H₂O₂ is recognized as a green oxidant and finds extensive applications in industrial processes, chemical synthesis, and environmental application [1]. The predominant method for the commercial production of H₂O₂ is the anthraquinone process. However, this method suffers from harsh reaction conditions and high energy consumption, which limits its widespread application [2]. In recent years, the solar-to-H₂O₂ energy conversion has emerged as a promising alternative for H₂O₂ production. In this approach, photogenerated e⁻ from light excitation participates in oxygen reduction reaction (ORR) to yield H₂O₂ (Eqs. 1 and 2) [3]. Zhang et al. proposed resin-based photo-self-Fenton system in which H₂O₂ was photocatalytic produced and activated into ^•OH for bisphenol A degradation [4]. Typically, the reaction between h⁺ and H₂O (Eq. 2) is thermodynamically and kinetically challenging, leading to insufficient e⁻-h⁺ separation, which subsequently hinders the ORR. Consequently, the requirement for aeration or sacrificial agents addition remains inevitable in previous reported works, which increases operation costs and complexity [5–7]. Actually, the direct oxidation by h⁺ earns a crucial place in antibiotics degradation. Conversely, antibiotics could also trigger the h⁺ scavenging effect, thereby promoting the ORR without the need for the addition of exogenous reagents. In this regard, we conceive of constructing dual sites for both direct h⁺ oxidation and ORR, thereby enabling the synergistic antibiotics degradation and H₂O₂ production.

$ \mathrm{O}_2+2 \mathrm{H}^{+}+2 \mathrm{e}^{-} \rightarrow \mathrm{H}_2 \mathrm{O}_2 $

(1)

$ 2 \mathrm{H}_2 \mathrm{O}+4 \mathrm{~h}^{+} \rightarrow \mathrm{O}_2+4 \mathrm{H}^{+} $

(2)

TiO₂, a widely studied photocatalyst, exhibits excellent redox properties and is considered as a promising candidate for the construction of bifunctional catalytic materials. However, it still suffers from limitations such as low mass transfer efficiency and insufficient O₂ affinity [8]. Crystallization by particle attachment, wherein appropriately selected precursor serve as fundamental building blocks for oriented assembly into mesocrystals, offers a potential solution [9]. Mesocrystals are characterized as superstructures of well-aligned nanoparticles, typically exhibiting regular morphology, uniform size, and excellent porosity [10]. In addition, vacancy engineering has been proved a powerful strategy to improve the photocatalytic performance. For instance, oxygen vacancies (Ov) can serve as shallow electron acceptors for electron-rich molecules, thereby enhancing O₂ affinity for ORR [11]. Hence, constructing TiO₂ mesocrystals with rich Ov could simultaneously realize antibiotics degradation and H₂O₂ production.

In this work, TiO₂ mesocrystals with Ov (meso–TiO_2-x) were successfully prepared through a facile pyromellitic diimide (PDI)-assisted hydrothermal process. Comprehensive characterizations were conducted to elucidate the detailed structure of the catalyst. The photocatalytic H₂O₂ production and ciprofloxacin (CIP, a fluoroquinolone antibiotic, chosen as a model pollutant) degradation were thoroughly investigated. Scavenger experiments and electron paramagnetic resonance (EPR) spectroscopy were employed to determine the primary reactive species involved. Additionally, density functional theory (DFT) calculations were carried out to reveal the underlying mechanisms responsible for the enhanced performance of meso–TiO_2-x. Overall, this study developed a bifunctional photocatalyst for synergistic antibiotics degradation and H₂O₂ production. The detailed information for chemicals and materials, catalysts preparation, characterization and analytic methods was presented in Texts S1-S6 (Supporting information). PDI-x represented as catalysts with a mass ratio (Ti-glycerol/PDI) of x. Specially, PDI-0 and PDI-6 were pristine TiO₂ and meso–TiO_2-x, respectively.

The microstructures of TiO₂ and meso–TiO_2-x were investigated using scanning electron microscopy (SEM) and transmission electron microscopy (TEM) in Figs. 1a and b. The pristine TiO₂ showed larger particle sizes with uneven size distribution. while meso–TiO_2-x exhibited a regular morphology with uniform particle size, which could be attributed to the superstructures of well-aligned mesocrystals. The hydrolysis of the imide bonds in the PDI structure generated carboxyl groups, which chelated with Ti, thereby providing nucleation sites for TiO₂ growth. This process effectively prevented the extensive aggregation of crystals, promoting uniform crystal growth, reducing particle size, and leading to the formation of mesocrystals. BET analysis in Fig. 1c, Fig. S1 and Table S1 (Supporting information) indicated that meso–TiO_2-x had higher porosity compared to TiO₂, with surface areas of 177.1 and 15.4 m²/g, respectively. The TEM images (Fig. 1d) revealed a lattice spacing of 0.35 nm, corresponding to the (101) crystal plane of anatase TiO₂ [12]. Furthermore, high-resolution TEM image showed a sharp contrast in intensity analysis, suggesting the existence of O_V, as highlighted in the yellow circle (Fig. 1e). As evidenced in the structural model (Fig. 1f), the center distance of 3.78 Å corresponded to the oxygen atom, indicating the Ov exposed on the surface of anatase TiO₂. Energy dispersive spectroscopy (EDS) mappings in Fig. S2 (Supporting information) confirmed the uniform distribution of N, C, Ti, and O elements on the surface, validating the participation of PDI in the formation of meso–TiO_2-x.

The X-ray diffraction (XRD) patterns (Fig. 2a) indicated that the pristine TiO₂, synthesized without PDI (PDI-0), exhibited a brookite crystal structure. As the amount of PDI increased, TiO₂ gradually transitioned from brookite to anatase. Upon the addition of PDI, the pH of the hydrothermal solution decreased from 13 to 8–9. As the hydrothermal reaction progressed, PDI continued to react and hydrolyze, leading to a further pH decrease, ultimately falling below 4 (Fig. 2b). In the hydrothermal process, Ti⁴⁺ formed complex octahedral configuration with OH⁻ or O²⁻. With the pH decrease, OH⁻ on the surface of growth units was replaced by O²⁻, which impeded the surface hydroxylation and promoted the growth of long chain polymers to form anatase TiO₂. Consequently, the pH reduction caused by the hydrolysis of the imide bonds in PDI was the main factor driving the crystal phase transition of TiO₂ [13]. In the solid-state EPR spectra (Fig. 2c), the signal at g = 2.003 corresponds to e⁻ trapped by Ov [14]. The stronger EPR signal observed in meso–TiO_2-x compared to TiO₂ further confirmed the formation of Ov. The binding energy of the O 1s X-ray photoelectron spectra (XPS) in Fig. 2d exhibited peaks corresponding to lattice O (529.7 eV), -OH (532.2 eV), and adsorbed O₂ (536.2 eV) [15]. The introduction of PDI induced the emergence of new characteristic peaks attributed to Ov adsorbed oxygen species, whose intensity exhibited a positive correlation with the PDI amount. Additionally, the negative shift observed in the Ti 2p spectra (Fig. 2e) suggested an increase in the electronic density of Ti due to the absence of adjacent O atoms [16]. These findings provided compelling evidence for the present of Ov, which was consistent with the HRTEM results. Photoelectric measurements were performed to evaluate the photoelectric properties of the catalysts. As displayed in UV–vis diffuse reflectance (DRS) spectra (Fig. 2f), meso–TiO_2-x exhibited enhanced light absorption in the range of 400–800 nm, with two distinct absorption edges, which may be attributed to midgap energy level induced by Ov [17]. According to the Kubelka-Munk formula, the energy band corresponding to these two absorption edges were calculated to be 2.44 and 3.20 eV, respectively. The valence band potential of meso–TiO_2-x was 2.07 eV as manifested by XPS-VB spectrum in Fig. S3 (Supporting information). And the conduction band potential was determined as −1.20 eV in Mott-Schottky curves (Fig. S4 in Supporting information) [18]. Steady-state photoluminescence (PL) spectroscopy, presented in Fig. S5a (Supporting information), revealed a significant decrease in the PL intensity for meso–TiO_2-x, suggesting a reduction in the recombination of photogenerated carriers. Time-resolved PL measurements (Fig. S5b and Table S2 in Supporting information) further indicated that the carrier lifetimes for TiO₂ and meso–TiO_2-x were 11.94 and 4.7 ns, respectively. The Ov acted as electron traps, thereby reducing the carrier lifetime and mitigating the issue of carrier recombination [19]. As shown Fig. S5c (Supporting information), the smaller arc radius in the Nyquist plots for meso–TiO_2-x indicated a lower charge transfer resistance. Additionally, photocurrent measurements (Fig. S5d in Supporting information) under intermittent irradiation further confirmed the higher carrier transfer rate of meso–TiO_2-x.

The catalytic performance of the as-synthesized catalysts was initially assessed through the photocatalytic degradation of ciprofloxacin (CIP) (Fig. 3a). UV–vis irradiation alone was ineffective in removing CIP (< 11.9%), indicating that the direct photolysis of CIP was negligible. And meso–TiO_2-x alone resulted in only a 35.1% removal via adsorption. In contrast, the combination of the catalysts and UV–vis led to a substantial CIP degradation. The catalytic performance of catalysts with different amount of PDI was evaluated (Fig. 3b), the trend of the pseudo-first-order kinetic rate constants (k) first decreased and then increased, with a considerable k value of 0.068 min⁻¹ observed for meso–TiO_2-x. Besides the contribution of Ov, the increased specific surface area of anatase mesocrystals induced by PDI, which facilitated mass transfer, also accounted for the enhanced performance. Moreover, the photocatalytic degradation of CIP over meso–TiO_2-x was tested in various water matrices, including tap water, Yangshan lake, and Qinhuai river (Table S3 in Supporting information). As demonstrated in Fig. 3c, the degradation of CIP was minimally affected, with complete removal occurring within 60 min. The effect of initial pH, catalyst dosage and co-existing substances on CIP degradation were systematically investigated. Complete removal of CIP was achieved within 40 min across a wide pH range of 4–11 (Fig. S6 in Supporting information). Also, effective CIP degradation could be maintained in the presence of coexisting inorganic anions (Cl⁻, SO₄²⁻ and NO₃⁻) and NOM (natural organic matter) (Fig. S7 in Supporting information). Besides, considering the aggregation and light shading effect, the optimal catalyst dosage was 0.2 g/L (Fig. S8 in Supporting information). The catalytic performance of meso–TiO_2-x showed no significant decline in activity across five consecutive cycles (Fig. 3d). It could be seen from the XRD patterns of fresh and used catalyst (Fig. S9 in Supporting information), the structure of meso–TiO_2-x maintained good stability. In addition to CIP, the photocatalytic degradation was also efficient for various antibiotics, such as tetracycline (TC), norfloxacin (NOR) and ofloxacin (OFL) (Fig. 3e). However, the adsorption and degradation of sulfamethoxazole (SMZ) were unsatisfactory, which might be attributed to the specific active species associated with the bifunctional sites of the catalyst.

Furthermore, the photocatalytic H₂O₂ production was evaluated. As shown in the Fig. S10 (Supporting information), PDI and TiO₂ could hardly produce H₂O₂ in the absence of aeration and sacrificial agent. The H₂O₂ production of commercial anatase TiO₂ (a-TiO₂) was merely 34.3 µmol g⁻¹ h⁻¹. In contrast, the photocatalytic H₂O₂ yield over meso–TiO_2-x reached 387.5 µmol g⁻¹ h⁻¹ under pure water condition, demonstrating that Ov significantly enhanced the production of H₂O₂. Besides, the formation rate constant (k_f) and decomposition rate constant (k_d) were calculated in Fig. 3f and Fig. S10 (Supporting information). Meso-TiO_2-x exhibited the highest k_f of 7.0 µmol L⁻¹ min⁻¹, which was 8748.9, 68.7 and 7.6 times higher than PDI, TiO₂ and commercial anatase TiO₂ (a-TiO₂), respectively. Further, H₂O₂ production under Ar purging was completely suppressed, indicating that H₂O₂ generation primarily originated from the ORR (Fig. 3g). The concentration of dissolved oxygen (DO) after purging Ar was recorded in Fig. S11 (Supporting information). The DO concentrations remained consistently below 0.1 mg/L under both dark and light irradiation conditions, with no significant difference observed between these states. Notably, H₂O₂ production was undetectable in either condition. These results conclusively excluded the possibility of water oxidation reaction (WOR) contributed in the reaction. Purging with O₂ significantly facilitated the ORR, leading to a H₂O₂ yield of 645.8 µmol g⁻¹ h⁻¹. Furthermore, the addition of CIP enhanced the H₂O₂ yield to 904.2 µmol g⁻¹ h⁻¹, which was 1.4 times higher than that of pure water condition, likely due to the consumption of h⁺ by CIP improving the ORR process. Such promoting effect was also existed in TC, SMZ, OFL and NOR, with H₂O₂ production reached 912.5, 579.1, 612.5 and 533.3 µmol g⁻¹ h⁻¹, respectively (Fig. S12 in Supporting information). And, the introduction of the commonly used sacrificial agent isopropanol led to a more pronounced increase in H₂O₂ yield due to its higher efficiency in quenching h⁺. To validate the conjecture of synergistic interaction between antibiotics degradation and H₂O₂ generation. H₂O₂ production was measured in a series of organics with different ionization potentials (Fig. S13 and Table S4 in Supporting information). It could be seen that the kinetic constant for H₂O₂ production (k_H2O2) was inversely proportional to the ionization potential (IP) of organics (Fig. 3h). Specifically, the lower the ionization potential, the easier it was to lose e⁻ and be oxidized by h⁺, which was more conducive to ORR. This observation further supported that antibiotics degradation and H₂O₂ production were mutually reinforcing processes, exhibiting a synergistic relationship. Comparative analysis was conducted with previous reported works under comparable conditions (Fig. 3i and Table S5 in Supporting information). Notably, previous works often struggled to simultaneously balance the antibiotics degradation and H₂O₂ production. In this study, the construction of bifunctional meso–TiO_2-x effectively facilitated the synergistic occurrence of both antibiotics degradation and H₂O₂ production, demonstrating distinct advantages [20–32].

Scavenger experiment was conducted to identify the main active species involved in the CIP degradation. As shown in Fig. 4a, p-benzoquinonep (PBQ), tert–butyl alcohol (TBA), and furfuryl alcohol (FFA), which act as quenchers for ^•O₂⁻, ^•OH and ¹O₂, respectively, exhibited minimal inhibition on CIP degradation. In contrast, EDTA-2Na served as a h⁺ scavenger, exhibiting significant inhibition, which indicated that CIP degradation primarily relied on the direct oxidation by h⁻. Additionally, the introduction of KBrO₃ enhanced CIP degradation, as it acted as an e⁻ scavenger, promoting further separation and oxidation reaction of h⁺. Furthermore, the EPR spectra (Fig. 4b) revealed a gradual decrease in the TEMPO signal intensity over time, which could be attributed to the reduction of photogenerated e⁻. This observation suggested that the e⁻-h⁺ pairs in the system were effectively separated and actively participated in subsequent redox reactions [33]. Concurrently, the prominent DMPO-^•O₂⁻ signal indicated that ^•O₂⁻ was an intermediate product of the ORR (Fig. 4c).

DFT calculations were executed to gain further insight into the underlying mechanism for the function of Ov. The partial and summed density of states (PDOS) was analyzed in Fig. 4d. Compared to pristine TiO₂, a sub-band resulted from Ov state was found in meso–TiO_2-x, which in accordance with the DRS results [34]. Moreover, O₂ adsorption mode were evaluated in Fig. 4e. O₂ adsorption was thermodynamically unfavorable on pristine TiO₂ with a positive adsorption energy of 3.48 eV. However, upon the introduction of Ov, the adsorption energy was calculated as −3.66 eV, suggesting that photocatalytic H₂O₂ production via the ORR was more favorable on the Ov of meso–TiO_2-x. Based on the discussion above, the synergistic CIP degradation and H₂O₂ production was elucidated in Fig. 4f. Upon light stimulation, e⁻ was excited to the conduction band, leaving h⁺ in the valence band, while part of e⁻ transferred to intermediate levels, which inhibited e⁻-h⁺ recombination. Simultaneously, O₂ adsorbed in Ov to produce H₂O₂ via the ORR, while CIP was degraded through the direct oxidation of h⁺. This synergistic process of antibiotics degradation and H₂O₂ production was facilitated by the bifunctional properties of meso–TiO_2-x.

Liquid chromatography-mass spectrometry (LC-MS) was employed to identify the intermediates formed during CIP degradation, as summarized in Table S6 and Figs. S14-S20 (Supporting information). Accordingly, possible degradation pathways were proposed, as illustrated in Fig. 5a. Initially, CIP degradation proceeded via a dihydroxylation reaction, yielding TP3. The cleavage of the carbonyl group then produced TP7, which subsequently underwent ring-opening to form TP10. Additionally, the piperazine group in CIP was cleaved, resulting in the formation of TP1 and TP2. TP2 further transformed into TP5, TP6 and TP9, which were followed by dealkylation and decarboxylation to generate TP12 and TP13. Meanwhile, TP1 was cleaved into TP4 and TP8. The dissociation of the cyclopropyl group, coupled with decarboxylation, led to the formation of TP11 and TP14. To evaluate the ecotoxicity of the degradation products, key toxicity indicators were assessed using the Toxicity Estimation Software Tool [35]. As shown in Figs. 5b–e, acute toxicity, developmental toxicity, bioconcentration factor, and mutagenicity were significantly alleviated, suggesting that the effluent after treatment was more compatible with environmental release, thus lowering the potential ecological risk.

In conclusion, TiO₂ mesocrystal with Ov was prepared through a facile PDI assisted hydrothermal process. The hydrolysis of the imide bonds in the PDI provided nucleation sites for TiO₂ growth, thus facilitating the formation of mesocrystals. The well-aligned superstructures of meso–TiO_2-x renders it excellent porosity for the mass transfer. Besides, the resulted Ov not only served as e⁻ trap to tackle the carrier recombination issue but also facilitated the O₂ adsorption for ORR. Based on it, photocatalytic H₂O₂ production reached 904.2 µmol g⁻¹ h⁻¹ without aeration or sacrificial agents. Meanwhile, photocatalytic CIP degradation through h⁺ oxidation achieved a k value of 0.068 min⁻¹. Environmental toxicity analysis suggested that ecological risk of CIP was significantly reduced after treatment. Overall, this work proposed a facile strategy of constructing bifunctional photocatalyst for synergistic antibiotics degradation and H₂O₂ production.

Declaration of competing interests

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

CRediT authorship contribution statement

Zhiling Du: Writing – review & editing, Writing – original draft, Validation, Methodology, Investigation, Data curation, Conceptualization. Zhiqi Zhou: Validation, Investigation, Data curation. Nan Sun: Investigation, Data curation. Cailiang Yue: Writing – review & editing, Validation, Software, Methodology, Conceptualization. Fuqiang Liu: Writing – review & editing, Supervision, Resources, Project administration, Conceptualization.

Acknowledgment

This work was supported by the National Outstanding Youth Science Fund Project of National Natural Science Foundation of China (No. 51522805).

Supplementary materials

Supplementary material associated with this article can be found, in the online version, at doi:10.1016/j.cclet.2025.111341.