1 Introduction

Antibiotics play a vital role in the prevention and treatment of diseases and infectious in both humans and livestock [1, 2]. However, their large amounts of discharge into natural water body have triggered serious aquatic environmental pollution and ecological risks due to the unrestricted abuse, mass production, and lack of effective treatment approaches. This has brought about considerable attention to antibiotics as emerging organic contaminants [3, 4]. Levofloxacin (LFX), a typical fluoroquinolones antibiotic, has been extensively utilized as the oral and intravenous formulations, thereby leading to its frequent detection in surface and ground water bodies, sewage and drinking water [5-7]. Meanwhile, LFX has triggered severe concerns due to its significant threat to the humans and water bodies. Therefore, it is emergently imperative to develop green and effective techniques for removing the refractory LFX contaminants from effluents [8].

Recently, numerous efforts have been concentrated on the conventional wastewater treatment (CWT) technologies, such as biological methods, reverse osmosis, flocculation, air stripping, adsorption and membrane filtration. These technologies are widely utilized due to their mature technology, easy implementation, and lots of practical applications [9]. However, CWT technologies are inefficient in eliminating LFX effluent due to its strong microbial toxicity and highly stable structure. In comparison, advanced oxidation processes (AOPs), including semiconductor photocatalysis, electrocatalysis and Fenton oxidation, show promising prospects for completely mineralizing recalcitrant antibiotic contaminants [10, 11]. This is mainly due to the production of reactive radicals with high oxidation activity, such as superoxide (O₂^•–), hydroxyl (·OH), sulfate radical (SO₄^•–) and singlet oxygen (¹O₂) [12]. Among these, semiconductor photocatalysis technology has drawn considerable attention for degrading refractory organic pollutants due to its preeminent advantages, such as high efficiency, environmentally friendly, and convenient operation [13-16]. During this process, the ·OH, O₂^•– or photogenerated hole (h⁺) are the commonly dominating reactive oxygen species (ROS) in responsible for the oxidization of contaminants. In comparison, the SO₄^•– has a stronger oxidation ability due to its higher oxidation-reduction potential (2.5-3.1 V) [17]. Generally, SO₄^•– is primarily produced through the activation of persulfates, including peroxymonosulfate (PMS, HSO₅^-) and peroxydisulfate (PDS, S₂O₈^2-) [18, 19]. It has been reported that persulfates can be activated by heat, irradiation, transition metal ions, homogeneous/heterogeneous semiconductors, ultrasound, light, etc [20]. Accordingly, using photocatalysts to activate PMS towards the removal of recalcitrant contaminants is a more promising technology. Quite obviously, it is quite crucial to explore photocatalyst materials with strong oxidative reactivity, especially the visible-light-responsive photocatalysts, for efficient PMS activation and contaminants removal.

Although numerous photocatalysts have been developed since the pioneers’ work by FUJISHIMA and HONDA in 1972 [21], the scale practical application of the photocatalytic technology for addressing the worldwide environmental and energy issues remains a formidable task. Recently, the graphite-like g-C₃N₄, a fascinating metal-free polymer semiconductor photocatalyst with visible-light-response, has drawn significant attention and been extensively investigated in various photocatalytic fields due to its numerous outstanding advantages, such as high chemical stability, facile synthesis, and unique optical and electronic characteristics [11, 22-26]. However, the pure bulk g-C₃N₄ is generally synthesized by direct calcination of the precursor (e.g., melamine, urea, thiourea), which commonly results in several defects, such as unfavorable agglomeration, limited specific surface area, insufficient light absorption, and high photogenerated charge recombination rate [27-29]. Therefore, it is imperative to rationally design and optimize the morphology, structure, and composition of the g-C₃N₄ to further improve its photocatalytic activity. Particularly, morphological regulation and heterogeneous structure construction are commonly deemed as the promising strategies for regulating the photocatalytic properties of the g-C₃N₄.

Carbon quantum dots (CQDs) refer to a category of carbon nanomaterials possessing granule sizes less than 10 nm, whose structure commonly contains the O/N surface functional group (defect state) and core (intrinsic state) sp³ hybrid matrix [30, 31]. CQDs possesse numerous desirable advantages such as low toxicity, high photostability, good biocompatibility and favorable conductivity, which make them suitable for various applications in biological and chemical sensing, energy storage, nanomedicine, photocatalysis and electrocatalysis [32-34]. Some recently published works in our group have successfully confirmed that the CQDs play a significant role in heterogeneous nucleation induction due to its unique molecular structure and nano size, endowing their distinctive morphology and superior catalytic activities of the resulting catalyst material by varying the crystallization and growth process [35]. In addition, zinc ferrite (ZnFe₂O₄) is an iron-based spinel-structured soft magnetic material, which has the huge potential for usage as a visible-light driven photocatalyst in view of its superior features, for instance a narrow band gap (~1.92 V), noticeable chemical stability, high coercivity, low toxicity, and high quantum efficiency [36, 37]. Recently, numerous efforts have successfully manifested that the construction of heterojunction composites by coupling ZnFe₂O₄ with other photocatalysts can efficiently promote efficient separation of photogenerated e^- and h⁺ pairs, finally resulting in significant enhancement for the ultimate photocatalytic activity [38-41].

In this study, in order to improve the photocatalytic activity of pure g-C₃N₄ toward LFX degradation by activating PMS, a novel ternary Z scheme ZnFe₂O₄/g-C₃N₄/CQDs (ZCC) heterojunction nanocomposite was synthesized by anchoring ZnFe₂O₄ onto the CNC via a combination of microwave-assisted hydrothermal/solvothermal and high temperature calcination methods. The morphology, chemical composition, structure and physicochemical properties of synthesized photocatalysts were characterized and compared in detail. Sequentially, the photocatalytic activity of ZCC was then evaluated by investigating its ability to degrade LFX via PMS activation. In addition, the stability and reusability of ZCC were also verified by conducting the multi-step cyclic degradation experiments. Radical capturing in combine with ESR measurements was utilized for elucidating the reactive oxidation species (ROS) responsible for the LFX degradation. Additionally, the possible photodegradation mechanism of LFX removal was presumably clarified through comprehensive analysis on the intermediates with LC-MS technique. Consequently, our work offers a novel route for improving the photocatalytic property of g-C₃N₄ and efficient strategy towards LFX removal in effluent using photocatalytic activation of PMS.

2 Materials and methods

2.6 Photocatalytic degradation experiment

The photocatalytic degradation experiments were performed in a photocatalytic reaction apparatus (BL-GHX-V, China) equipped with a cooling circulation apparatus for maintaining a constant room temperature (around 25 ℃). A 500 W xenon lamp was selected as the light source for persistent supply of energy. Typically, 50 mL of LFX stock solution with designated concentration was transferred into a quartz light reaction reactor. Sequentially, photocatalysts with certain amounts were uniformly dispersed into the above LFX solution under magnetically stirring and then the pH was adjusted to a preset value. After that, the suspension was retained in the dark condition for 30 min to reach adsorption-desorption equilibrium. After dispersing PMS with specific proportion into the suspension, the photocatalytic degradation was operated by immediately turning on the xenon lamp light source. During the photocatalytic operation process, 6 mL of aliquots was withdrawn at a pre-scheduled time intervals of 30 min and then centrifuged to completely eliminate the solid photocatalyst. Afterward, the clarified filtrate was collected and measured with a UV-vis spectrophotometer (TU-1900, China) at the maximum absorption wavelength (λ=290 nm) of LFX to analyze the concentration of residual LFX. Finally, the LFX degradation rate of was evaluated according to the following Eq. (1):

(1)

where C₀ and Ct are the concentrations of LFX at time 0 and t, respectively, while A₀ and At refer to the absorbance intensities of the LFX solution at time 0 and t.

The total organic carbon (TOC) removal efficiency was measured to reflect the mineralization degree of LFX under the action of photocatalysis by ZCC. The TOC removal rate (η) was calculated based on the formula (2):

(2)

where C(TOC)₁ and C(TOC) are the TOC concentrations in initial LFX solution and degradation solution at time t, respectively.

Furthermore, the photocatalytic degradation process of LFX was investigated utilizing the pseudo first-order reaction kinetic model as the following formula (3):

(3)

where k_obs stands for the apparent rate constant (min^-1).

It is noted that the photocatalytic operation parameters were detailly optimized by varying the FLX initial concentration (5, 10, 20, 30 and 40 mg/L), catalysts dosage (5, 10, 20, 30 and 40 mg), solution pH (3, 5, 7, 9 and 11), utilization amounts of PMS (0.5, 1.0, 1.5, 2.0, 2.5 mL), respectively.

In addition, the degradation intermediates of LFX were identified by the liquid chromatography-mass spectrometry (LC-MS) technique. The stability and reusability of the catalysts were investigated by performing multiple recycle degradation experiments under identical conditions. The used catalysts were collected with centrifugation approach and washed with deionized water for removing the residual substance. Furthermore, the FT-IR, XRD, XPS and SEM of the catalyst before and after utilization were compared to assess the variation of composition, structure and morphology.

The reactive oxygen species (ROS) generated in the photodegradation process were analyzed by free radical capture (FRC) assays, in which the scavengers were selected as methanol (MeOH) for capturing hydroxyl radical (·OH) and sulfate radical (SO₄^•–), isopropanol (IPA) for trapping ·OH, p-benzoquinone (p-BQ) for scavenging the superoxide radical anion (O₂^•–), L-histidine (L-his) for checking the singlet oxygen (¹O₂), potassium iodide (KI) for obliterating the surface-bound free radicals and L-ascorbic acid (L-AA) for quenching the universal radical, respectively. Noteworthily, the FRC assay was performed at the same operation conditions with the above photodegradation reaction besides the addition of selected scavengers. For further unravelling the possible ROS, the ESR spectra were recorded on a Bruker model EPR A300 spectrometer equipped with a 300 W Xe lamp at 9.44 GHz at 300 K. In this measurement, 5, 5-dimethyl-1-pyrrolin-n-oxide (DMPO) in aqueous was utilized for capturing the O₂^•-, ·OH and SO₄^•- radicals.

Additionally, the chemical probe conversion experiments were performed to quantitatively check the produced reactive radicals. The nitroblue tetrazolium (NBT, 20 μmol/L) and terephthalic acid (TA, 20 μmol/L) were utilized as the probes of O₂^•– and ·OH, respectively. The produced O₂^•– can be quantified by investigating the decrease of the characteristic adsorption peak intensity at 259 nm of NTB. Moreover, the quantification of ·OH can be accomplished by exploring the increase of the fluorescence peak intensity at 425 nm of the converted 2-phthalic acid (TAOH) which originated from the reaction between TA and ·OH. In this work, we merely calculated and compared the concentration of O₂^•– for simplifying the experiment. The steady-state concentration of O₂^•– is determined using the formula below [43]:

r (4)

where r represents the initial conversion rate of NBT (µmol·L^-1·s^-1), K_O2•–+NBT denotes the reaction rate constant (5.88×10⁴ mol^-1·L·s^-1), C_NBT corresponds to the initial concentration of NBT (20 µmol/L), and [O₂^•–]_ss signifies the steady-state concentration of O₂^•–.

3 Results and discussion

As schematically illustrated in Figure 1(a), the synthesis of ZCC heterojunctions mainly involves 3 steps, which was detailedly described in the experimental section. 1) The reaction between melamine and cyanuric acid in a solution at 80 ℃ initially produces the precursors for preparation of g-C₃N₄. 2) Under the induction of CQDs, the binary CNC was sequentially gained by integrating CQDs into the g-C₃N₄ backbone via consecutive immersion, rotary evaporation and calcination treatment under N₂ atmosphere. During this process, the CQDs participate in the self-polymerization reaction of the precursors due to their abundant hanging groups. 3) Finally, ZCC was obtained by anchoring ZnFe₂O₄ onto the formed CNC using microwave-assisted hydrothermal method with the formed CNC, Zn(NO₃)₂ and Fe(NO₃)₂ as the feedstocks.

Figure 1全屏查看

(a) Schematic illustration of synthesis process of ternary ZCC; SEM images of (b) binary CNC and (c) ternary ZCC; (d) TEM and (e, f) high-resolution TEM (HRTEM) images of ZCC; (g) SAED pattern of ZCC; (h) EDS elemental mapping images of ZCC

The morphologies and microstructures of fabricated binary CNC and ternary ZCC heterojunctions were revealed by performing SEM and TEM measurements. As can be observed in Figure 1(b), the formed CNC displays a distinctive tubular-like morphology, while the pristine g-C₃N₄ (Figure S1(b)) has an irregular bulk morphology with severe agglomeration. This indicates that the induced CNC possesses a larger specific area than the pristine g-C₃N₄, possibly providing more reactive sites for photocatalytic reactions. During this process, the CQDs with their abundant external functional groups can react with the g-C₃N₄ precursors, leading to a variation in the crystallization process of CNC, which was also confirmed in other published reports [35]. In Figure 1(c), the SEM image of ZCC shows that the formed ZnFe₂O₄ microspheres, with a granule size of 100 nm, are immobilized onto the surface or inside of the binary CNC. This results in the formation of a heterojunction between the binary CNC and ZnFe₂O₄. However, we cannot identify the CQDs in the CNC and ZCC, likely owing to their small particle size (Figure S3(b)) and trace amounts in the binary CNC and ternary ZCC. In Figure 1(d), the TEM image of ZCC reveals the simultaneous existence of nanosheets and nanoparticles. However, due to the significantly larger particle size of CNC, we cannot identify the tubular-like CNC in the TEM image of ZCC. It can be inferred that the nanosheets correspond to the CNC, while the nanoparticles are related to ZnFe₂O₄, indicating the successful synthesis of the ZCC heterojunction. The HRTEM image of ZCC (Figure 1(e)) shows distinguished lattice fringes with an interlayer spacing of around 0.254 nm and 0.275 nm (Figure 1(f)), relative to the (311) lattice plane of ZnFe₂O₄ and the (200) lattice plane of g-C₃N₄, respectively, powerfully confirming the formation of the ZCC heterojunction. The selected area electron diffraction (SAED) pattern in Figure 1(g) displays distinguishable bright circular patterns, revealing the polycrystalline state feature of pure g-C₃N₄ and ZnFe₂O₄ in ZCC. The four clear diffraction rings indexed in the SAED pattern can be related to the (220) crystal plane of g-C₃N₄ and the (311), (422), and (511) crystal planes of ZnFe₂O₄, proving the formation of the ZCC heterojunction. Furthermore, the relative energy-dispersive X-ray (EDX) mapping images (Figure 1(h)) show that the ZCC is mainly composed of five elements with a uniform distribution, including Fe, Zn, C, N and O. This further demonstrates the successful synthesis of the ternary ZCC heterojunction.

The crystalline structure and phase composition of the synthesized samples were measured using X-ray powder diffraction (XRD). As displayed in Figure 2(a), the XRD pattern of pure g-C₃N₄ exhibits the obvious characteristic diffraction peaks at 12.3° and 26.3°, corresponding to the (100) and (220) crystal planes of g-C₃N₄. As for the ZnFe₂O₄, the characteristic XRD peaks at 29.8°, 35.3°, 42.8°, 53.3°, 56.5° and 62.1° are ascribed to the (220), (311), (400), (422), (511) and (440) crystal planes of cubic spinel structure, respectively, in good line with the standard JCPDS data (JCPDS No. 22-1012). From the XRD pattern of ZCC, we can clearly observe all the characteristic peaks of ZCC, we can clearly observe all the characteristic XRD peaks of ZnFe₂O₄ and g-C₃N₄, indicating the successful synthesis of ZCC heterojunction. Moreover, the absence of CQDs characteristic diffraction peaks in the XRD of ZCC composites may be relative to its low content, which is in consistent with the above-mentioned SEM measurement. Figure 2(b) shows the FT-IR spectra of the synthesized samples. The distinct peak located at 808 cm^-1 of pure g-C₃N₄ in FT-IR spectrum correlates to the respiratory mode of the triazine ring units, which represents the characteristic units of g-C₃N₄ [44]. Moreover, the FT-IR spectrum of ZnFe₂O₄ displays two significant absorption peaks at 606 and 467 cm^-1, which are attributed to the bending vibration of Fe—O and Zn—O, respectively [12]. A series of typical peaks in the range 1200 to 1750 cm^-1 are caused by the stretching vibrations of C—N and C=N heterocycles. As for CQDs, the characteristic peaks at 2830 and 2745 cm^-1 correspond to the stretching vibration of C—H bond and the peak at approximately 1134 cm^-1 is assigned to the tensile vibration of C—O [35]. Furthermore, the wide absorption band centered at 3420 cm^-1 for all samples is consistent with the stretching vibration of the O—H bond of the adsorbed water molecules. Importantly, all the characteristic FT-IR peaks of the aforementioned samples can also be identified in the FT-IR spectrum of ZCC, directly indicating the formation of the ZCC heterojunction by anchoring the ZnFe₂O₄ onto the CQDs induced CNC.

Figure 2全屏查看

(a) XRD patterns and (b) FT-IR spectra of ZnFe₂O₄, g-C₃N₄, CQDs, and ZCC

The chemical composition and surface chemical of the prepared samples were verified by performing the X-ray photoelectron spectroscopy (XPS) measurements. As demonstrated in the XPS survey spectrum (Figure 3(a)), the ternary ZCC composite is primarily composed of five elements including C, N, O, Fe and Zn, indicating the successful integration of CQDs, g-C₃N₄ and ZnFe₂O₄. From the high-resolution C 1s XPS spectra in Figure 3(b), we can easily discern two distinctive peaks located at 284.6 eV and 287.8 eV, which are attributed to the signals of C—C, C=C and C—N, correlating to the g-C₃N₄ or CQDs. In Figure 3(c), the main peaks centered at around 389.2 eV and 399.8 eV are associated with the C—N=C bond and N—(C)₃ bond of g-C₃N₄. The peaks deconvoluted from the high-resolution O 1s XPS spectrum in Figure 3(d) are attributed to surface hydroxyl groups (530.9 eV) and surface lattice oxygen of Fe—O and Zn—O (529.2 eV), separately. The high-resolution Fe 2p XPS spectrum in Figure 3(e) confirms the presence of Fe³⁺ in the ZCC heterojunction, with characteristic main peaks at 710.5 eV and 724.1 eV attributed to the Fe 2p_1/2 and Fe 2p_3/2 orbitals, respectively, and satellite peaks at 719.5 eV and 732.5 eV [45]. The high-resolution Zn 2p XPS spectrum in Figure 3(f) shows two typical peaks at 1021.3 and 1044.3 eV, which are classified as Zn 2p_1/2 and Zn 2p_3/2 respectively, unraveling the presence of Zn²⁺ in ZCC [46]. Furthermore, the C 1s and N 1s XPS characteristic peaks of ZCC composite show distinct shift toward a lower binding energy in comparison with those of the g-C₃N₄, while the O 1s, Fe 2p and Zn 2p XPS peaks of ZCC composite display apparent shift to a higher binding energy than those of ZnFe₂O₄ monomer [47], indicating the successful formation of heterojunction between ZnFe₂O₄ and CNC. In a word, the above XPS analyses further confirmed the simultaneous presence of ZnFe₂O₄, g-C₃N₄, and CQDs in ZCC.

Figure 3全屏查看

(a) XPS spectrum of the ternary ZCC composite; Comparison of high-resolution (b) C 1 s, (c) N 1 s and (d) O 1s XPS spectra between g-C₃N₄ and ZCC, (e) Fe 2p and (f) Zn 2p XPS spectra between ZnFe₂O₄ and ZCC

The optical properties of the prepared photocatalysts were explored through UV-vis diffuse reflectance spectroscopy (DRS) technique. As shown in Figure 4(a), the pure g-C₃N₄ has an absorption edge at approximately 460 nm, while the pristine ZnFe₂O₄ displays a strong visible light absorption ability in entire visible light region. When CQDs were introduced during the preparation of g-C₃N₄, the derived binary CNC composite shows a significantly improved visible light adsorption ability, with the absorption edge extending to around 650 nm compared to the pure g-C₃N₄. Moreover, after anchoring the ZnFe₂O₄ onto CNC, the formed ternary ZCC possesses stronger visible light response than those of the g-C₃N₄ and monomers. Noticeably, the obvious colors difference of the prepared catalysts shown in inset of Figure 4(a) also provides a direct proof on the light adsorption response variation. Quite obviously, the prominent optical property of the formed ternary ZCC heterojunction greatly contributes to its photocatalytic activity in activating PMS for LFX degradation. Figure 4(b) shows the Tauc plots of g-C₃N₄ and ZnFe₂O₄ semiconductors, which were obtained by plotting linear relationships between (αhν)^1/n and hv based on the Eq. (5):

Figure 4全屏查看

(a) UV-vis diffusion reflectance spectra and (b) derived Tauc plots of ZnFe₂O₄, g-C₃N₄, CNC and ZCC; (c) N₂ adsorption-desorption isotherm curves and (d) corresponding pore size distribution plots of g-C₃N₄, CNC and ZCC

(5)

where α stands for the absorption coefficient; h represents the Planck constant; v is the vibration frequency; A refers to relative to the proportional constant; E_g correlates with the forbidden band width of the semiconductor, and n denotes a variable value either 1/2 for indirect semiconductors or 2 for direct semiconductors.

In view of the direct semiconductors of pure ZnFe₂O₄ and g-C₃N₄, the band gaps of ZnFe₂O₄ and g-C₃N₄ were calculated to be approximately 1.91 eV and 2.72 eV, which were gained by extending the tangent to coordinate axis on band gap site, respectively.

The specific surface area and pore structure of the prepared catalysts were investigated by analyzing the N₂ adsorption-desorption isotherms. As revealed in Figure 4(c), all samples exhibited typical type Ⅳ isotherms with a type H3 hysteresis loop in the broad p/p₀ range of 0.4-1.0 according to the IUPAC classification, manifesting the presence of mesopores with a slit-shaped structure [48]. The BET specific surface area (S_BET) and corresponding pore parameter values of the prepared catalysts are summarized in Table S1. It can be observed that the S_BET values of ZnFe₂O₄, g-C₃N₄, CNC, and ZCC are 25.272, 14.366, 70.225 and 56.379 m²/g, respectively. Among them, CNC possesses the largest S_BET due to its distinctive tubular structure compared to the pure g-C₃N₄ with the smallest S_BET which suffers from the adverse agglomeration. Quite obviously, the induction of CQDs plays an extremely important role for this phenomenon and also the huge S_BET of CNC is favorable for the adsorption of pollutants and the exposure of more reactive sites [49]. After the introduction of ZnFe₂O₄, the S_BET value of ZCC composite material shows slight decrease than that of CNC, which may be related to the attachment of ZnFe₂O₄ nanospheres on the surface of and inside the tublar-shaped CNC. Apparently, the decrease in S_BET value is an indirect verification for the heterojunction formation between ZnFe₂O₄ and g-C₃N₄ due to the improved interface contact. The pore size distribution parameters of all samples were calculated as shown in Figure 4(d) and Table S1 according to the Barrett-Joyner-Halenda (BJH) method. It can be easily seen that the pore size distributions of all samples is concentrated in the mesoporous region.

The photocatalytic activities of the above-synthesized photocatalysts were assessed through activating PMS toward LFX photodegradation under visible light. Initially, for obtaining the best compositional catalysts, the mass ratio of CQDs in CNC and mass ratio of ZnFe₂O₄ in ZCC were optimized by exploring the LFX removal. As shown in Figure S4(a), when the added mass of CQDs reaching 0.01 g, the formed CNC-0.01 exhibits the highest removal rate reaching 55.3% toward LFX. The inferior removal rate is possibly ascribed to the incomplete morphology conversion into nanotube (low specific surface area) at a lower added mass <0.1 g and conversely the decreased reactive sites of the predominant catalyst ingredients (g-C₃N₄) at a higher introduction mass of CQDs> 0.1 g. Furthermore, with increase for the mass variation of the anchored ZnFe₂O₄ in ZCC, the removal rate of LFX shows significant difference and the 0.2 g ZnFe₂O₄ anchored in ZCC has a highest removal rate as high as 67.7% toward LFX. The lower removal rate is primarily caused by the inadequate heterojunction formation at insufficient introduction of ZnFe₂O₄ (<0.2 g) and shielding of the active site under excessive addition of ZnFe₂O₄ (>0.2 g). Subsequentially, the catalytic properties of the prepared catalysts were further evaluated by investigating the LFX photodegradation via PMS activation. As shown in Figure 5(a), the pure g-C₃N₄ and ZnFe₂O₄ deliver lower removal rate toward LFX degradation with merely 12.9% and 19.9% at 180 min, respectively. The binary CNC and ternary ZCN composites endow slightly improved removal rates reaching 57.3% and 47.8%, respectively. In view of the CNC, the improvement of the catalytic activity is mainly attributed to increment of specific surface area and the reactive sites due to the transform of well-dispersed nanotubes from the aggregated g-C₃N₄ nanosheet. Additionally, the CQDs have better conductivity due to the resembled structure feature with graphite-like carbon materials, which can commonly apply as a charge transfer bridge in the composite catalysts [26]. Quite obviously, the introduced CQDs in CNC can accelerate the transmission of the photogenerated electrons/holes (e^-/h⁺) pair. As for ZCN, the formation of heterojunction between g-C₃N₄ and ZnFe₂O₄ monomers is the main reason for the enhancement of the catalytic properties toward LFX removal, which can accelerate the photogenerated e^-/h⁺ pair and finally lead to the reduction of the e^-/h⁺ recombination. Significantly, the optimal ternary ZCC composite achieves the highest photodegradation rate as high as 81.3%, confirming 7.3 times that of the neat g-C₃N₄. Apparently, the CQDs induction coupled with heterojunction construction with ZnFe₂O₄ can result in a synergistic action for further improving the photocatalytic activity of neat g-C₃N₄, which may be related to the increased reactive sites due to the induced nanotube-shaped morphology, accelerated e^-/h⁺ transmission and inhibited recombination due to the presence of CQDs charge transfer bridge and formed n-n type heterojunction. Figure 5(b) shows the fitted kinetic curves based on the above-gained results in Figure 5(a). Apparently, the good linear relationship between ln(C_t/C₀) and reaction time (t) manifests that all the above-mentioned photodegradation process of LFX well follows a quasi-first-order kinetics process. The apparent rate constants (k_ops) obtained by calculating the slopes show an order of pure g-C₃N₄ (0.00079 min^-1) < ZnFe₂O₄ (0.00118 min^-1) < CNC (0.00450 min^-1) < ZCN (0.00327 min^-1)<ZCC (0.00782 min^-1). Based on the above experimental results, 0.2-ZCC was deemed as the optimal photocatalyst in the subsequent experiments.

Figure 5全屏查看

(a) Photodegradation curves in PMS system of LFX with g-C₃N₄, ZnFe₂O₄, CNC, ZCN, ZCC and in ZCC system without PMS and (b) the kinetic fitting curve of Figure 5(a)

Subsequently, the photocatalytic operation parameters for PMS activation toward LFX degradation were systematically optimized under visible light irradiation. As shown in Figure 6(a), with the increase of initial concentration of LFX from 5 to 40 mg/L, the removal rate of LFX displays a trend of initial increase at a lower concentration of LFX (less than 20 mg/L) and then decrease at a higher concentration of LFX (30 and 40 mg/L), accompanying with a biggest removal rate ~84.7% at the optimal concentration of 20 mg/L. Apparently, the higher concentration of LFX is not conducive to obtaining a high degradation rate and many possible reasons are capable to explain this phenomenon. Firstly, the existence of limited catalysts in the degradation system is insufficient to provide sufficient reactive sites and adsorption centers for PMS activation toward LFX degradation. Secondly, the more intermediates produced during the degradation process will produce competitive reactions with LFX. Besides, the higher LFX concentration would block the light from penetration into the solution and cover the catalyst active center. In contrast, a lower concentration of LFX will not be sufficient to make sufficient contact with the catalyst and participate in the reaction. Likewise, this photodegradation processes still satisfied the pseudo first-order kinetics (Figure 6(b)) and the corresponding k_obs values are 0.00801 min^-1 (5 mg/L), 0.00903 min^-1 (10 mg/L), 0.01024 min^-1 (20 mg/L), 0.00971 min^-1 (30 mg/L) and 0.00752 min^-1 (40 mg/L) (Figure 6(c)), respectively. Consequently, 20 mg/L of LFX was selected as the optimal concentration for subsequent experiments.

Figure 6全屏查看

Optimization on the photodegradation parameters of LFX by PMS activation with ZCC: (a) Initial concentration of LFX; (d) ZCC dosage; (g) Solution pH; (j) PMS addition amount; (b, e, h, k) Corresponding first-order kinetic curves derived from Figures 6(a), (d), (g) and (h); (c, f, i, l) Histograms of k_ops values derived from Figures 6(b), (e), (h) and (k)

Figure 6(d) exhibited the influence of applied dosage of ZCC on the PMS activation toward LFX degradation. With the increase of applied dosage of ZCC range from 5 to 20 mg, the removal rate of LFX shows a significant increase from 66.9% to 85.9%. This can be attributed that a small amount of catalyst can result in insufficient reactive centers and available photon, resulting in insufficient generation of ROS. Oppositely, with the further improvement of the applied dosage of ZCC range from 20 mg to 40 mg, the removal rate of LFX displays a significantly decline trend range from 85.9% to 68.9%. This can be related to the fact that the introduction of excessive catalyst might enhance the solution turbidity, thereby reducing the light utilization efficiency and resulting in insufficient production of the ROS. Meanwhile, this photocatalytic process also comply for the quasi-first-order kinetics model (Figure 6(e)) and the k_obs values (Figure 6(f)) shows a coincident order with the above-confirmed removal rate of LFX: 0.00660 min^-1<0.00873 min^-1<0.01105 min^-1>0.00749 min^-1>0.00709 min^-1. To sum up, 20 mg ZCC is the optimal dosage, resulting in the highest removal rate reaching 85.9%.

Consecutively, the influence of solution pH on the photodegradation efficiency of LFX was checked by utilizing ZCC (20 mg) to activate PMS toward LFX (20 mg/L) degradation under visible light. As the increase of solution pH from 3 to 5, the removal rate of LFX shows a significant increase from 77.7% to 94.4% (Figure 6(g)) and the corresponding k_obs values increase from 0.00844 to 0.01612 min^-1 (Figure 6(h)). Subsequently, with the persistent increase of pH to 7, 9 and 11, the removal rates of LFX by ZCC present prominent decline tendency to 84.2%, 63.7%, and 36.4%, as well demonstrating dramatically reduced value of k_obs values of 0.01011, 0.00621 and 0.00265 min^-1 (Figure 6(i)). Quite obviously, the weakly acidic condition is conducive to the removal of LFX and the strong acid and strong base conditions lead to deteriorated degradation of LFX. Many aspects can explain this phenomenon as follows. Firstly, the LFX owns dissociation constants (pKa) of 5.7 and 7.9, manifesting that LFX primarily presents three kinds of ionic forms involving a cation (LFX⁺) as the pH<5.7, an anion (LFX^-) as the pH>7.9, and zwitterion as the pH between 5.7 and 7.9 [6]. In addition, the surface charge of ZnFe₂O₄ is negative under strong acidic condition and positive under strong alkaline condition due its point of zero charge (pH_pzc)~5.9 [38]. Thereby, there is electrostatic repulsion between LFX and ZCC owning to their coincident charge feature at extremely acidic (pH=3) or alkaline (pH=11) conditions, which hinders the contact of ZCC and LFX contaminants, finally leading to adverse influence on the degradation efficiency. Besides, PMS has two pKa values according to reported publications [34, 35]. The pKa1 has a low value below 0 and the pKa2 value is about 9.4 [50]. Apparently, PMS mainly exists in the form of H₂SO₅ under extreme acidic conditions (pH=3), which is unconducive to the generation of high active sulfate radicals. Under alkaline conditions, PMS mainly has two forms of HSO₅^- or SO₅^2-. Thus, there is also electrostatic repulsion between LFX and PMS, which dramatically impedes the degradation of LFX. In addition, the protonated and deprotonated equilibrium of O₂^•– will be broken under strong acid condition, finally triggering the consumption of O₂^•–. On a contrary, under high pH, the generated SO₄^•– from PMS activation will react with OH^- to convert into ·OH, resulting in reduced oxidation capacity. Furthermore, it is reported that the cleavage reaction of piperazine rings is one of the most important degradation pathways of LFX and the cationic LFX is more prone to piperazine ring-opening reaction, finally leading to higher degradation rate of LFX under weak acidic conditions. To sum up, the optimal pH is ~5, which leads to the highest removal rate of LFX up to 94.4%.

The PMS concentration determines the generated amounts of ROS. As depicted in Figure 6(j), when the PMS concentration increases from 0.015 to 0.03 mg/L, the removal rates of LFX exhibit significant growing tendency from 65.9% to 95.3%, and the k_obs also shows a dramatical enhancement from 0.00634 to 0.01898 min^-1 (Figure 6(k)), confirming the promotion role for the LFX degradation at an appropriately increased amount of PMS. With further increment of the PMS concentration to 0.045, 0.060 and 0.075 mg/L, the obtained removal rates of LFX gradually decrease to 86.7%, 79.3% and 75.2%; likewise the k_obs gradually decreases to 0.01063, 0.0092 and 0.00722 min^-1 (Figure 6(l)). This can be rationally interpreted by the fact that insufficient amount of SO₄^•– generation at small amounts addition of PMS (<0.030 mg/L), which results in the inadequate contact between ROS and LFX. Conversely, the excessive PMS in the degradation system will react with the generated e^- and radicals (e.g., SO₄^•–) to convert into another low reactive radicals (e.g., SO₅^•–), leading to an adverse competitive reaction in the degradation system. Additionally, the excessive existence of SO₄^•– will result in a self-quenching effect, which can decline the oxidative capability. Hence, on the basis of above investigation, 0.030 mg/L addition of PMS is the optimal concentration. By further comparison with the previous developed g-C₃N₄ composite photocatalysts (Table S2), this ZCC photocatalyst possesses considerable catalytic activity toward LFX degradation, confirming its huge application prospect in treatment of antibiotic wastewater.

Besides the superior photocatalytic properties, the reusability and stability of photocatalyst are another priority consideration for the practical application. The stability and reusability were investigated by performing the multiple cycling experiments toward LFX degradation. As shown in Figure 7(a), after 5 cycles, the final removal rate toward LFX did not decrease apparently in comparison with that (95.3%) in the first running and still retained up to 85.5%, verifying the prominent stability and recyclability of the ZCC catalyst. At the same time, the TOC removal rate was calculated by measuring the TOC content of the photodegradation solution in the recycling experiments to monitor the mineralization degree of LFX. As can be seen from Figure 7(a), in the first cycle of photodegradation experiment, the TOC removal rate of ZCC for LFX was 60.2%, which slightly decreased to 54.8% after five cycles of experiments [51]. From Figure S6, we can easily see that the characteristic absorption peak intensity of LFX shows gradual decrease with the extension of degradation time for the first cycling experiment, confirming the gradual degradation of LFX pollutants. Furthermore, from the XRD patterns (Figure 7(b)) and FT-IR spectra (Figure 7(c)) and SEM image (Figure S7(a)) and XPS energy spectra (Figure S7(b)), we cannot observe that the ZCC photocatalyst has the obvious discrepancy before and after utilization, further manifesting that the ZCC photocatalyst has abundant stability.

Figure 7全屏查看

(a) Recycle tests and mineralization degree evaluation for the LFX photodegradation by ZCC composite; (b) XRD patterns and (c) FT-IR spectra of ZCC before and after utilization; (d) FRC analysis of photodegradation LFX by 0.2-ZCC under different scavengers; (e) ESR signal of DMPO-O₂^•– adducts and (f) ESR signal of DMPO-·OH adducts in ZCC system; (g) ESR signal of DMPO-SO₄^•– adducts in ZCC-PMS system under dark and light conditions, respectively; (h) Reactive kinetic curves of NBT photodegradation under different catalysts; (i) Quantitative determination of generated O₂^•– during photocatalytic process

Generally, the active radicals are dominantly responsible for the photocatalytic removal of contaminants. Therefore, free radical capture (FRC) assays were performed using p-benzoquinone (p-BQ) as the scavenger of O₂^•– (k=9.0×10⁸ mol^-1·L·s^-1), the methanol (MeOH) as the capturing agent of ·OH (k=9.7×10⁸ mol^-1·L·s^-1) and SO₄^•– (k=2.5×10⁷ mol^-1·L·s^-1), isopropanol (IPA) as the trapping agents for ·OH and L-Histidine (L-his) as the scavenger for ¹O₂ (k=3.2×10⁸ mol^-1·L·s^-1), respectively. In addition, potassium iodide (KI) and L-ascorbic acid (L-AA) are commonly applied as a quencher of the surface-bound free radicals and a universal radical trapping agent in water during the AOPs. As shown in Figure 7(d), without introduction of any scavenger, the normal photodegradation of LFX by ZCC/PMS system delivers a removal rate reaching 95.8%. When IPA was introduced into the system, the removal rate of LFX decreased to 74.3%. However, the addition of MeOH and L-AA solution makes the removal rates of LFX further decline to 21.4% and only 17.6%, separately. Therefore, it can be reasonably inferred that SO₄^•– is the predominant reactive radical and ·OH triggers a subordinative role in the photodegradation system of LFX by ZCC/PMS system. When L-his was added to the solution, the removal rate of LFX was slightly reduced, unraveling that ¹O₂ possesses the weakest contribution to degradation reaction. As for the addition of p-BQ (O₂^•–), the removal rate of LFX shows a significant inhibition, proving that the O₂^•– is one of the main free radicals in the degradation process. To sum up, all the relative radicals including SO₄^•–, O₂^•–, ·OH and ¹O₂ participated in the photodegradation of LFX, and the resulted contribution order was SO₄^•–> O₂^•–> ·OH > ¹O₂. Additionally, the photodegradation reaction was almost completely hindered when L-AA was introduced into the system, indicating that the ROS could be regarded as deriving from ZCC/PMS system [35]. After the addition of KI, the removal rate decreased by about 24.6%, revealing that the free radicals participated in photodegradation reaction were also produced on the surface binding layer of ZCC [38]. To sum up, it can be concluded that the removal of LFX by ZCC/PMS system is a complex catalytic process involving many free radicals.

For further confirming presence of the above-mentioned predominant reactive radicals during the photodegradation, the ESR tests were conducted using 5, 5-dimethyl-1-pyrrolin-n-oxide (DMPO) as the spin-trapping agent. As can be seen in Figure 7(e), the ZCC system (without PMS) possesses four clearly continuous ESR signal peaks with identical intensity of 1:1:1:1 under light irradiation, which are attributed to the formation of DMPO-O₂^•– adducts. Likewise, Figure 7(f) also shows four continuous ESR signal peaks with various intensity of 1:2:2:1, indicating the formation of DMPO-·OH of the above ZCC system. Besides, no ESR signal under dark conditions for the above-mentioned ZCC system verifies that the ZCC catalyst does not produce reactive radicals under dark. After adding PMS in ZCC system, the ESR signal peaks in Figure 7(g) show obvious difference with the above-confirmed ESR signal peaks (Figures 7(e) and (f)), suggesting the simultaneous presence of ·OH and SO₄^•– in ZCC/PMS system under light irradiation [52]. This can mainly be attributed that the generated photoelectron from ZCC preferentially react with PMS to produce SO₄^•– in ZCC/PMS system rather than reduce O₂ to O₂^•– and thereby the generated holes can oxidize the H₂O to produce ·OH, which are also demonstrated in the radical capture experiments (Figure 7(d)). Consequently, the results of ESR test further manifested that O₂^•–, ·OH and SO₄^•– were the main reactive substances in the photocatalytic degradation of LFX by ZCC/PMS system.

Furthermore, the generated O₂^•– was quantitatively analyzed utilizing a chemical probe conversion experiment. Generally, the transformation of NBT probe makes the characteristic absorption peak at 259 nm vanishment, which was commonly employed to quantitively determine the generation of O₂^•– in the photocatalytic system. As shown in Figures S8(a) and (b), it is evident that the intensity of the absorption peak of NBT gradually diminishes with the reaction time elongation under existence of CNC and ZCC. By comparison with the CNC, the ZCC leads to more significant decrease for the characteristic absorption peak of NBT within 180 min, indicating a larger amount of production of O₂^•– by ZCC. From the reaction kinetic curves (Figure 7(h)), the reaction rate constant k shows a tendency ZCC/NBT (0.00412 min^-1)>CNC/NBT (0.00118 min^-1), indicating the more rapid reaction between NBT and the generated O₂^•– by ZCC than CNC. After calculation, the steady-state concentrations of O₂^•– for CNC and ZCC are 4.03×10^-10 mol/L and 14.05×10^-10 mol/L (Figure 7(i)), respectively. Moreover, non-fluorescent terephthalic acid (TA) can react with the ·OH to generate a unique and stable 2-phthalic acid (TAOH) with intense fluorescent characteristics at 425 nm, which was commonly employed as a molecular probe to detect the ·OH. As shown in Figure S8(c), TA+ZCC system demonstrates a significantly higher fluorescence intensity than that of the TZ+CNC system, indicating production of a larger number of ·OH radicals. This also corresponds to the higher photodegradation performance of ZCC.

Photoluminescence (PL) spectroscopy is a commonly-used approach for unveiling the separation and recombination process of photogenerated e^-/h⁺ pairs [53]. Commonly, low PL intensity stands for high separation efficiency of e^-/h⁺ for photocatalyst, finally leading to superior photodegradation activity [54]. As manifested in Figure 8(a), in comparison with the binary CNC and monomer ZnFe₂O₄, the ternary ZCC possesses the lowest PL strength, confirming that the ZCC heterojunction composite shows the highest photogenerated e^-/h⁺ pairs separation and transmission efficiency, finally resulting in the highest photocatalytic performance. This is in good line with the aforementioned photodegradation experiment for LFX removal. It is well-known that the efficient separation and low recombination rate of e^-/h⁺ pairs result in higher transient photocurrent response. As shown in Figure 8(b), it can be easily observed that the rank of photocurrent intensities under light-on model is ZCC>CNC>ZnFe₂O₄, revealing that the ternary ZCC photocatalyst has the highest separation efficiency and lowest recombination for the photogenerated e^-/h⁺ pairs. Figure 8(c) depicts the electrochemical impedance (EIS) spectra of the prepared photocatalysts. Generally, the smaller EIS semicircle in the low-frequency region represents the lower chargers transfer resistance and faster carriers transfer rate, finally leading to declined e^-/h⁺ recombination. As can be seen, both of the monomers including ZnFe₂O₄ and g-C₃N₄ deliver bigger EIS arc than the composites, indicating the bigger carriers transfer resistance of the ZnFe₂O₄ and g-C₃N₄ monomers. The lowest EIS resistance of CNC with feature of the smallest arc is related to the intrinsic high conductivity of CQDs and g-C₃N₄. After integration of CNC with ZnFe₂O₄, the ternary ZCC possesses the slightly higher EIS resistance than CNC. Quite obviously, the highest resistance of ZnFe₂O₄ and the lowest resistance of CNC can mainly be responsible for the moderate resistance of ZCC. The catalytic activity of the catalyst mainly depended on the number of reactive sites on the surface of catalyst, which is closely related to the electrochemical active surface area (ECSA) of the catalyst. Based on the analyses on the measured CV curves in Figure S5, the C_DL values (Figure 8(d)) of g-C₃N₄, ZnFe₂O₄, CNC, and ZCC were calculated to be 14.79, 15.89, 15.99 and 17.69 mF/cm², finally obtaining the ECSA values of 0.369, 0.397, 0.399 and 0.442, respectively. Obviously, ZCC has the largest ECSA value, leading to more active sites for photocatalytic reactions. Mott-Schottky measurement was performed for investigating the physicochemical property of the prepared semiconductor photocatalysts and further unraveling the possible photocatalytic mechanism of PMS activation toward LFX removal. As shown in Figures 8(e) and (f), the positive slope represents n-type semiconductor characteristics of ZnFe₂O₄ and g-C₃N₄. By extrapolating the line to 1/C ²=0, the flat-band potential (E_fb) values of ZnFe₂O₄ and g-C₃N₄ can be determined by the values of the intersection with the X-axis, which is -0.35 eV (vs NHE) and 0.05 eV (vs NHE), respectively. It is well known that the conduction band (CB) value of n-type semiconductors is approximately equal to the E_fb value. Thus, the conduction band (CB) positions of ZnFe₂O₄ and g-C₃N₄ are -0.35 eV (vs NHE) and 0.05 eV (vs NHE), respectively. Combining the above-calculated E_g values of 1.91 eV (ZnFe₂O₄) and 2.72 eV (g-C₃N₄), the formula (E_VB=E_CB+E_g) is used to calculate the valence band potential values (E_VB) of ZnFe₂O₄ and g-C₃N₄, where E_VB represents the VB potential value (eV), E_CB is the CB potential value (eV), and E_g is the energy band energy (eV). The valence band (E_VB) values of ZnFe₂O₄ and g-C₃N₄ are 1.56 eV and 2.77 eV, respectively.

Figure 8全屏查看

(a) The PL spectra and (b) transient photocurrent response spectra of ZnFe₂O₄, CNC and ZCC; (c) The EIS spectra and (d) relationship of current densities and scan rates derived from CV curves for calculating C_DL values; The Mott-Schottky curves of (e) ZnFe₂O₄ and (f) CNC

Based on the band edge potentials of the g-C₃N₄ and ZnFe₂O₄ together with the confirmed reactive free radicals in the above experiments, the possible photodegradation mechanism of LFX coupled with the PMS activation by ZCC was presumably proposed and depicted in Figure 9. According to the estimated band positions of g-C₃N₄ and ZnFe₂O₄, two possible charge transfer models in ZCC heterojunction can be utilized to analyze the photocatalytic process of LFX through PMS activation, namely type II and Z scheme. Assume that the charge transfer path complies for the traditional type II heterojunction (Figure 9(a)). After absorbing photons by ZCC (Eq. (4)), the electrons will be excited from the VB to the CB of ZnFe₂O₄ and the holes will remain in its VB. Due to the more negative CB potential (-0.35 eV (vs NHE)) of ZnFe₂O₄ than that of g-C₃N₄ (0.05 eV (vs NHE)), the photogenerated electrons on the CB of ZnFe₂O₄ will transfer to CB of g-C₃N₄. Unfortunately, the photogenerated electrons on CB of g-C₃N₄ cannot react with the adsorbed O₂ to produce O₂^•– due to the fact that CB potential of g-C₃N₄ (0.05 eV (vs NHE)) is more positive than the redox potential of O₂/O₂^•– (-0.046 eV (vs NHE)), which vividly contradicts with the existence of O₂^•– detected in above free radical trapping experiment and ESR test. Likewise, the holes on the VB of g-C₃N₄ will move to the VB potential of ZnFe₂O₄. Since the VB potential value of ZnFe₂O₄ (1.56 eV (vs NHE)) is more negative than that of H₂O/·OH (2.27 eV), holes cannot oxidize H₂O into ·OH, which does not accord with ·OH formation confirmed in the above experiments. Apparently, the above-described type II heterojunction cannot successfully realize the degradation of LFX by ZCC. Conversely, as illustrated in Figure 9(b), after absorbing photons by ZCC (Eq. (6)), both of the electrons in the VBs of g-C₃N₄ and ZnFe₂O₄ will initially be excited to their CBs and then the electron in the CB of g-C₃N₄ will directly flow to the VB of ZnFe₂O₄ and quickly recombine with the holes in the VB of ZnFe₂O₄, thereby leading to improved photogenerated charges separation efficiency. During this process, the CQDs in ZCC can be utilized as a charge transfer bridge for accelerating the photogenerated electrons transfer due to its prominent conductivity, which is conducive to the improvement of photocatalytic activity of ZCC. Quite obviously, the CB potential (-0.35 eV (vs NHE)) of ZnFe₂O₄ is more negative than that of O₂/O₂^•– (-0.046 eV (vs NHE)), and thus it is feasible that the photogenerated electrons can react with O₂ to form O₂^•– (Eq. (7)). The VB of g-C₃N₄ possesses more positive potential in comparison with the H₂O/·OH (2.27 eV (vs NHE)) and the abundant holes in the VB of g-C₃N₄ can easily oxidize H₂O to produce ·OH (Eq. (8)). It can be concluded that the ZCC well fits the Z scheme heterostructure, which not only facilitates the charges separation and transmission, but also provides a high redox activity, hence enhancing the photocatalytic activity.

Figure 9全屏查看

Possible photocatalytic degradation mechanism of LFX by 0.2-ZCC in PMS system: (a) General type II heterojunction; (b) Z-scheme heterojunction

Apart from these, the PMS activation in this system will endow another extraordinary promotion role for further improving the photodegradation efficiency of LFX, which can be accomplished as follows. Initially, the photoexcited electrons in the CB of g-C₃N₄ can be captured by the PMS to form SO₄^•– (Eq. (9)), and then a part of SO₄^•– may react with H₂O to produce ·OH (Eq. (10)). Furthermore, based on the presence of Fe³⁺ in the ZCC revealed in XPS spectra (Figure 3(e)), it can be reasonably inferred that Fe³⁺ can be reduced to Fe²⁺ by O₂^•– (Eq. (11)) and also react with HSO₅^- to form Fe²⁺ and SO₅^•– (Eq. (12)). Subsequently, the resulted Fe²⁺ can also be oxidized into Fe³⁺ by reacting with HSO₅^-, finally leading to the formation of SO₄^•– (Eq. (13)). The results indicate that these reactions trigger the cycle of Fe³⁺ and Fe²⁺ redox, which benefits for the continuous and rapid production of huge free radicals. In addition, the free radical trapping experiment results unveil that non-free radical ¹O₂ also exists in the ZCC/PMS system. In fact, ¹O₂ can be formed by O₂^•– reacting with h⁺, H₂O, and ·OH (Eqs. (14)-(16)). Consequently, the efficient charge separation of the Z scheme ZCC heterojunction photocatalyst can provide abundant photogenerated electrons for PMS activation and promoting the redox cycle of Fe³⁺ and Fe²⁺, thus further rapidly activating PMS. Simultaneously, the recombination of photogenerated e^-/h⁺ pairs is hindered and finally the photocatalytic performance is improved. The synergistic effect between photocatalysis of ZCC and PMS activation is in favor for the consecutive production of more ROS in the reaction system. Finally, under the combined action of SO₄^•–, O₂^•–, ·OH, h⁺ and ¹O₂, LFX was degraded to CO₂ and H₂O (Eq. (17)) in ZCC and PMS catalytic systems.

2O₂^•-+2H₂O→¹O₂+H₂O₂+2OH^-

(15)

The possible intermediates of LFX photodegradation were identified by performing liquid chromatography-mass spectrometry (LC-MS) measurements. The obtained mass spectra of intermediates are shown in Figure S9 and Table S3 lists their corresponding molecular structure formula. On the basis of analyses on the above-discerned intermediates and the previous reports, five possible degradation pathways were reasonably inferred under the attack of reactive radicals (SO₄^•-, O₂^•-, ·OH, h⁺, ¹O₂) (Figure 10) [55-59]. In pathway I, LFX (m/z=361) was initially converted into P1 (m/z=335) owing to the decarboxylation process of the methylmorpholine group and then P2 (m/z=280) was generated due to the further oxidation of n-methylpiperazine ring in P1 (m/z=335). Sequentially, the intermediate P2 further suffered from dealkylation reaction to produce the intermediate P3 (m/z=252), which was then decomposed into P4 (m/z=165). In pathway II, P5 (m/z=318) was produced by the decarboxylation of LFX and further oxidized into P6 (m/z=278) by ring-opening and demethylation and then P7 (m/z=263) was gained by deamination of P6. In pathway III, P8 (m/z=318) was initially obtained due to the decarboxylation process of LFX. Subsequently, the products P9 (m/z=218) and P10 (m/z=212) are formed by demethylation of P8, in which the piperazine bond was destroyed and F was replaced with the OH^- group through the attack of the SO₄^•– and ·OH. In pathway IV, some short chain carboxylic acids were also produced due to the formation of reducing raw materials such as H₂. According to the natural valence orbital theory, pathway IV began in a heterocyclic ring with —C=C— breaking. The F atom was then replaced by a hydroxyl group and a carboxyl group was removed to produce P11 (m/z=338). Under the attack of generated radicals such as O₂^•-/∙OH, the —C—C— may be attacked by the resulting O₂^•–/∙OH, which gave rise to the ring-opening reaction of piperazine, finally resulting in the generation of intermediates P12 (m/z=240) and P13 (m/z=211). In pathway Ⅴ, the aldehyde product P14 (m/z=392) was produced by oxidation of LFX (m/z=362) and was further oxidized to convert into P15 (m/z=364). After that, P15 was changed into P16 (m/z=279) due to the removal of the piperazine group induced by the hydroxylation of the piperazine group. Then the intermediate P16 was consecutively degraded to form P17 (m/z=263). Finally, under the action of reactive radicals, the generated intermediate products during the degradation process continued to decompose and mineralize into small organic molecules, and completely mineralize into inorganic ions, CO₂ and H₂O under prolonged reaction time.

Figure 10全屏查看

Potential pathways of photodegradation for LFX

4 Conclusions

In summary, a novel Z scheme heterojunction ZnFe₂O₄/g-C₃N₄/CQDs was successfully developed by anchoring ZnFe₂O₄ on CQDs induced tubular-like g-C₃N₄ via microwave-assisted hydrothermal and solvothermal methods. In comparison with the neat ZnFe₂O₄ and pure g-C₃N₄, the prepared ZCC possesses a distinctive hollow tube morphology of g-C₃N₄ induced by CQDs, strengthened UV-vis light response ability, and ameliorated specific surface area and pore structure. Therefore, ZCC delivers a higher photocatalytic activity for activating PMS toward LFX degradation, delivering a removal rate of 95.3% at optimal operation conditions. The formation of a Z scheme heterojunction between ZnFe₂O₄ and induced g-C₃N₄ along with CQDs acting as the photogenerated charges (e^- and h⁺) transmission bridge, is the primary reason for the improved photocatalytic activity of ZCC toward LFX removal. This results in positive promotion role for the accelerated charges transfer and inhibited recombination of e^- and h⁺. SO₄^•– and O₂^•– play a predominant role and ·OH and ¹O₂ are the subordinate ROS in the degradation process of LFX. Liquid chromatography-mass spectrometry (LC-MS) technique was utilized to identify the intermediates to further propose the feasible degradation pathways of LFX. Consequently, the development of Z scheme photocatalyst along with CQDs induction is efficient strategy for improving the photocatalytic activity of resultant photocatalysts.