Heterogeneous UV–Fenton Process by Maize-Straw-Templated TiO₂/Fe₃O₄ for the Degradation of Tetracycline: Optimization Using Response Surface Methodology

Abstract

The heterogeneous Fenton-like catalysts TiO₂/Fe₃O₄ were fabricated using maize straw as template (MST-TiO₂/Fe₃O₄) by calcination followed by the hydrothermal method. The characterization showed that higher Fe₃O₄ particle dispersion, closer interaction between TiO₂ and Fe₃O₄, stronger electron transfer ability, and lower leaching of Fe ions of MST-TiO₂/Fe₃O₄ catalyst resulted in higher catalytic activity towards the degradation of tetracycline (TC) compared to pure Fe₃O₄. The best conditions for TC degradation were initial pH = 6.74, 11.52 mmol/L of H₂O₂, 0.38 g/L of MST-TiO₂/Fe₃O₄, and a reaction time of 56.63 min according to the response surface methodology (RSM) result based on the Box–Behnken design (BBD). The quadratic model was well-fitted to the experimental data with R2 (0.9843) and adj-R2 (0.9660) by the analysis of variance (ANOVA). Under the optimum reaction conditions, a maximum removal rate of 98.67% was achieved. The findings of the present study revealed that heterogeneous UV–Fenton process catalyzed by MST-TiO₂/Fe₃O₄ was a suitable way for the degradation of TC from aqueous environment.

1. Introduction

Antibiotic contamination is a potential threat to humans and the environment [1]. Tetracyclines (TCs) are the world’s second-most frequently used antibiotic that is commonly employed in livestock farming or added to livestock feed as growth promoters, and are designed to be stable and biorefractory [2,3]. Fenton-like and photocatalytic processes in advanced oxidation technology have shown great advantages. The essence of the Fenton-like reaction consists of Fe²⁺ and H₂O₂ reacting to generate high redox potential for the hydroxyl radial (•OH) and complete the cycle of Fe²⁺ and Fe³⁺. The heterogeneous Fenton technology that uses magnetic materials as catalysts has always attracted attention as a well-known example of AOPs. Compared to the homogeneous Fenton technology, the main advantages of this technology are as follows: no iron-containing sludge is produced, the suitable pH range is wider, the catalyst can be recycled, and the stability is better [4,5]. Fe₃O₄ is a type of iron oxide that has good magnetic properties; it is also non-toxic, reusable, and has a stable catalytic performance. The nanomodification of the catalyst is beneficial to increase its specific surface area and can indirectly increase catalytic activity. However, nano-sized Fe₃O₄ particles are easily agglomerated in the liquid phase, which affects its catalytic activity [6].

In addition, the catalytic activity of pure heterogeneous Fenton is relatively weak, because studies have shown that Fe³⁺/Fe²⁺ circulation rate on the catalyst surface under neutral or alkaline conditions is lower than that under acidic conditions. Considering that the photo-generated electrons generated by photocatalysis described in the previous section can promote the conversion of Fe³⁺ to Fe²⁺, the synergistic effect of photocatalysis and Fenton technology has a good oxidation effect on tetracycline. Therefore, Fe₃O₄ can be combined with TiO₂ to form a photocatalytic and heterogeneous Fenton system [7,8].

Orthogonal and single-factor tests are the more common experimental methods in the Fenton process [9]. However, these two methods are not able to analyze the interaction effects between the influencing factors, yet examining the interaction effects between factors is also the key to assessing the effect of each reaction parameter on the removal of pollutants by Fenton reaction. RSM is a method for the comprehensive analysis of test results using data obtained from a small number of tests [10]. It uses a reasonable experimental design to combine experimental, statistical, and mathematical models to fit a multiple quadratic regression equation function between the test factors and the test results, so as to analyze the influence of each test factor on the test results. Therefore, the RSM is used to design the experimental process, evaluate the individual and interaction effects of the process parameters, optimize the parameters with a minimum number of experiments, and predict the experimental results [11,12].

In this research, the MST-TiO₂/Fe₃O₄ composite catalysts were prepared by calcination followed by a hydrothermal method and TC was selected as a model pollutant to study the hybrid process involved in the UV–Fenton and TiO₂ photocatalytic oxidation of MST-TiO₂/Fe₃O₄. RSM was applied to model and optimize TC degradation efficiency based on four main parameters including initial pH, H₂O₂ dosage, catalyst dosage, and reaction time. This study did not merely synthesize an effective catalyst but also optimized the UV–Fenton process’ procedure, energy, and material usage.

2. Materials and Methods

2.1. Reagents and Chemicals

All chemicals used in this study were analytical grade reagents and were used without further purification. Tetracycline (TC) HCl (C₂₂H₂₄N₂O₈ HCl) was obtained from Sigma (Sigma–Aldrich Chemical Co., Beijing, China). Sodium thiosulfate pentahydrate (Na₂S₂O₃·5H₂O), ferrous sulfate (FeSO₄·7H₂O), sulfuric acid, sodium hydroxide (NaOH), and hydrogen peroxide 30% (H₂O₂) were purchased from Shenyang East China Reagent Plant. All solutions were prepared in deionized water.

2.2. Synthesis and Characterization of MST-TiO₂/Fe₃O₄

MST-TiO₂/Fe₃O₄ and MST-TiO₂ were prepared using our previously reported procedures [13] and Fe₃O₄ was obtained through the hydrothermal synthesis method.

The phase structure characterization of all powders was carried out on a Rigaku D/Max 2550 diffractometer using Cu Kα radiation (λ = 1.54056 nm) at a scanning speed of 10° min⁻¹ with the 2θ angle from 20° to 80°. The acceleration voltage and applied current were 3 kV and 20 mA, respectively. Microstructure and morphology of the synthesized samples were characterized by field emission scanning electron microscopy (FE-SEM, NOVA-230, FEI, Hillsboro, OR, USA). The chemical compositions were investigated by X-ray photoelectron spectroscope (XPS, Thermo Fisher Scientific 250Xi, Shanghai, China). Electrochemical impedance spectroscopy (EIS) was obtained by a computer-controlled impedance measurement unit (CHI760E, Shanghai Chenhua Instrument Co., Ltd., Shanghai, China). The ultraviolet-visible absorption properties of the samples were characterized using a Shimadzu U-4100 UV-Vis spectrophotometer. Barium sulfate was used as the reference standard, with a scanning wavelength range of 200–800 nm.

2.3. Experimental Procedure and Analysis

The batch UV–Fenton catalytic experiments were conducted in a 250 mL glass beaker using a magnetic agitator (750 rpm). A 10 w UVC-lamp (L: 220 mm, D: 80 mm, λ = 220–275 nm, Kongche, Shenzhen, China) was used as the UV source. H₂SO₄ and NaOH were added to adjust the pH of the solution to the desired value and the UV–Fenton reaction was initiated by adding the catalyst and H₂O₂ to the TC solution. At given time intervals, 4 mL of solution was collected and filtered immediately with 0.45 μm membrane filters for measurement. The RSM based on the BBD was used to analyze the relationship between the response (TC degradation rate) and four independent variables (initial pH, H₂O₂ dosage, catalyst dosage, and reaction time) with the Design-Expert Software (version 8.0.6).

The absorbance of TC was measured at 375 nm via an ultraviolet-visible spectrophotometer (752, Shanghai Shunyu Instrument Co., Ltd., Shanghai, China). The H₂O₂ concentration was determined at 400 nm on a UV-visible spectrophotometer by a potassium titanium oxalate method [14]. The amount of iron-leaching concentration was determined by the atomic absorption spectrophotometer (AAS, PE, Analyst 400, Springfield, IL, USA). Total organic carbon (TOC) was measured using a TOC-L series total organic carbon analyzer (Shimadzu Corporation, Kyoto, Japan). During the TC degradation process, 20 mL of water samples were collected at regular intervals. A specific amount of 10% sodium sulfite solution was added to each sample to quench residual H₂O₂, and after which the samples were placed into the continuous autosampler tray for analysis.

3. Result and Discussion

3.1. Characterization of MST-TiO₂/Fe₃O₄

The X-ray diffraction (XRD) pattern of TiO₂ shows a typical anatase phase of MST-TiO₂ (PDF No. 73-1764) in Figure 1. New characteristic peaks indexed to the magnetite phase with a cubic structure of Fe₃O₄ (PDF No.88-0866) were revealed from the MST-TiO₂/Fe₃O₄ sample.

Figure 1. XRD patterns of Fe₃O₄, TiO_2, and MST-TiO₂/Fe₃O₄.

XPS characterization is performed to reveal the chemical state of the elements in the MST-TiO₂/Fe₃O₄ sample and to identify the interaction between MST-TiO₂ and Fe₃O₄(Figure 2). The Ti 2p, Fe 2p and O 1s envelops of MST-TiO₂, Fe₃O₄ and MST-TiO₂/Fe₃O₄ are shown in Figure 2b–d and their deconvolution was performed by using a Gaussian–Lorentzian peak shape after performing a Shirley background correction. From the XPS survey spectrum of MST-TiO₂/Fe₃O₄ (Figure 2a), peaks of Fe 2p, O 1s, Ti 2p and C 1s were identified, which are in accordance with the binding energies of Fe, O, Ti and C, respectively. It should be noted that the C found in the samples resulted from the surface adventitious carbon from XPS instruments itself. In the Ti 2p XPS spectrum (Figure 2b), the two peaks at 464.2 eV and 458.4 eV of MST-TiO₂ correspond to the Ti 2p 1/2 and Ti 2p 3/2, respectively, while the peaks of MST-TiO₂/Fe₃O₄ shift 0.3 eV towards a lower binding energy. Also noticed from the Fe 2p XPS spectrum of Figure 2c is that the peaks at binding energy of 725.5 eV and 711.9 eV, which can be attributed to the Fe 2p 1/2 and Fe 2p 3/2, respectively, both consisted of Fe³⁺ (Fe₂O₃) and Fe²⁺ (FeO) peaks which are typical characteristics of Fe₃O₄ [15]. The Fe³⁺/Fe²⁺ ratio calculated from the peak areas is about 2:1, which is consistent with the characteristics of Fe₃O₄. The two peaks (Fe 2p 1/2 and 3/2) in the MST-TiO₂/Fe₃O₄ nanocomposites have a shift of 0.32 eV towards higher binding energy. In the O1s spectra of TiO₂, Fe₃O₄ and MST-TiO₂/Fe₃O₄ composites (Figure 2d), the O1s spectra of TiO₂ and Fe₃O₄ both contain lattice oxygen (O-Ti or O-Fe) and adsorbed oxygen (O-H), which are consistent with previous reports [16]. However, the lattice oxygen of the MST-TiO₂/Fe₃O₄ sample shifts towards higher energy in O1s spectrum. The down-shifts in Ti 2p peaks and the up-shifts in Fe 2p and lattice oxygen of MST-TiO₂/Fe₃O₄ can be traced to the change in the chemical environment arising from the close interaction between TiO₂ and Fe₃O₄ [17,18], which demonstrates the well-established Fe-O-Ti bond of MST-TiO₂/Fe₃O₄ [19]. This is because Fe₃O₄ enters the lattice of TiO₂, which reduces the binding energy of TiO₂, and this change makes TiO₂ in the MST-TiO₂/Fe₃O₄ sample easier to excite with light.

Figure 2. Survey and high-resolution XPS spectra of Fe₃O₄, TiO₂ and MST-TiO₂/Fe₃O₄. (a) Survey spectra; (b) Ti 2p spectra of TiO₂ and MST-TiO₂/Fe₃O₄; (c) Fe 2p spectra of Fe₃O₄ and MST-TiO₂/Fe₃O₄; (d) O 1s spectra of Fe₃O₄, TiO₂ and MST-TiO₂/Fe₃O₄.

The morphology of MST-TiO₂, Fe₃O₄ and MST-TiO₂/Fe₃O₄ particles were revealed by SEM as shown in Figure 3. MST-TiO₂ clearly displayed its honeycomb morphology and was composed of the MST-TiO₂ nanosheet (Figure 3a). Pure Fe₃O₄ exhibited nanoparticles with a size range of 150∼250 nm, and high-resolution SEM images of MST-TiO₂ reveal that the wall thickness of MST-TiO₂ nanosheets is approximately 150–200 nm. (Figure 3b), in accordance with the former report. However, the pure Fe₃O₄ particles were agglomerated due to the magnetic attraction between the grains. The SEM image of MST-TiO₂, Fe₃O₄ in Figure 3c shows that the smooth surface of MST-TiO₂ turned rough after Fe₃O₄ nanoparticles were loaded on it; meanwhile, Figure 3d obviously shows that the Fe₃O₄ nanoparticles have good dispersity on MST-TiO₂ plate. From the SEM image, we randomly selected some particles of Fe₃O₄ and statistically determined the average primary diameter of Fe₃O₄ nanoparticles to be 30∼50 nm (Figure 3d), which is much smaller than that of pure Fe₃O₄ nanoparticles (Figure 3b), and the image changed from nearly spherical polyhedron to cubic shape. This indicated that the assembling of Fe₃O₄ nanoparticles on MST-TiO₂ nanosheets can inhibit the growth of Fe₃O₄ particles, which is conducive to the formation of smaller particles of Fe₃O₄ during the crystallization process and has a certain dispersion effect.

Figure 3. SEM of (a) MST-TiO₂ (b), Fe₃O₄ (c), and MST-TiO₂/Fe₃O₄.

Furthermore, the electrochemical impedance spectrum (EIS) is employed to investigate the separation efficiency of the photo-excited charge carriers. A smaller impedance arc radius on the EIS Nyquist plot indicates a smaller electric resistance of the electrode [20,21]. Figure 4 shows the Nyquist curves of MST-TiO₂, Fe₃O₄ and MST-TiO₂/Fe₃O₄. The impedance arc radius for MST-TiO₂ and MST-TiO₂/Fe₃O₄ are much smaller than Fe₃O₄ from Figure 4. This implies a faster surface reaction rate with better charge transfer of MST-TiO₂/Fe₃O₄. For the UV–Fenton system, it can promote the conversion between Fe³⁺ and Fe²⁺, thereby improving catalytic reaction activity.

Figure 4. The EIS plots of MST-TiO₂/Fe₃O₄, MST-TiO₂ and Fe₃O₄ catalysts.

The ultraviolet and visible light absorption wavelengths of photocatalytic materials are key characteristics. Figure 5 shows the UV-vis diffuse reflectance spectra of the prepared MST-TiO₂ and template-free TiO₂ (N-TiO₂) across the 200 nm–800 nm wavelength range. As shown in Figure 5a, MST-TiO₂ and N-TiO₂ exhibit remarkably similar absorption spectra, both demonstrating strong absorption in the ultraviolet region (λ < 400 nm). This is primarily attributed to the bandgap energy (3.2 eV) excitation of the anatase TiO₂. The figure indicates that the absorption edge of the N-TiO₂ sample is approximately 387.96 nm, while that of MST-TiO₂ is red-shifted to 424.35 nm. The bandgap width of a semiconductor catalyst influences its absorption edge. As shown in Figure 5b, the Kubelka–Munk equation yields bandgap widths of 2.922 eV and 3.196 eV for MST-TiO₂ and N-TiO₂, respectively. The trend in bandgap width changes and aligns with that of the absorption edge. Specifically, as the bandgap width decreases, the absorption edge of the sample shifts toward the red end of the spectrum, thereby extending its visible light absorption capability. This indicates that elemental doping (e.g., C element) in the straw template effectively modifies the bandgap width of MST-TiO₂ samples, enhancing their photoresponsivity.

Figure 5. UV-vis diffuse reflectance spectra of MST-TiO₂ and N-TiO₂ catalysts. (a) UV–vis DRS spectra, (b) Bandgap width.

3.2. UV–Fenton Catalytic Activity for the Degradation of TC

3.2.1. The Degradation Performance of TC in Different Reaction Systems

To validate the UV–Fenton catalytic activity of the samples, several advanced oxidation experiments were performed for TC degradation and mineralization under various conditions (Figure 6a,b). Under dark conditions, it appeared that approximately 5% TC (negligible TOC removal rate) could adsorb to the surface of MST-TiO₂/Fe₃O₄ in 90 min. The degradation rate of TC in the UV/H₂O₂ system after 90 min was 69.34%. Additionally, in the absence of H₂O₂, the activity of UV/MST-TiO₂ under UV light irradiation had weak ability in terms of catalytic degradation and mineralization of TC under the tested conditions. It is, however, not surprising that Fe₃O₄/H₂O₂ without UV irradiation led to a slight TC degradation of 20% after 90 min (1.8% for TOC removal rate in 2 h). Once UV–Fenton catalysts were introduced, degradation of TC obviously increased, and faster TC degradation contributed to faster ·OH production from the reaction between H₂O₂ and UV–Fenton catalysts. As for pure Fe₃O₄, approximately 89.2% degradation of TC occurred after 90 min and when 37.5% TOC removal rate in 2 h was reached. As for the UV/MST-TiO₂/Fe₃O₄/H₂O₂ system, when compared with pure Fe₃O₄, the content of Fe₃O₄ was less under the same catalyst dosage, but the TC tended to be almost completely degraded after 60 min of reaction, and the mineralization rate was 63.5% after 120 min. In contrast, the removal rate of UV/Fe₃O₄/H₂O₂ system was only 79.8% after 60 min of reaction. This indicates that the MST-TiO₂/Fe₃O₄ formed after the introduction of TiO₂ can greatly enhance the catalytic activity of the UV–Fenton system.

Figure 6. Degradation effect of tetracycline under different reactions. (a) Degradation efficiency and (b) mineralization. Experimental conditions: initial pH = 7.0, [catalyst]₀ = 0.3 g/L, [H₂O₂]₀ = 10 mM, [TC]₀ = 50 mg/L and T = 25 ± 1 °C.

3.2.2. H₂O₂ Consumption

It is well known that H₂O₂ is the oxidant in Fenton reaction, and the study of H₂O₂ consumption is essential to evaluate UV–Fenton catalytic efficiency. Figure 7 depicts the consumption of H₂O₂ by samples as a function of reaction time under various conditions. As seen from Figure 6, the consumption of H₂O₂ in the multiphase Fenton system catalyzed by Fe₃O₄ with no light irradiation under neutral conditions is extremely low, and the decomposition rate of H₂O₂ at 60 min of the reaction is 14.1%. While in the UV/H₂O₂ system, the decomposition rate of H₂O₂ was slightly increased, and the decomposition rate was 54.2% at 60 min of the reaction. When Fe₃O₄ was introduced, the decomposition rate of H₂O₂ in the multiphase UV–Fenton system increased rapidly, and the decomposition rate at 60 min of the reaction increased to 82.4%. With the introduction of MST-TiO₂/Fe₃O₄, the decomposition rate of H₂O₂ was further increased to 92.9%. This result is consistent with the degradation effect test of TC, and the main reason for this is that TiO₂ transfers photogenerated electrons to Fe³⁺ on the surface of Fe₃O₄ to promote its conversion to Fe²⁺, thus indirectly accelerating the rate of Fenton reaction, which results in the increase in the decomposition rate of H₂O₂.

Figure 7. H₂O₂ consumption under different processes. Experimental conditions: initial pH = 7.0, [catalyst]₀ = 0.3 g/L, [H₂O₂]₀ = 10 mM, [TC]₀ = 50 mg/L and T = 25 ± 1 °C.

3.2.3. Leaching of Fe Ions

The leaching of active components is an important indicator to examine the stability of catalysts, so the changes in Fe ions in UV/MST-TiO₂/Fe₃O₄/H₂O₂ and UV/Fe₃O₄/H₂O₂ systems were investigated, and the results are shown in Figure 8. As seen from Figure 8, the concentration of Fe ions in both UV/MST-TiO₂/Fe₃O₄/H₂O₂ and UV/Fe₃O₄/H₂O₂ systems showed a trend of first increasing and then decreasing with an increase in reaction time. It is mainly because the Fe ions on the catalyst surface will be captured by the intermediate metabolites of TC degradation and form complexes, which will be free in the solution, thus making the concentration of leached Fe ions in the system increase. However, as the reaction time continues to increase, the complexes of intermediate metabolites and Fe ions will be gradually mineralized to CO₂ and H₂O, while Fe ions will be released. Therefore, the leaching amount of Fe ions showed a trend of increasing and then decreasing during the reaction. In addition, the amount of Fe ions leached in both reaction systems is relatively small, with the highest leaching amounts being 0.46 mg/L and 1.23 mg/L, respectively, whereas the amount of Fe ions required in a typical homogeneous Fenton reaction is tens to hundreds of mg per liter [22,23], indicating that the degradation of TC in both reaction systems is caused by non-homogeneous Fenton. In addition, the figure shows that Fe-ion concentration in the UV/MST-TiO₂/Fe₃O₄/H₂O₂ system is always lower than that in the UV/Fe₃O₄/H₂O₂ system during the reaction, and the residual Fe-ion concentration in the system after 180 min was 0.3 mg/L (0.42% by weight) and 0.81 mg/L (0.48% by weight), respectively. From the results, it can be concluded that the introduction of TiO₂ can improve the stability of MST-TiO₂/Fe₃O₄ composite catalyst.

Figure 8. Fe-leaching amount of UV/MST-TiO₂/Fe₃O₄/H₂O₂ and UV/Fe₃O₄/H₂O₂ in the solution. Experimental conditions: initial pH = 7.0, [catalyst]₀ = 0.3 g/L, [H₂O₂]₀ = 10 mM, [TC]₀ = 50 mg/L and T = 25 ± 1 °C.

3.2.4. Effect of Single-Factor on TC Oxidation in MST-TiO₂/Fe₃O₄-Catalyzing UV–Fenton

The solution pH of a Fenton reaction is a crucial factor, so the effect of initial pH from 3.0 to 9.0 on TC oxidation by MST-TiO₂/Fe₃O_4-catalyzing UV–Fenton was studied and the results are presented in Figure 9a. At the initial pH values of 1.0, 3.0, 5.0, 7.0, and 9.0, the degradation rate of TC followed a trend of increasing and then decreasing with the increase in pH, and the removal rates were 87.2%, 95.5%, 98.1%, 99.2%, and 68.7%, respectively. Although the catalysts are effective in the pH range of 1 to 9, the most suitable pH range is from 3.0 to 7.0, with the highest degradation rate at pH 7.0 (99.2%). It can therefore be interpreted that pH affects the catalyst activity and the Fenton-like synergistic reaction mainly due to the generation of free radicals, especially •OH, and the adsorption properties of the catalyst on TC. On the one hand, the rate-limiting step in the Fenton-like system is the Fe³⁺/Fe²⁺ cycle, so it is especially important to accelerate the conversion of Fe³⁺ to Fe²⁺. When the pH is too low, there is a large amount of H⁺ in the solution, so the reduction of Fe³⁺ to Fe²⁺ (Equations (2) and (3)) can be inhibited. Moreover, the excess H⁺ is trapped by H₂O₂ in the system to produce stable H₃O₂⁺ (Equation (4)). In addition, •OH is scavenged by H⁺ and e⁻ (Equation (5)). This explains the superior oxidation capacity in the UV–Fenton system at pH 5.0 and 7.0 compared to pH 3.0. But when the pH is too high, it will affect the •OH formation in Equation (1), while Fe³⁺ will precipitate as hydroxides and reduce the catalytic performance, and the colloid formed will also hinder the penetration of light. On the other hand, the pH also affects the charge on the surface of TiO₂ and its photoexcitation properties [24]. The isoelectric point of TiO₂ in water is approximately 6.5 [25]. When the pH is low, the surface of TiO₂ has a positive charge dominated by H⁺, and it has a negative charge dominated by OH⁻ at high pH [26], and when the pH is 7.0 the TiO₂ surface has a small amount of negative charge. Under medium and acidic conditions, TC is positively charged in cationic form; and TiO₂, with a small amount of negative surface charge, will have a positive effect on the adsorption of TC. In contrast, under alkaline conditions, TC is negatively charged and TiO₂ also has a large amount of negative charge on its surface, which is not favorable for adsorption [9]. Theoretically, the best adsorption performance of TiO₂ on TC is achieved at pH 6.5∼7.0. The mechanism of TiO₂ photocatalysis is the reaction of h⁺ with H₂O to form •OH (Equation (4)), but this reaction is only effective when it is transported to the catalyst surface. Acidic conditions are not conducive to the reaction of h⁺ with H₂O and reduce the transport of h⁺. At pH 7, the TiO₂ surface is almost free of H⁺ and other positive charges and thus does not affect h⁺ transport. A small amount of OH⁻ promotes the production of •OH (Equation (5)) and does not unduly affect the production and transport of photogenerated e⁻.

Fe²⁺ + H₂O₂ → Fe³⁺ + •OH + OH⁻

(1)

Fe³⁺ + H₂O₂ → •HO₂ + Fe²⁺ + H⁺

(2)

$${\mathrm{Fe}}^{3+} + {\mathrm{H}}_2\mathrm{O} + \mathrm{h}\mathsf{\nu } \to {\mathrm{Fe}}^{2+} + •\mathrm{OH} + {\mathrm{H}}^{+}$$

(3)

h⁺ + H₂O → •OH + H⁺

(4)

h⁺ + OH⁻ → •OH

(5)

H₂O₂ + H⁺ → H₃O₂

(6)

•OH + e⁻ + H⁺ → H₂O

(7)

Figure 9. Effect of different parameters on the removal rate of tetracycline: (a) initial pH, (b) H₂O₂ dosage, (c) catalyst dosage, and (d) reaction time on the removal of TC.

Figure 9b shows the effect of H₂O₂ dosage on the UV–Fenton reaction with MST-TiO₂/Fe₃O₄. When the dosage of H₂O₂ was 5 mM, 10 mM and 20 mM, the removal rates of TC were 55.2%, 74.1% and 79.9%, respectively. In contrast, the removal rate of TC decreased slightly (77.9%) when the H₂O₂ dosage was increased to 30 mM. This is because within a certain dosage range, the more H₂O₂ is added, the more •OH is generated and the greater the oxidation efficiency. However, when the dosage of H₂O₂ is too high, it intensifies the side reaction of quenching •OH by H₂O₂ (Equations (1)–(3)), which decreases the degradation efficiency [4,27].

Figure 9c shows the effect of catalyst dosage on UV–Fenton reaction. It can be seen from the figure that the increase in the dosage of MST-TiO₂/Fe₃O₄ from 0.1 g/L to 0.5 g/L can significantly improve the removal rate of TC, because the increase in catalyst dosage also increases the area of Fenton reaction, which increases the active sites on the catalyst surface and in turn increases the yield of •OH [12]. However, the increase in TC removal was not significant as the catalyst dosage continued to increase, because the amount of H₂O₂ became a limiting factor for the reaction when the catalyst concentration reached saturation. On the other hand, there are some negative effects of high doses of the catalyst, such as nanoparticle agglomeration and scattering effects on light [28].

Figure 9d shows the influence of reaction time on the degradation rate of TC. The degradation of TC went through two different phases: fast degradation phase (0–20 min); slow degradation phase (20–90 min). The reaction at 60 min basically reached the maximum TC removal rate.

3.2.5. Effect of Initial TC Concentration

The effect of initial TC concentration on its degradation rate was investigated. As shown in Figure 10. Under identical reaction conditions, increasing the TC concentration from 25 mg/L to 100 mg/L resulted in removal rates consistently exceeding 90% after the reaction commenced.

Figure 10. Effect of initial concentration on degradation. Experimental conditions: initial pH = 7.0, [catalyst]₀ = 0.3 g/L, [H₂O₂]₀ = 10 mM and T = 25 ± 1 °C.

3.2.6. Optimization of the TC Degradation Using RSM

RSM is an accumulation of a statistical and mathematical method employed to analyze the importance of different process variables [29]. The Design-Expert Software (version 8.0.6) was used to analyze the simultaneous effect of initial pH, H₂O₂ dosage, catalyst dosage, and the reaction time in the UV–Fenton degradation process. It is also used to evaluate the interactions between the studied variables and calculate the predicted results with the experimental design [10,12]. Therefore, the four factors could be identified with three levels (−1, 0, +1) (Table 1) and a series of experimental runs were designed with TC degradation efficiency as the response, as shown in Table 2.

Table 1. Factor levels for the experiment.

Levels	Factors
	Initial pH	H2O2 Dosage (mmol/L)	Catalyst Dosage (g/L)	Reaction Time (min)
−1	5	5	0.1	45
0	7	10	0.3	60
1	9	15	0.5	75

Table 2. Experimental design and results based on response surface methodology.

Run	A	B	C	D	Degradation Rate (%)
1	−1 (5)	0 (10)	0 (0.3)	1 (75)	88.05
2	0 (7)	0 (10)	−1 (0.1)	−1 (45)	77.57
3	−1 (5)	1 (15)	0 (0.3)	0 (60)	89.62
4	1 (9)	0 (10)	1 (0.5)	0 (60)	63.47
5	1 (9)	1 (15)	0 (0.3)	0 (60)	65.76
6	0 (7)	1 (15)	0 (0.3)	1 (75)	97.67
7	0 (7)	0 (10)	1 (0.5)	1 (75)	99.79
8	1 (9)	−1 (5)	0 (0.3)	0 (60)	34.07
9	0 (7)	−1 (5)	1 (0.5)	0 (60)	53.43
10	0 (7)	−1 (5)	−1 (0.1)	0 (60)	47.68
11	0 (7)	0 (10)	0 (0.3)	0 (60)	97.25
12	1 (9)	0 (10)	0 (0.3)	−1 (45)	59.66
13	0 (7)	1 (15)	−1 (0.1)	0 (60)	96.71
14	−1 (5)	−1 (5)	0 (0.3)	0 (60)	51.91
15	−1 (5)	0 (10)	−1 (0.1)	0 (60)	79.18
16	−1 (5)	0 (10)	1 (0.5)	0 (60)	81.86
17	0 (7)	0 (10)	0 (0.3)	0 (60)	95.41
18	0 (7)	−1 (5)	0 (0.3)	1 (75)	57.53
19	0 (7)	1 (15)	0 (0.3)	−1 (45)	85.34
20	0 (7)	−1 (5)	0 (0.3)	−1 (45)	44.08
21	1 (9)	0 (10)	−1 (0.1)	0 (60)	60.04
22	0 (7)	0 (10)	1 (0.5)	−1 (45)	83.75
23	1 (9)	0 (10)	0 (0.3)	1 (75)	71.77
24	0 (7)	1 (15)	1 (0.5)	0 (60)	98.49
25	−1 (5)	0 (10)	0 (0.3)	−1 (45)	76.92
26	0 (7)	0 (10)	−1 (0.1)	1 (75)	93.2
27	0 (7)	0 (10)	0 (0.3)	0 (60)	96.87

Table 2 shows the obtained actual responses (i.e., TC degradation efficiency) with respect to the 27 designed experiments. After model fitting by the software, the second-order polynomial equations for the coding factors and the actual results are shown in Equations (8) and (9), respectively, as a simultaneous function of the initial pH value (A), the H₂O₂ dosage (B), the catalyst dosage (C), and the reaction time (D) [30].

R₁ = 96.51 − 9.40A + 20.41B + 2.20C + 6.72D − 1.51AB + 0.19AC + 0.25AD − 0.99BC − 0.28BD + 0.10CD − 18.70A² − 18.70B² − 4.59C² − 4.57D²

(8)

R₂ = −339.6266 + 61.61688A + 20.61675B + 84.44792C + 2.85553D − 0.1505AB + 0.46875AC + 8.16667 × 10⁻³ AD − 0.9925BC − 3.73333 × 10⁻³ BD + 0.034167CD − 4.67438A² − 0.748B² − 114.75C² − 0.020311D²

(9)

The analysis of variance (ANOVA) was used to identify the adequacy of the model and its importance (Table 3). The results show that the model has a high F-value of 53.83 and low p-value (less than 0.0001), indicating that the model was highly statistically significant for TC degradation [5]. The lack of fit p-value was 0.0595 (>0.05), indicating that the lack of fit was not significant and that model was reasonably predictable. The fit of the model was verified by the determination of coefficient (R²). The closer the R² is to 1, the stronger the model and the better it predicts the response. For TC degradation, the coefficients and adjusted coefficients were 0.9843 and 0.9660, respectively, indicating that the model response was in good agreement with the experimental values. The coefficient of variance (C.V.) for TC degradation was 4.73%, demonstrating a high degree of precision and reliability of the model with the experimental values. In addition, the value of Adeq precision reached 25.891 (>4), which revealed that the model was well-fitting and that the-signal-to-noise ratio was satisfactory. It was also found that the p-values of the linear coefficient terms A, B, C, and D are all less than 0.05, with small p-values (p < 0.001) for A, B, and D. This indicates that each of the four factors had statistically significant changes in the response value, which is consistent with the results of the single-factor test.

Table 3. Variance analysis of regression equation.

Source	Sum of Squares	df	Mean Square	F-Value	p-Value
Model	9711.56	14	693.68	53.83	<0.0001	Significant
A	1059.76	1	1059.76	82.23	<0.0001
B	4997.59	1	4997.59	387.79	<0.0001
C	58.12	1	58.12	4.51	0.0452
D	542.57	1	542.57	42.1	<0.0001
AB	9.06	1	9.06	0.7	0.4182
AC	0.14	1	0.14	0.011	0.9185
AD	0.24	1	0.24	0.019	0.8937
BC	3.94	1	3.94	0.31	0.5905
BD	0.31	1	0.31	0.024	0.8786
CD	0.042	1	0.042	0.00326	0.9554
A2	1864.51	1	1864.51	144.68	<0.0001
B2	1865.01	1	1865.01	144.72	<0.0001
C2	112.36	1	112.36	8.72	0.0121
D2	111.39	1	111.39	8.64	0.0124
Residual	154.65	12	12.89
Lack of fit	152.76	10	15.28	16.19	0.0595	Not significant
Pure error	1.89	2	0.94
Core total	9866.21	26

C.V. = 4.73%; Predicted R² = 0.9843; Adjusted R² = 0.966; Adeq Precision = 25.891.

The positive and negative signs of the model coefficients of the linear coefficients indicate that the coefficients have a facilitating or inhibiting effect on the response values, and the absolute value of the coefficients of the model coefficients of the linear coefficients indicates the degree of the facilitating or inhibiting effect of the coefficients on the response values. The order of their influence on the response values is as follows: H₂O₂ dosage (20.41) > initial pH (9.40) > reaction time (6.72) > catalyst dosage (2.20). It is worthy noting that under the UV–Fenton system catalyzed by MST-TiO₂/Fe₃O₄ with the experimental set parameters, H₂O₂ dosage can change the TC degradation effect by influencing the reaction between Fe²⁺ and H₂O₂ and the photolysis of H₂O₂, and then the generation of •OH, so the precise and strict control of H₂O₂ dosage is more important. In addition, the pH of the system not only determines the redox potential of •OH, but also affects the surface electrical and adsorption properties of the catalyst, the presence of Fe²⁺, and the status and number of active sites, as well as the quantity and status of each reaction material in the solution. Therefore, the control of the initial pH also makes an important change in the reaction process. Again, the promotion of the reaction time on the degradation performance of TC is due to the fact that the UV–Fenton degradation reaction occurs positively with increasing time, and this reaction process is basically irreversible and fitted as a pseudo primary reaction, while the reaction time is explored as a more important reaction parameter. However, the factor with the least degree of influence on the response values was the catalyst dosage, which is due to the fact that the different catalyst dosages in this system had little effect on the reaction results, and the catalyst played a significant catalytic role as a reaction condition without the need for a large dosage, indicating the high efficiency of the catalyst.

The p-values of quadratic terms A², B², C², and D² are all less than 0.05, indicating that they all had statistically significant effects on the changes in response value. The interaction terms AB, AC, AD, BC, BD, and CD, all had p-values greater than 0.05, which illustrated the impact of the interaction of the six sets of relationships (between the initial pH and H₂O₂ dosage, initial pH and catalyst dosage, initial pH and reaction time, H₂O₂ dosage and catalyst dosage, H₂O₂ dosage and reaction time, and catalyst dosage and reaction time) on the degradation rate of TC. Meanwhile, the initial pH, H₂O₂ dosage, catalyst dosage, and reaction time have no significant antagonistic and synergistic effects with each other, and possess a certain degree of independence, which proves the feasibility and significance of the single-factor experiment, and the model’s wider significance of use.

To interpret the deviation between the actual and predicted responses, the comparison curves of the actual and predicted values were plotted, and the results are shown in Figure 11a. The results show that the 27 points of data fit well with the comparison baseline, and the deviation between the predicted and actual values is small, which further proves that the regression model fits the experimental points well. Figure 11b shows the plot of the normal probability distribution of the residuals of the response surface, from which it can be seen that the residuals of all the experimental results are basically distributed in a straight line. In addition, from Figure 11c, it observed that there is no regular distribution relationship between the residuals and the predicted values, indicating that the residuals are uniformly distributed, which proves the reliability of the quadratic regression model for the degradation rate of TC.

Figure 11. Normal probability plot of degradation of TC. (a) Predicted vs. actual. (b) Normal plot residuals. (c) Residuals vs. predicted.

According to the experimental parameter optimization function of the Design-Expert (Version 8.0.6) software, the four parameters of initial pH, H₂O₂ dosage, catalyst dosage, and reaction time were optimized, and the best degradation conditions were determined with the maximum degradation rate of TC and the shortest reaction time. The optimum degradation conditions were determined as follows: the initial pH value was 6.74, the H₂O₂ dosage was 11.52 mmol/L, and the catalyst dosage was 0.38 g/L. The resulting reaction time was 56.63 min, and the predicted TC degradation rate was 98.13%. The experimental results are shown in Figure 12. The deviation between the experimental value (98.67%) and the predicted value (98.13%) of TC removal rate is small, and almost complete removal of TC can be achieved in 58–60 min. As observed from this study, the established regression model is more accurate in predicting the performance of UV–Fenton degradation of TC catalyzed by MST-TiO₂/Fe₃O₄ materials, and the optimal operating conditions in the experimental range were derived.

Figure 12. TC degradation effect under optimized conditions.

3.2.7. Catalytic Mechanism

Based on the above results, the UV–Fenton system catalyzed by MST-TiO₂/Fe₃O₄ exhibits synergistic effects between photocatalysis and UV/Fenton in the degradation of TC, with the synergistic mechanism illustrated in Figure 13. The degradation of TC in the reaction system is primarily driven by •OH radicals, with the UV–Fenton process serving as the main pathway for •OH generation (Equation (1)). Additionally, UV/H₂O₂ (Equation (10)) and photocatalysis (Equations (4) and (11)) also contribute to •OH production. On one hand, upon UV irradiation, electrons in the valence band of MST-TiO₂ are excited to the conduction band, forming photo-generated electrons and holes in the conduction and valence bands, respectively. Photo-generated holes react with H₂O to produce •OH, while photo-generated electrons react with O₂ to generate O₂•⁻. On the other hand, Fe³⁺ generated in the UV–Fenton system acts as an electron acceptor, capturing photoelectrons produced by UV-irradiated MST-TiO₂. This not only suppresses recombination of photo-generated holes and electrons but also accelerates the conversion of Fe³⁺ to Fe²⁺ in the system (Equation (12)). Consequently, the oxidative capacity of the Fenton reaction is enhanced, promoting the generation of more free radicals in the system, thereby enhancing the degradation efficiency of TC (Equation (13)).

$${\mathrm{H}}_2{\mathrm{O}}_2 \stackrel{hv}{\to } •\mathrm{OH}$$

(10)

$$\mathrm{MST}-{\mathrm{TiO}}_2 \stackrel{hv}{\to } {\mathrm{h}}^{+} + {\mathrm{e}}^{-}$$

(11)

Fe³⁺ + e⁻ → Fe²⁺

(12)

TC + •OH → Products

(13)

Figure 13. The reaction mechanism of MST-TiO₂/Fe₃O₄.

4. Conclusions

In this study, the XRD, XPS, SEM, and EIS analysis confirmed that MST-TiO₂/Fe₃O₄ composite catalyst was successfully synthesized by calcination followed by a hydrothermal method. Compared to pure Fe₃O₄, the introduction of MST-TiO₂ in MST-TiO₂/Fe₃O₄ not only prevents the agglomeration of Fe₃O₄ particles, but also significantly improves electron transfer ability which contributed to higher catalytic activity towards the degradation of TC in the UV–Fenton system. The RSM based on BBD was employed to model and predict the effects of initial pH, H₂O₂ dosage, catalyst dosage, and reaction time on TC degradation rate. The predicted degradation efficiency of 50 mg/L TC reached 98.13% and matched well with the experimental values (98.67%) under optimum conditions. Moreover, the initial pH, catalyst dosage, H₂O₂ dosage, and reaction time all significantly affected the catalytic efficiency of UV–Fenton degradation of TC, and the order of significance was as follows: H₂O₂ dosage > initial pH > reaction time > catalyst dosage. The interaction between the influencing factors had no significant effect on the degradation rate of TC. However, the initial pH has an effect on the degradation of TC by the UV/MST-TiO₂/Fe₃O₄/H₂O₂ system, primarily because it determines the presence of H₂O₂, •OH, Fe³⁺/Fe²⁺ cycle in the system and the surface charge and photoexcitation properties of the TiO₂ catalyst.