Zongyi
Yu‡
a,
Shichang
Li‡
a,
Yufeng
Wu
a,
Cunfei
Ma
a,
Jianing
Li
a,
Liyuan
Duan
a,
Zunchao
Liu
a,
Huinan
Sun
a,
Guofeng
Zhao
a,
Yue
Lu
a,
Qilei
Liu
*a,
Qingwei
Meng
*ab and
Jingnan
Zhao
*a
aState Key Laboratory of Fine Chemicals, Frontiers Science Center for Smart Materials Oriented Chemical Engineering, Department of Pharmacy, School of Chemical Engineering, Dalian University of Technology, Dalian 116024, China. E-mail: Mengqw@dlut.edu.cn
bNingbo Institute of Dalian University of Technology, Dalian University of Technology, Ningbo 315000, China
First published on 11th July 2024
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.
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 BiVO416 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).
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.
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 photocatalysis42–46 (eqn (1), Fig. 1B and C). The regression results demonstrated that the dipole moment47–49 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)”.
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 CO 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.
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).
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.
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.
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.
Entrya | 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.
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.
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 process55–57 (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.
2H+ + H2O2 + 2I− ↔ 2H2O + I2 | (a) |
2S2O32− + I2 ↔ S4O62− + 2I− | (b) |
(c) |
(d) |
Footnotes |
† Electronic supplementary information (ESI) available. See DOI: https://doi.org/10.1039/d4gc02385d |
‡ Equivalent to first author. |
This journal is © The Royal Society of Chemistry 2024 |