ethanol (487.1 mM h⁻¹), n-butanol (361.9 mM h⁻¹), and 2-butanol (307.2 mM h⁻¹) exhibited high H₂O₂ production rates. However, for high-viscosity compounds like diols or aromatic alcohols, magnetic stirring was unable to fully unleash their potential as HADs, resulting in relatively lower H₂O₂ production rates. Therefore, the development of a reactor that can maintain good lighting conditions while achieving efficient mass transfer of gas–liquid two-phase is crucial for the efficient photocatalytic production of H₂O₂. This challenge is also a common issue faced by various photocatalytic reactions.

Scheme 3 Expansion of hydrogen atom donors.  Unless otherwise specified, HADs 1 (3 mL), and DBAQ (0.1 mol%), were added to a quartz beaker equipped with a stirring bar. The mixture was sonicated to dissolve completely. Then, the mixture was stirred under 1.0 atm O₂ atmosphere with exposure to 427 nm blue LED at room temperature for 1 h.

Continuous photocatalytic production of H₂O₂

The photocatalytic amplification effect poses a significant challenge for replacing the industrial anthraquinone process in H₂O₂ production. As the reaction vessel size increases, light illumination efficiency decreases sharply, affecting gas–liquid mass transfer. Continuous flow strategies in photocatalysis, utilizing microchannel reactors, have gained traction for their efficiency in reducing light exposure distance and liquid layer thickness. Transparent microchannel reactors, like plate, fixed-bed, and coiled-tube types, enhance gas–liquid mass transfer through internal design or fillers, improving specific surface area and generating turbulence. A novel single-pass high-throughput continuous photocatalytic microchannel reactor, using an optimized catalyst, enables efficient continuous H₂O₂ production.

Establishment of a single-pass high-throughput continuous photocatalytic setup (SPHT-CP setup)

The SPHT-CP setup comprises a high-pressure plunger pump, gas mass flow meter, Y-shaped gas–liquid mixer, gas–liquid pre-cooling module, bead-filled PFA coiled tube reaction module, tail-end cooling module, and air-cooling module. The PFA tube used has an inner diameter of 2.0 mm and an outer diameter of 3.0 mm, with glass beads of 1.5 mm outer diameter used for filling.

Traditional PFA coiled tube reactors achieve gas–liquid mixing at the Y-shaped mixer by adjusting flow rates, followed by plug flow in the coiled tube module. However, gas–liquid mass transfer is limited to surface diffusion segments. To enhance mass transfer, transparent glass beads were added to promote turbulent flow, ensuring continuous gas–liquid mixing and maximizing mass transfer efficiency.

However, for a simple flow channel, filling it with glass beads is relatively straightforward. In our SPHT-CP setup, the total length of the tubing reaches 112 meters (Scheme 4C). Filling the entire length with glass beads in such an extensive tubing system is a challenging task. Nevertheless, we accomplished this by modular filling in four cylindrical continuous micro-packed reactors, each with an outer length of 50 cm and a diameter of 6 cm. These reactors allow for the flexible selection of the reaction channel's length based on the required reaction residence time, ranging from 28 m to 112 m.

Scheme 4 The schematic diagrams and physical images of SPHT-CP setup.

The reserved volume of each reactor ranges from 40 mL to 160 mL.

In the realm of continuous microchannel photocatalysis, balancing enhanced gas–liquid mass transfer efficiency with sufficient light intensity for efficient single-pass H₂O₂ production is challenging. Our system addresses this by enhancing Taylor gas–liquid mass transfer using a PFA reaction tubing filled with glass beads, coiled around a quartz cylinder. Two layers of reflective films inside and outside the reactor, coupled with two 425 nm Kessil light sources, ensure optimal light intensity (64.3 mW cm⁻²) on the PFA surface for efficient illumination and reaction (Scheme 4B and D).

In the single-pass continuous reactor described above, we conducted an extended study on various HADs that were previously tested under batch conditions. It was observed that, regardless of whether alcohols or toluene and its homologs were used as HADs (Table 1, entries 9 and 10), the production rate of H₂O₂ per unit time significantly increased in continuous conditions, ranging from 4 to 403-fold improvement. In particular, when isopropanol was used as the HAD, the concentration of H₂O₂ at the outlet reached 2.01 wt% single-run time of 7 minutes, with a space–time yield of 7.90 [mol (L h)⁻¹]. This corresponds to the production of 107.5 g of a 30 wt% H₂O₂ solution per hour (pure H₂O₂ yield = 32.3 g h⁻¹) (Table 1, entry 2). The H₂O₂ concentration at the outlet already met the requirements of medical disinfectants, surpassing our previous reported photocatalytic systems and currently represents the highest level reported in the field of photocatalytic H₂O₂ production.

Table 1 Expansion of hydrogen atom donors in continuous flow strategy

Entry^a	1	Residence time (min)	H2O2 (wt%)	Generation rate of H2O2		STY
				(mM min^−1)	(mM h^−1)	[mol (L h)^−1]
a Unless otherwise specified, HADs 1 (100 mL), and DBAQ (0.1 mol%), were added to a beaker and sonicated to dissolve completely. Then, the mixture was thoroughly mixed with oxygen through a Y-shaped mixer and then pumped into the photocatalytic reactor at room temperature, vgas : vliq = 28 : 4 (mL min^−1).
1		8.5	1.39	38.0	2278.7	4.56
2		7.0	2.01	65.8	3950.6	7.90
3		12.0	1.04	20.8	1245.9	2.49
4		12.0	1.09	21.6	1294.6	2.59
5		12.0	2.14	46.2	2772.9	5.55
6		14.0	2.71	50.4	3022.1	6.04
7		18.0	1.60	27.2	1630.6	3.26
8		19.0	0.95	14.9	892.5	1.79
9		8.5	0.45	13.4	804.9	1.61
10		8.5	0.53	15.9	953.5	1.91

---

The longer reaction channel caused higher pressure (up to 0.8 MPa) due to the viscous nature of cyclohexanol in continuous conditions. However, by optimizing the reaction liquid composition, efficient hydrogen supply from cyclohexanol-type hydrogen atom donors (HADs) was achieved. The resulting product, KA oil (mixture of cyclohexanol and cyclohexanone), has industrial applications and can be oxidatively ring-opened to obtain adipic acid, crucial for nylon 6 production. In intermittent photocatalytic reactions, the H₂O₂ yield from cyclohexanol was only 8.54 mM h⁻¹ (Scheme 3, 1g). In continuous flow strategy, by optimizing the cyclohexane : cyclohexanol molar ratio to 2 : 1 led to a significant increase in H₂O₂ production rate, reaching 3022.1 mM h⁻¹, which is 403-fold improvement over batch conditions (Table 1, entry 5). This equates to producing 82.2 g of a 30 wt% H₂O₂ solution per hour (pure H₂O₂ yield = 24.7 g h⁻¹). H₂O₂ can be separated using a continuous extraction apparatus,³⁷ and the organic phase containing the catalyst can be recycled, enabling continuous and on-demand H₂O₂ production.

Moreover, various low-carbon alcohols achieved high H₂O₂ production efficiency (from 1245.9 to 2278.7 mM h⁻¹) in this continuous flow strategy (Table 1). Higher viscosity HADs also saw significant production rate increases (from 892.5 to 1630.6 mM h⁻¹), showcasing improved gas–liquid mass transfer and photoefficiency in the continuous flow device, leading to enhanced catalyst turnover numbers (TON) and efficient hydrogen atom transfer (HAT) processes for different HADs.

In a 48-hour uninterrupted reaction, a continuous photocatalytic device for H₂O₂ production was evaluated. Isopropanol served as the hydrogen atom donor, oxygen as the oxygen source, and DBAQ as the photocatalyst at 0.1 mol% loading. The gas–liquid flow rate ratio was 28 : 2 (ml min⁻¹), and the reaction pressure was 0.3 MPa. For the initial 6 hours, H₂O₂ concentration in the outlet solution was monitored hourly, with a consistent production rate of 44.2 mM min⁻¹ (2652.0 mM h⁻¹) and sample variance of 0.34 (mM min⁻¹)² (Fig. 3A). The stable operation allowed sampling every 2 hours from 6 to 48 hours. Results showed stable and efficient H₂O₂ production throughout, with a production rate of 44.8 mM min⁻¹ (2688 mM h⁻¹) over the entire 48 hours. Sample variance for the production rate from 27 data points was 0.92 (mM min⁻¹)². The H₂O₂ mass concentration remained at 1.56 wt%, with a low sample variance of 0.0011.

Fig. 3 (A) Concentration and production rate of H₂O₂ in DBAQ system for continuous photocatalytic preparation of H₂O₂ for 48 hours in SPHT-CP setup. (B) Impact of reduced reactor pressure on photocatalytic hydrogen peroxide production. (C) Comparison of ¹H-NMR spectra of DBAQ before and after photocatalysis. (D) The performance of the recovered DBAQ catalyst in the photocatalytic production of H₂O₂.

Reducing the pressure within the reactor decreases the compression of oxygen, thereby enhancing the turbulence of the gas–liquid two-phase flow in the reaction pathway. Therefore, at pressures of 0.2 MPa and 0.1 MPa, the H₂O₂ mass fractions at the outlet increased to 1.82 wt% and 2.24 wt%, respectively. Correspondingly, the H₂O₂ production rates were elevated to 52.2 mM min⁻¹ and 64.2 mM min⁻¹ (3132.4 mM h⁻¹ and 3852.3 mM h⁻¹) (Fig. 3B).

Recovery of HAD isopropanol and the DBAQ catalyst through rotary evaporation and crystallization. ¹H-NMR spectra showed that the catalyst's structure remained unchanged, achieving a recovery rate of 93.6% (Fig. 3C). A continuous photocatalytic experiment was conducted for 4 hours using the recovered DBAQ. The results revealed that the average production of H₂O₂ within the 4-hour period was 45.2 mM min⁻¹ (Fig. 3A and D). Remarkably, under identical experimental conditions, the H₂O₂ production remained consistent with the results obtained from the 48-hour continuous experiment. Ultimately, a 310 g working solution of H₂O₂ with a concentration of 21 wt% was obtained.

Mechanism exploration

Based on previous research, we initially conducted a blank control experiment under standard experimental conditions. The results indicated that, in the absence of light or a catalyst, H₂O₂ production was almost negligible (Fig. 4A). To further investigate the mechanism of H₂O₂ generation, we introduced various scavengers into the standard reaction conditions, including the free radical inhibitor TEMPO (2,2,6,6-tetramethylpiperidoxyl), singlet oxygen quencher DABCO (1,4-diazabicyclo[2.2.2]octane), electron scavenger AgNO₃, hole scavenger NaC₂O₄. Experimental results revealed that in the group with the addition of the singlet oxygen quencher, H₂O₂ production was significantly suppressed, with a H₂O₂ yield of only 0.5 mM h⁻¹, compared to the control group's 49.1 mM h⁻¹ (Fig. 4B). In the group with the addition of the free radical scavenger TEMPO, the reaction was also partially inhibited, suggesting the generation of free radical intermediates in the H₂O₂ production process.

Fig. 4 (A) Blank control experiment (under intermittent standard conditions). (B) The effect of various scavengers on the photocatalysis process (switching to cyclohexane as the HAD on the basis of intermittent standard conditions). (C) EPR test of the photocatalytic system to detect the generation of singlet oxygen (¹O₂) under the irradiation. (D) The quench of DBAQ* with different concentrations of isopropanol. (E) Plausible mechanism for the production of H₂O₂ under DBAQ photocatalytic system.

To further investigate the reaction mechanism, we employed EPR analysis to capture reactive oxygen species (ROS) during the reaction process⁵² (Fig. 4C). EPR signals corresponding to singlet oxygen radicals was captured using TEMP (2,2,6,6-tetramethylpiperidine). The fluorescence quenching experiment confirmed that as the concentration of isopropanol in the reaction system increased, the fluorescence intensity of the reaction system gradually decreased, indicating that isopropanol can effectively quench the excited state of DBAQ (Fig. 4D). Additionally, high-resolution mass spectrometry was employed to capture critical isopropanol radical intermediates during the course of the reaction (Fig. S4†).

In addition, when utilizing EAQ as photocatalyst, TEMPO can capture the respective radical intermediates. However, this method failed to capture the relevant intermediates in the DBAQ photocatalytic H₂O₂ production system due to DBAQ's superior hydrogen atom abstraction and HAT efficiency in its excited state, preventing chemical capture of the semi-hydrogenated state of the catalyst radical intermediate.^37,53

In summary, we propose a possible photocatalytic mechanism for H₂O₂ generation. In this mechanism, the ground-state photocatalyst DBAQ is excited to its excited state photocatalyst DBAQ* upon exposure to light.^36,54 This excited-state photocatalyst can first undergo energy transfer (ET) for quenching and catalyst regeneration. The singlet-state photocatalyst facilitates the activation of triplet oxygen (³O₂) during intersystem crossing (ISC), resulting in the more oxidizing singlet oxygen (¹O₂).

Simultaneously, a portion of the excited-state photocatalyst, through a hydrogen atom transfer process^55–57 (HAT), abstracts a hydrogen atom from the hydrogen atom donor, forming the corresponding hydrogenated anthraquinone free radical intermediate (I, HDBAQ). This intermediate rapidly reacts with the photochemically generated singlet oxygen to efficiently produce H₂O₂. HDBAQ then oxidatively quenches to return to the ground state, completing the catalyst cycle. The HAD, isopropanol, generates a corresponding radical (II) after losing one hydrogen atom, and this radical, upon interaction with one molecule of O₂ or ¹O₂, forms a superoxide free radical intermediate (III). This intermediate, in turn, abstracts an active hydrogen from HDBAQ,^1,54 eventually yielding superoxide (IV) and, ultimately, releasing one molecule of H₂O₂ to form acetone.

Experimental

Synthesis of catalyst

Alkylation of anthracene for the synthesis of 2,6-di-tert-butylanthracene. The synthesis of 2,6-di-tert-butylanthracene³⁸ involved the reflux of anthracene with tert-butyl alcohol in trifluoroacetic acid as a solvent for 24 hours, as depicted in Fig. S1.†

Oxidative synthesis of 2,6-di-tert-butylanthraquinone (DBAQ) from 2,6-di-tert-butylanthracene. In a 500 mL round-bottom flask,³⁹ 55.0 mmol (15.96 g) of 2,6-di-tert-butylanthraquinone and 160 g of glacial acetic acid were combined. The reaction mixture was then heated to 124 °C and maintained at this temperature for 15 minutes before cooling it to room temperature, as depicted in Fig. S2.†

Experimental steps for photocatalytic production of H₂O₂

Intermittent standard conditions. Precisely measure and transfer the hydrogen atom donor and the catalyst, DBAQ, into a 10 mL vial equipped with a magnetic stirrer. Before initiating the reaction, evacuate the sealed vial to establish a vacuum inside. Connect a syringe filled with pure oxygen to the vial to ensure a pure oxygen environment within the system. Throughout the reaction, employ forced air cooling to maintain room temperature conditions during the reaction. Keep a constant distance of 4.0 cm between the light source and the vial. Activate the light source and the magnetic stirrer to commence the reaction. This photocatalytic H₂O₂ production strategy is a homogeneous photocatalysis process.

Continuous standard conditions

Single-pass high-throughput continuous photocatalytic setup (SPHT-CP setup). To prepare the photocatalytic reaction working solution, dissolve 1.30 mmol (0.4164 g) of DBAQ in 100 mL of isopropanol using ultrasonication until complete dissolution (0.013 M). The single-pass high-throughput continuous photocatalytic setup (SPHT-CP setup) consists of high-pressure plunger pump, gas mass flow meter, Y-shaped gas–liquid mixer, gas–liquid pre-cooling module, packed-bed PFA coiled tube reaction module, trailing cooling module, an air cooling module and LED light sources. The inner diameter of the PFA tubing used is 2.0 mm, and the outer diameter is 3.0 mm. Glass beads with an outer diameter of 1.5 mm are filled inside the PFA tubing. The PFA tubing is coiled around the outer wall of a quartz tube with a length of 50 cm and an outer diameter of 8 cm. During the reaction, the flow rate of the gas–liquid mixture and reaction time are controlled using a high-pressure plunger pump and a gas mass flow meter.

Titration analysis of H₂O₂

Titration reaction procedure^36,37. Take a small sample of the reaction mixture in a beaker and accurately measure its mass. Add 8 ml of 2 M sulfuric acid solution, followed by 5 drops of ammonium molybdate solution to enhance color development. Next, add 3 ml of starch indicator and 6 ml of potassium iodide solution. If H₂O₂ is produced in the reaction, the solution will turn dark purple. Finally, titrate the test solution with a standard sodium thiosulfate solution, shaking the beaker while adding the titrant until the solution fades when adding the last half-drop. Record the volume of the standard solution used and calculate the concentration of H₂O₂. The specific titration reaction equation and the concentration of H₂O₂ are shown in eqn (a)–(d).^40,41

2H⁺ + H₂O₂ + 2I⁻ ↔ 2H₂O + I₂ (a)

2S₂O₃²⁻ + I₂ ↔ S₄O₆²⁻ + 2I⁻ (b)

Conclusion

The PMAQ-PHP-EPDM model is developed for AQ catalyst designs based on DFT calculations and validated through the synthesis and application of DBAQ. DBAQ achieved the highest reported H₂O₂ production rates in both batch (621.2 mM h⁻¹) and continuous (3950.6 mM h⁻¹) conditions at a low catalyst loading of 0.1 mol%. In our self-constructed high-throughput single-pass continuous reactor (SPHT-CP setup), using a working solution comprising cyclohexanol and cyclohexane (molar ratio 2 : 1) with DBAQ, at a retention time of 14.0 minutes, the outlet H₂O₂ concentration reached 2.71 wt%, with a space–time yield of 6.04 (mol (L h)⁻¹) and a turnover number (TON) of 112. This efficient continuous photocatalytic H₂O₂ production strategy improved stability and catalytic efficiency while achieving single-pass continuous H₂O₂ production, offering a solution for on-demand synthesis. It also introduced working solutions for oxidation platforms and a modular continuous oxidation setup, advancing research on higher-value-added products.

Data availability

The authors confirm that the data supporting the findings of this study are available within the article [and/or] its ESI.†

Conflicts of interest

There are no conflicts to declare.

Acknowledgements

We are grateful for financial support from the National Natural Science Foundation of China (U20A20143). We would like to thank the State Key Laboratory of Fine Chemicals and the Fundamental Research Funds for China Central Universities (DUT22LAB608).

References

1. J. M. Campos-Martin, G. Blanco-Brieva and J. L. G. Fierro, Angew. Chem., Int. Ed., 2006, 45, 6962–6984.
2. X. Zeng, Y. Liu, X. Hu and X. Zhang, Green Chem., 2021, 23, 1466–1494.
3. R. J. Lewis, K. Ueura, X. Liu, Y. Fukuta, T. Qin, T. E. Davies, D. J. Morgan, A. Stenner, J. Singleton, J. K. Edwards, S. J. Freakley, C. J. Kiely, L. Chen, Y. Yamamoto and G. J. Hutchings, ACS Catal., 2023, 13, 1934–1945.
4. P. Tang, B. Ji and G. Sun, Carbohydr. Polym., 2016, 147, 139–145.
5. Q. Chen, Chem. Eng. Process., 2008, 47, 787–792.
6. H. Li, B. Zheng, Z. Pan, B. Zong and M. Qiao, Front. Chem. Sci. Eng., 2018, 12, 124–131.
7. T. Freese, J. T. Meijer, B. L. Feringa and S. B. Beil, Nat. Catal., 2023, 6, 553–558.
8. H. Tan, P. Zhou, M. Liu, Q. Zhang, F. Liu, H. Guo, Y. Zhou, Y. Chen, L. Zeng, L. Gu, Z. Zheng, M. Tong and S. Guo, Nat. Synth., 2023, 2, 557–563.
9. X. Zhang, H. Su, P. Cui, Y. Cao, Z. Teng, Q. Zhang, Y. Wang, Y. Feng, R. Feng, J. Hou, X. Zhou, P. Ma, H. Hu, K. Wang, C. Wang, L. Gan, Y. Zhao, Q. Liu, T. Zhang and K. Zheng, Nat. Commun., 2023, 14, 7115.
10. M. Volokh and M. Shalom, Nat. Catal., 2021, 4, 350–351.
11. Z. Teng, Q. Zhang, H. Yang, K. Kato, W. Yang, Y. Lu, S. Liu, C. Wang, A. Yamakata, C. Su, B. Liu and T. Ohno, Nat. Catal., 2021, 4, 374–384.
12. X. Wang, P. Zhou, Q. Zhou, Q. Zhang, H. Ning, M. Wu and W. Wu, Chem. Eng. J., 2023, 476, 146488.
13. Z. Wang, Y. Zhao, Y. Zhou, X. Wang, H. Huang, Y. Liu, M. Shao and Z. Kang, Chem. Eng. J., 2022, 427, 131972.
14. Y. Shiraishi, S. Kanazawa, D. Tsukamoto, A. Shiro, Y. Sugano and T. Hirai, ACS Catal., 2013, 3, 2222–2227.
15. L. Wang, J. Zhang, Y. Zhang, H. Yu, Y. Qu and J. Yu, Small, 2022, 18, 2104561.
16. T. Liu, Z. Pan, J. J. M. Vequizo, K. Kato, B. Wu, A. Yamakata, K. Katayama, B. Chen, C. Chu and K. Domen, Nat. Commun., 2022, 13, 1034.
17. H. Hou, X. Zeng and X. Zhang, Angew. Chem., Int. Ed., 2020, 59, 17356–17376.
18. R. Wang, M. S. Shi, F. Y. Xu, Y. Qiu, P. Zhang, K. L. Shen, Q. Zhao, J. G. Yu and Y. F. Zhang, Nat. Commun., 2020, 11, 4465.
19. R. Ma, L. Wang, H. Wang, Z. Liu, M. Xing, L. Zhu, X. Meng and F. Xiao, Appl. Catal., B, 2019, 244, 594–603.
20. Y. Isaka, Y. Kondo, Y. Kawase, Y. Kuwahara, K. Mori and H. Yamashita, Chem. Commun., 2018, 54, 9270–9273.
21. S. Wang, Z. Xie, D. Zhu, M. Bonn, H. I. Wang, Q. Liao, H. Xu, X. Chen and C. Gu, Nat. Commun., 2023, 14, 6891.
22. Y. Shiraishi, T. Takii, T. Hagi, S. Mori, Y. Kofuji, Y. Kitagawa, S. Tanaka, S. Ichikawa and T. Hirai, Nat. Mater., 2019, 18, 985–993.
23. Y. Su, K. Urayama, S. Kitagawa, L. Huang, S. Furukawa and C. Gu, J. Am. Chem. Soc., 2024, 146, 15479–15487.
24. X. Xu, Y. Sui, W. Chen, W. Huang, X. Li, Y. Li, D. Liu, S. Gao, W. Wu, C. Pan, H. Zhong, H. Wen and M. Wen, Appl. Catal., B, 2024, 341, 123271.
25. C. Chu, D. Huang, Q. Zhu, E. Stavitski, J. A. Spies, Z. Pan, J. Mao, H. L. Xin, C. A. Schmuttenmaer, S. Hu and J. Kim, ACS Catal., 2019, 9, 626–631.
26. Q. Zhi, W. Liu, R. Jiang, X. Zhan, Y. Jin, X. Chen, X. Yang, K. Wang, W. Cao, D. Qi and J. Jiang, J. Am. Chem. Soc., 2022, 144, 21328–21336.
27. M. Deng, J. Sun, A. Laemont, C. Liu, L. Wang, L. Bourda, J. Chakraborty, K. Van Hecke, R. Morent, N. De Geyter, K. Leus, H. Chen and P. Van Der Voort, Green Chem., 2023, 25, 3069–3076.
28. F. Hao, C. Yang, X. Lv, F. Chen, S. Wang, G. Zheng and Q. Han, Angew. Chem., Int. Ed., 2023, 62, e202315456.
29. Q. Liao, Q. Sun, H. Xu, Y. Wang, Y. Xu, Z. Li, J. Hu, D. Wang, H. Li and K. Xi, Angew. Chem., Int. Ed., 2023, 62, e202310556.
30. Y. Isaka, Y. Kawase, Y. Kuwahara, K. Mori and H. Yamashita, Angew. Chem., Int. Ed., 2019, 58, 5402–5406.
31. Y. Zhang, C. Pan, G. Bian, J. Xu, Y. Dong, Y. Zhang, Y. Lou, W. Liu and Y. Zhu, Nat. Energy, 2023, 8, 361–371.
32. A. Gopakumar, P. Ren, J. Chen, B. V. M. Rodrigues, H. Y. V. Ching, A. Jaworski, S. V. Doorslaer, A. Rokicińska, P. Kuśtrowski, G. Barcaro, S. Monti, A. Slabon and S. Das, J. Am. Chem. Soc., 2022, 144, 2603–2613.
33. L. Xu, Y. Liu, L. Li, Z. Hu and J. C. Yu, ACS Catal., 2021, 11, 14480–14488.
34. M. Ren, M. Mao, X. Duan and Q. Song, J. Photochem. Photobiol., A, 2011, 217, 164–168.
35. G. Xu, Y. Liang and F. Chen, J. Mol. Catal. A: Chem., 2016, 420, 66–72.
36. H. B. Vibbert, C. Bendel, J. R. Norton and A. J. Moment, ACS Sustainable Chem. Eng., 2022, 10, 11106–11116.
37. Z. Yu, J. Zhao, Y. Wu, L. Guo, C. Ma, H. Zhu, J. Li, L. Duan, Z. Liu, H. Sun, G. Zhao and Q. Meng, ACS Sustainable Chem. Eng., 2023, 11, 13320–13332.
38. J. Lee, Y. Chen, J. S. Lin, C. Wu, C. Chen, C. Dai, C. Chao, H. Chen and W. Liau, Tetrahedron, 2011, 67, 1696–1702.
39. M. V. N. Raju, M. E. Mohanty, P. R. Bangal and J. R. Vaidya, J. Phys. Chem. C, 2015, 119, 8563–8575.
40. Y. Wei, J. Zhang, Q. Zheng, J. Miao, P. J. Alvarez and M. Long, Chemosphere, 2021, 279, 130556.
41. N. V. Klassen, D. Marchington and H. C. E. McGowan, Anal. Chem., 1994, 66, 2921–2925.
42. J. S. Murray and P. Politzer, Wiley Interdiscip. Rev.: Comput. Mol. Sci., 2011, 1, 153–163.
43. N. Argaman and G. Makov, Am. J. Phys., 2000, 68, 69–79.
44. P. Politzer and J. S. Murray, Theor. Chem. Acc., 2002, 108, 134–142.
45. 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 and H. Nakatsuji, Gaussian Inc., Wallingford CT, 2016, 1, 572.
46. T. Lu and F. Chen, J. Comput. Chem., 2012, 33, 580–592.
47. A. St Amant, W. D. Cornell, P. A. Kollman and T. A. Halgren, J. Comput. Chem., 1995, 16, 1483–1506.
48. J. Zhang and T. Lu, Phys. Chem. Chem. Phys., 2021, 23, 20323–20328.
49. T. Lu and Q. Chen, J. Mol. Model., 2020, 26, 315.
50. J. E. Bachman, L. A. Curtiss and R. S. Assary, J. Phys. Chem. A, 2014, 118, 8852–8860.
51. L. Capaldo and D. Ravelli, Eur. J. Org. Chem., 2017, 2056–2071.
52. Y. Nosaka and A. Y. Nosaka, Chem. Rev., 2017, 117, 11302–11336.
53. Organic Photochemistry, ed. A. Padwa, CRC Press, 1st edn, 1991,  DOI:10.1201/9780203744826.
54. D. Jiang, Q. Zhang, L. Yang, Y. Deng, B. Yang, Y. Liu, C. Zhang and Z. Fu, Renewable Energy, 2021, 174, 928–938.
55. L. Capaldo and D. Ravelli, Eur. J. Org. Chem., 2017, 2056–2071.
56. C. Tang, X. Qiu, Z. Cheng and N. Jiao, Chem. Soc. Rev., 2021, 50, 8067–8101.
57. L. Capaldo, D. Ravelli and M. Fagnoni, Chem. Rev., 2022, 122, 1875–1924.

---

Footnotes

† Electronic supplementary information (ESI) available. See DOI: https://doi.org/10.1039/d4gc02385d

‡ Equivalent to first author.

---