Optimization of ciprofloxacin removal from aqueous solutions using the TiO2/Fe3O4/UV process and toxicity assessment via resazurin reduction test Authors: Mansour Ghaderpoori, Maryam Yaghobi, Faramarz Azimi, Bahram Kamarehie, Ali Toolabi Published: 2026 ABSTRACT AbstractCiprofloxacin (CIP) is a widespread environmental pollutant due to its high chemical stability and frequent presence in aquatic environments. Even at low concentrations, it poses serious risks to human health and ecosystems. This study aimed to optimize the photocatalytic degradation of CIP and evaluate the toxicity of the treated effluent from aqueous solutions. The experiments were conducted in batch mode using a 3 L photocatalytic reactor. Experimental optimization was performed using the Central Composite Design (CCD), considering pH, contact time, CIP concentration, and nanoparticle dose as variables. TiO2/Fe3O4 nanocomposites were synthesized via the sol-gel method. Their structure, morphology, and elemental composition were analyzed using Scanning Electron Microscopy (SEM), Fourier Transform Infrared Spectroscopy (FTIR), X-Ray Diffraction (XRD (and Energy Dispersive X-Ray Spectroscopy (EDX). CIP residual concentrations were measured using High-Performance Liquid Chromatography )HPLC( according to the experimental design. Effluent toxicity was assessed using a resazurin colorimetric assay based on Escherichia coli activity. The results of the study showed that the synthesized nanoparticles were spherical, with a homogeneous size distribution below 100 nm, smooth surface morphology, and no observable aggregation, indicating successful synthesis of structurally stable and well-dispersed nanoparticles. Based on the CCD model, the optimal conditions were a contact time of 67.50 min, pH 7.50, an initial CIP concentration of 1 mg/L, and a nanocomposite dose of 255 mg/L. Under these conditions, CIP removal reached approximately 95%. ANOVA indicated that a second-order regression model with a high coefficient of determination (R2 = 0.9895) best fit the data. The untreated CIP solution exhibited high toxicity toward E. coli, with a Half Maximal Effective Concentration (EC50) of 1.30 mg/L. After photocatalytic treatment, EC50, 100% mortality concentration, and No Observed Effect Concentration (NOEC) in the treated effluent were 1.37, 2.27, and 0.92 mg/L, respectively; the reduction in toxicity was statistically significant (p < 0.05). These results indicate that the synthesized TiO2/Fe3O4 nanocomposite, combined with UV-C ) power = 125 W, intensity = 20 mW/cm2, λmax = 254 nm) radiation, is an effective and stable method for treating wastewater contaminated with antibiotics. The integration of effluent toxicity assessment with experimental modeling (CCD and statistical analysis) represents a novel aspect of this study. This approach has the potential to be scaled up for large-scale environmental applications and to ensure compliance with effluent discharge standards. However, several limitations, including dependence on operational conditions, the formation of potentially toxic intermediate compounds, and challenges in scalability, require further investigation.Keywords: Ciprofloxacin, Photocatalytic processes, Optimization, Resazurin, E. coliSubject terms: Chemistry, Environmental sciences Introduction The rapid population growth and the expansion of industrial activities have led to the pollution of water resources1–3. Water pollution can result from heavy metals, dyes, pharmaceuticals, pesticides, antibiotics, organic compounds, and other hazardous substances, even at very low concentrations4–7. During the past decades, the fate and risks of antibiotics residues in aquatic environments as emerging pollutants have caused concern to communities4,8,9. The consumption of antibiotics in the world is estimated between 100,000 and 200,000 tons per year. Of this amount, about 50% is consumed by humans10. Antibiotics are not completely metabolized in the body, and 30–90% of them are excreted in an active form. Consequently, it is estimated that 100,000 to 200,000 tons of antibiotics enter the environment annually. Among these, fluoroquinolones represent a class of broad-spectrum synthetic antibacterial agents that are extensively employed in aquaculture, agriculture, and the production of human and veterinary pharmaceuticals10,11. These compounds are released into the environment predominantly through human and veterinary consumption, as well as via wastewater and pharmaceutical effluents. As a result, human exposure may occur through contaminated water, food, or occupational contact. Comprehensive guidelines have been established by the United States Environmental Protection Agency (EPA) to assess human exposure, encompassing data evaluation, exposure modeling, and uncertainty analysis12. Likewise, the World Health Organization (WHO) stresses the prudent use of antibiotics and advocates for strategies aimed at minimizing environmental contamination associated with pharmaceutical production and veterinary applications13. Ciprofloxacin (CIP) is one of the most commonly prescribed fluoroquinolones for the treatment of bacterial infections worldwide. Unfortunately, more than 95% of the administered dose is excreted from the body as active metabolites, ultimately entering municipal wastewater14,15. In recent years, researchers have reported widespread use of CIP in Switzerland16–18, China19,20, the United States21, Australia9, and other countries22,23. The presence of CIP in aquatic environments may pose serious threats to the ecosystem, human health, and interference in the biological treatment process of wastewater14,24. According to previous studies, the concentration of CIP in surface water, hospital effluents, has been estimated at 529 ng/g 25. The presence of CIP, even at low concentrations, is harmful and can cause side effects such as diarrhea, mental disorders, allergic reactions (including skin rash and swelling), shortness of breath, and general weakness26,27. American environmental standards have specified that the concentration of antibiotics in the effluent should be less than 1 mg/L 28. Various methods have been proposed to remove the antibiotic CIP from aqueous solutions. Common methods of CIP removal include photolysis by ultraviolet light29, adsorption30, activated carbon coated with zinc hydroxide nanoparticles31, composites nanostructure by ultrasonic waves32, natural carbon33, use of various photocatalytic processes29,34,35, and magnetic nano-biocomposites36. According to the studies conducted, less than 10% of CIP is removed in conventional wastewater treatment37. Meanwhile, advanced oxidation processes (AOPs) are highly effective methods for removing antibiotics from aqueous solutions31,38,39. AOPs decompose pollutants through the generation of hydroxyl radicals. When electrons (e−) and protons (H+) are produced on the catalyst surface, H+ interacts with water molecules to generate hydroxyl radicals, while e− reacts with dissolved oxygen to form superoxide ions (O2−•). Subsequent reactions with water molecules produce hydroxide ions and hydroperoxyl radicals (•OOH) (Fig. 1). Schematic illustration of electron excitation and the degradation pathways of CIP through the TiO2/Fe3O4/UV-C process, this figure was created by authors using paint v. 2024. In the next step, these hydroperoxyl radicals react with H+ ions to produce additional hydroxyl radicals (•OH). •OH are produced by the following reactions (Eqs. 1–5) 28,40–45. TiO2/Fe3O4 nanocomposites offer a balanced combination of strong photocatalytic performance and easy magnetic recovery46. The magnetic core facilitates quick separation of the catalyst after use, while the TiO2 shell provides high photoactivity and durability. This synergy enhances reaction efficiency, minimizes catalyst loss, and supports repeated applications in aqueous solutions. Following the photocatalytic degradation of CIP, it is critical to evaluate the residual toxicity of the reactor effluent from an environmental standpoint. In this study, particular emphasis was placed on bioassays employing the bacterial species E. coli, owing to its high sensitivity, ease of implementation, and rapid response in toxicity assessment. Among the available approaches for evaluating toxic effects, enzymatic activity assays are considered reliable indirect indicators of bacterial physiological status. The assessment of microbial viability is a useful approach for evaluating the toxicity of treated effluents (Fig. 2). One widely used method involves the resazurin bioassay, which monitors dehydrogenase activity associated with the electron transport system (ETS)35,47. Resazurin, a non-toxic tetrazolium-based dye, is reduced to resorufin by bacterial enzymes, producing a color change proportional to cellular metabolic activity. This method allows rapid and reliable estimation of effluent toxicity, providing insights into the biological impact of residual compounds generated during photocatalytic processes48,49. A schematic of effluent toxicity assessment using resazurin reduction on E. coli, this figure was created by authors using paint v. 2024. Recent research in photocatalysis has largely focused on TiO2- and Fe3O4-based materials due to their high efficiency in degrading various pollutants. However, very limited studies have simultaneously addressed the decomposition of CIP using chemical oxidation methods combined with toxicity assessment via E. coli ATCC 25,922 dehydrogenase enzyme activity. The novelty of this study lies in the integration of Central Composite Design (CCD) optimization with bioassay-based toxicity evaluation, which allows for the identification of optimal operational conditions for effective CIP removal while ensuring the environmental safety of the treated effluent. This study hypothesizes that TiO2/Fe3O4 enhances photocatalytic efficiency and reduces the residual toxicity of CIP. Therefore, the aim of this study was to identify the optimal conditions for CIP removal using the TiO2/Fe3O4/UV process, while concurrently assessing the toxicity of the treated effluent. This integrated strategy provides a comprehensive and sustainable approach for photocatalytic wastewater treatment. Materials and methods Ethical approval This study was approved by the Ethics Committee of Lorestan University of Medical Sciences (LUMS) under the code IR.LUMS.REC.1403.018. Figures 1, 2, 3, and 4 were created by the authors via paint v. 2024. Schematic illustration of nanoparticle synthesis: (A) Fe3O4 NPS, (B) TiO2/Fe3O4 NPS, this figure was created by authors using paint v. 2024. Schematic diagram of the photocatalytic reactor setup, this figure was created by authors using paint v. 2024. Preparation of material CIP) ≥ 99%), resazurin powder (≥ 98%), acetic acid (≥ 99.7%), ethanol (99.9%), iron(III) chloride (≥ 97%), tetraethyl orthosilicate (95%), tetra-n-butyl chlorotitanate (≥ 97%), sodium bicarbonate (≥ 99%), potassium dihydrogen phosphate (≥ 99%), dipotassium hydrogen phosphate (≥ 99%), dimethyl sulfoxide (DMSO, ≥ 99.9%), n-amyl alcohol (≥ 99%), glucose (≥ 99.5%), and sodium acetate (≥ 99%) were purchased from Merck, Germany. HCl–phthalate buffer and other reagents were obtained from readily available laboratory materials. The standard strain of E. coli ATCC 25922 was obtained from the Microbiology Laboratory of the Razi Research Institute, Tehran, Iran. The physicochemical properties of CIP and resazurin are summarized in Table 1. Table 1Physical and chemical properties of CIP and resazurin.Material nameMaterial propertiesMolecular structureCiprofloxacinChemical Formula: C17H18FN3O3Molecular weight: 331.347 g·mol/LSolubility in water: YesToxicity to organisms: Yes ResazurinChemical formula: C12H7NO4Molecular weight: 229.19 g·mo/LSolubility in water: YesToxicity to organisms: none Physical and chemical properties of CIP and resazurin. Chemical Formula: C17H18FN3O3 Molecular weight: 331.347 g·mol/L Solubility in water: Yes Toxicity to organisms: Yes Chemical formula: C12H7NO4 Molecular weight: 229.19 g·mo/L Solubility in water: Yes Toxicity to organisms: none Synthesis of Fe3O4 nanoparticles (NPS) In this study, Fe3O4 nanoparticles (NPs) were synthesized using the co-precipitation method. Briefly, 11.68 g of FeCl3·6 H2O and 4.31 g of FeCl2·6 H2O were dissolved in 200 mL of distilled water and stirred at 85 °C for 30 min. To prevent oxidation and facilitate degassing, nitrogen gas was bubbled through the solution. The pH of the solution was then adjusted to 9, followed by the dropwise addition of 20 mL of ammonia solution (30% w/v) until the solution color changed from orange to black. The supernatant was removed, and after several washes with ethanol and distilled water, the Fe3O4 NPs were collected37,50. Synthesis of nanocomposite TiO2/Fe3O4 For the synthesis of Fe3O4/TiO2 NPS, the Fe3O4 NPS obtained in the previous step were dispersed in a solution of tetra-n-butyl orthotitanate (P25 with 80% anatase and a surface area of 50 m2/g) (Fig. 3). The mixture was then subjected to ultrasonic treatment for 30 min. After sonication, the solution was allowed to stand undisturbed for 2 h, during which the pH gradually increased to approximately 8. The resulting NPs were first washed four times with ethanol and distilled water to remove impurities. The Fe3O4/TiO2 composite was then prepared with a Fe3O4:TiO2 ratio of 1:3 (w/w). Subsequently, the prepared catalysts were calcined at 773 K for 4 h at a heating rate of 5 °C min−1 under atmospheric pressure35,37,50–53. An 0.8 Tesla (T) magnetic field was applied to achieve complete separation of NPs during both synthesis and photocatalytic experiments. Characterization of TiO2/Fe3O4 NPS The characteristics of the synthesized TiO2/Fe3O4 NPS including surface morphology, crystal structure, elemental analysis, chemical composition, particle size distribution, were investigated by using Scanning Electron Microscopy (SEM) techniques, Energy Dispersive X-Ray Spectroscopy (EDX), Fourier Transform Infrared Spectroscopy (FTIR), and X-Ray Diffraction (XRD (techniques. Reactor setup The photocatalytic experiments were conducted under the conditions determined by CCD in a photoreactor with an approximate volume of 3 L and dimensions of 25 × 10 × 10 cm. For the photocatalytic experiments, the reactor was equipped with a UV-C lamp (length = 10 cm, power = 125 W, intensity = 20 mW/cm2, λmax = 254 nm), which was placed inside a quartz tube and positioned 10 cm above the solution surface. To ensure uniform dispersion of nanoparticles and prevent their sedimentation, a mechanical stirrer operating at 70 rpm was used during the experiments (Fig. 4). Stock solutions of CIP and TiO2/Fe3O4 NPS were first prepared. Then, 1 L of the CIP solution and the synthesized nanoparticles at the designed concentrations were added to the reactor. The mixture was then homogenized using a magnetic stirrer. During the photocatalytic process, samples were collected at specific intervals to measure the residual concentration of CIP. Quantitative analysis of CIP was performed using high-performance liquid chromatography (HPLC; Agilent Technologies 1260 Infinity) equipped with a C18 column (250 mm×4.6 mm). The injection volume was 20 µL. The mobile phase consisted of acetonitrile and water (70:30, v/v), the flow rate was 1.0 mL/min, and the column temperature was maintained at 30 °C. Detection was carried out at 278 nm. A standard calibration curve was constructed using known concentrations of CIP to determine the concentrations in the samples37, In all experiments, the pH was adjusted using standard solutions of 0.1 M HCl and 0.1 M NaOH. Modeling and statistical analysis In this study, Methodology Response Surface (RSM) was used to optimize experiments. RSM includes various models such as Behnken Design, CCD, Factorials Method, and Box-D-Optimal Design. Among them, the CCD was chosen due to saving time, high accuracy in determining the sensitivity of the response to each factor, determining the number of required samples, reducing the number of samples, and analyzing the effects of two or multiple variables simultaneously54–56. The quadratic model is mathematically represented by the following equation (Eq. 6) 57,58. where Y = Predicted response, β0 = Constant coefficient, βi = Linear regression coefficients, βii = Quadratic coefficients, βi = Interaction coefficients, Xi and Xj = Coded values of the independent variables. According to CCD model, firstly, independent variables including pH (3–12), contact time (15–120 min), initial concentration of CIP (1–50 mg/L) and the dose of NPS (10–500 mg/L) were defined as inputs to the model (Table 2). In the next step, output responses (CIP removal efficiency) were analyzed based on input variables. According to the CCD model and the number of four independent input variables, 30 runs were designed (Table 3), Expert Design version 13 was used for data analysis. Table 2Experimental range and levels of independent process variables.VariablesUnit− ∞− 10+ 1+∞Concentration of CIP(mg/L)113.2525.537.7550Contact time(min)1541.2567.593.75120pH–35.257.59.7512Dose of NPs(mg/L)10132.5255377.5500 Experimental range and levels of independent process variables. Table 3CCD matrix in coded terms along with experimental and predicted values for CIP removal.RunorderInitialconcentration(mg/L)Contacttime(min)pHDoseof NPs(mg/L)RE,%PredictedRE,%113.2593.755.25132.5090.0090.80213.2593.759.75377.5087.2082.20337.7593.759.75377.5071.0070.20437.7593.755.25132.5072.0076.30525.5067.507.50255.0091.1089.90613.2541.259.75132.5085.2084.70725.5067.507.5010.0083.0089.90813.2541.259.75377.5086.5084.70925.5015.007.50255.0081.5081.501050.0067.507.50255.0070.0068.201137.7541.255.25377.5075.0072.60121.0067.507.50255.0095.0095.501325.5067.5012.00255.0060.0059.701437.7541.259.75377.5073.0071.801513.2541.255.25132.5084.0087.901625.5067.57.50255.0090.5089.901737.7593.759.75132.5070.0070.201813.2541.255.25377.5089.9087.901913.2593.755.25377.5093.0090.802025.5067.507.50255.0090.9089.902113.2593.759.75132.5075.0082.202237.7593.755.25377.5077.0076.302325.5067.507.50255.0089.0089.902425.5067.507.50255.0090.0089.902525.5067.507.50255.0091.2089.902625.5067.503.00255.0070.0069.002725.5067.507.50500.0095.0089.902825.50120.007.50255.0084.0082.702937.7541.259.75132.5068.0071.803037.7541.255.25132.5071.0072.60 CCD matrix in coded terms along with experimental and predicted values for CIP removal. Effluent toxicity assessment In this study, the inactivation rate of E. coli was assessed by measuring bacterial growth inhibition before and after the photocatalytic process. Resazurin salt was used to evaluate toxicity (Table 1). The toxicity of CIP was determined using effective EC, NOEC, and 100% mortality parameters. Resazurin reduction by bacterial dehydrogenase was measured in nutrient broth medium enriched with potassium phosphate monobasic, dipotassium phosphate, glucose, and sodium acetate. Briefly, 2.75 mL of enriched nutrient broth was mixed with 2.25 mL of dimethyl sulfoxide, 1 mL of E. coli suspension, 1 mL of resazurin solution, and 1 mL of CIP. The mixture was incubated at 28 °C. After incubation, 10 mL of n-amyl alcohol and 1 mL of 0.05 M HCl-phthalate buffer were added. The upper alcohol layer was transferred to a microplate containing sodium bicarbonate. The absorbance of resazurin was measured at 610 nm. In the presence of active bacteria, dehydrogenase reduces resazurin to resorufin, changing the solution color from blue to pink35,47,50. Finally, the decrease in dehydrogenase enzyme activity during the conversion of resazurin was calculated using Eq. 7: where A= represents the dehydrogenase enzyme activity in the control sample, B= represents the dehydrogenase enzyme activity in the test (real) sample. Results and discussion Characterization of photocatalyst Morphological characterization by SEM Figure 5 presents the SEM images of the TiO2/Fe3O4 NPs, revealing nearly spherical Fe3O4 NPs with sizes below 100 nm uniformly distributed over the TiO2 surface. The TiO2 phase appears as dense, interconnected nanoscale clusters, which significantly enhance the interfacial contact and specific surface area of the composite. The homogeneous dispersion of Fe3O4 NPS and absence of agglomeration demonstrate the effective control of the sol-gel synthesis parameters, ensuring structural integrity and stability. Additionally, the presence of well-distributed fine pores on the surface suggests improved reactivity and potential enhancement in photocatalytic efficiency. These morphological features indicate favorable conditions for strong phase interactions between Fe3O4 and TiO2, contributing to the nanocomposite’s superior magnetic and photocatalytic properties. Overall, the morphology analysis confirms that the sol-gel method successfully produces uniformly dispersed, nanoscale particles with a stable structure, outperforming other synthesis techniques such as hydrothermal or chemical precipitation. SEM images of A, TiO2 and B, TiO2/ Fe3O4 NPs. Elemental composition analysis by EDX The EDX spectrum depicted in Fig. 6 validates the elemental composition of the TiO2/Fe3O4 NPs, demonstrating clear signals for Fe, Ti, and O. The distinct peaks of iron and titanium correspond well with the expected phases, confirming the successful incorporation of both components in the composite structure. The prominent oxygen peak indicates the formation of metal oxides, essential for the material’s photocatalytic and magnetic functionalities. Importantly, the absence of extraneous elemental peaks implies that the synthesis process yielded a high-purity product, free from contaminants. These findings align with SEM observations, which showed uniform particle distribution and morphology, collectively affirming the effectiveness of the employed sol-gel synthesis method. Generally, the elemental analysis confirms the successful synthesis of a homogeneously composed TiO2/Fe3O4 NPs with high purity. The verified elemental uniformity and oxide formation are fundamental to achieving enhanced photocatalytic performance and magnetic properties. This compositional integrity underlines the reliability of the sol-gel process for producing advanced multifunctional nanomaterials. EDX spectrum of the TiO2/Fe3O4 NPs, demonstrating the elemental composition with prominent peaks for Fe, Ti, and O. Functional group analysis by FTIR The FTIR spectrum of the TiO2/Fe3O4 NPs was recorded in the range of 451.15 to 3765.92 cm−1, revealing distinct and significant absorption bands (Fig. 7). A broad absorption band in the range of 3200–3600 cm−1 is typically attributed to O–H stretching vibrations, corresponding to adsorbed water molecules and surface hydroxyl groups in the TiO2/Fe3O4 NPs. According to recent reports, this band may include O–H vibrations in different hydrogen-bonding states as well as surface hydroxyl species59–61. The signal observed at around 3765.92 cm−1 in this study lies at the upper boundary of this range and is likely related to isolated surface –OH groups or instrumental resolution/overlapping effects. Therefore, the O–H related vibrations are interpreted within the range reported in previous studies62,63. These active surface groups play a crucial role as reactive sites in photocatalytic processes and surface interactions. Another absorption band at 3118.26 cm−1 further confirms the presence of hydroxyl groups and surface moisture, which significantly contribute to the improved dispersion and stability of the nanocomposite64. Additionally, the band observed at 1620.86 cm−1 is attributed to the bending vibrations of adsorbed water molecules (H-O-H), indicating the presence of moisture within the sample structure64,65. A weaker band near 1401.65 cm−1 is assigned to carbonate groups or C = O stretching vibrations, likely resulting from the adsorption of atmospheric CO2 or surface-related side products66. FTIR spectrum of the TiO2/Fe3O4 NPs. In the lower frequency region, characteristic bands at 895.33 cm−1 and 583.24 cm−1 correspond to the Ti-O-Ti and Fe-O bond vibrations in TiO2 and Fe3O4 structures, respectively, clearly confirming the coexistence of the two main phases in the nanocomposite64,66. Furthermore, the absorption band around 451.15 cm−1 is attributed to metal–oxygen bond vibrations, indicating a robust and stable oxide framework within the sample66. Overall, the FTIR results suggest that the TiO2/Fe3O4 NPs are physically combined without significant chemical interactions between the two phases. The presence of hydroxyl groups and surface moisture not only enhances sample stability but also potentially improves the photocatalytic activity and other surface-related properties of the nanocomposite. Crystalline structure analysis by XRD The XRD pattern (Fig. 8) indicates the presence of two crystalline phases: titanium dioxide (TiO2) in the anatase phase and magnetite (Fe3O4). The characteristic TiO2 peaks (JCPDS 04-0477) appear at 2θ = 25.3°, 37.8°, 48.0°, and 62.6°, corresponding to the (101), (0.0.4), (200), and (204) planes, confirming the tetragonal anatase structure. Magnetite peaks (JCPDS 01-1111) are observed at 2θ = 30.1°, 35.5°, 43.1°, 57.0°, and 62.6°, consistent with the cubic structure of Fe3O4. The full width at half maximum (FWHM) of the peaks (0.4–0.9°) indicates small crystallite size and good crystallinity. The crystallite size of the anatase phase was estimated using the Scherrer formula67,68, Eq. 8: XRD of the TiO2/Fe3O4 NPs. where D= crystallite size, K= shape factor (0.9), λ = X-ray wavelength of Cu Kα radiation (0.15406 nm), β = FWHM of the diffraction peak (in radians), and θ = Bragg diffraction angle. The calculation was performed using the most intense anatase (101) peak at 2θ = 25.36°. The FWHM of this peak was 0.907°, and the average crystallite size was calculated to be approximately 9.0 nm, confirming the nanocrystalline nature of the synthesized material. The high intensity of the (101) peak at 25.3° suggests the dominant presence of the anatase phase, which plays a key role in photocatalytic activity. Meanwhile, the presence of Fe3O4 imparts magnetic properties, facilitating easy separation of the nanocomposite from the reaction medium46,69,70. Overall, XRD results confirm the formation of a TiO2–Fe3O4 two-phase nanocomposite with tetragonal and cubic structures. This combination, due to its synergistic properties, can enhance structural stability, improve light absorption, and boost photocatalytic performance. Statistical analysis and model fitting for CIP removal In this study, CCD was employed to investigate and optimize the effects of key variables on the photocatalytic removal of CIP. The input variables and their respective levels are listed in Table 3. Each variable was examined at five levels: − ∞, − 1, 0, + 1, and +∞. The model fitting results and lack-of-fit analysis are presented in Table 4. According to these results, the experimental data demonstrated a good fit with the proposed polynomial model. The coefficient of determination (R2), adjusted R2, and predicted R2 for the model were 0.9895, 0.9797, and 0.9451, respectively. The adjusted R2 indicates the proportion of variance in the response that is explained by the model after accounting for the number of predictors. The predicted R2 reflects the model’s predictive capability for new data. A smaller gap between the adjusted R2 and predicted R2 values suggests a higher reliability of the model. In this study, the relatively small difference between these two values (0.9797 and 0.9451) confirms the robustness and acceptability of the model. The subsequent step involved conducting an ANOVA to evaluate the statistical significance of the model terms. Table 4Lack of fit and model summary statistics.SourceSum ofsquaresdfMean squareF valuep-valueProb > FA- Lack of Fit TestsLinear1265.71372063.285790.5374< 0.00012FI1243.59621488.8283127.0791< 0.0001Quadratic24.7717102.47723.54390.0876SuggestedCubic9.8824.947.06720.0349AliasedPure Error3.49550.699SourceStd. Dev.R-SquaredAdjustedR-SquaredPredictedR-SquaredPRESSB- Model summary statisticsLinear7.12520.52790.45230.34791753.17402FI8.10160.53610.29190.20252143.9837Quadratic1.37270.98950.97970.9450147.7176SuggestedCubic1.38230.9950.97940.46891427.7528Aliased Lack of fit and model summary statistics. The results of the ANOVA are presented in Table 5. Using analysis of ANOVA and its results, the models presented for the response variable can be evaluated. Also, based on these results, the regression coefficients for linear, interaction, and quadratic terms can be estimated. The coefficient of determination or goodness of fit, R2, is one of the tests used in ANOVA analysis to identify the superior model. Researchers use this test to measure accuracy35,58. In evaluating model performance, the Adjusted R2 is considered a key indicator, as it accounts for the number of predictors in the model. Unlike the regular R2, the adjusted form does not necessarily increase with the addition of new variables. In fact, incorporating irrelevant or non-contributing factors may lead to a decline in this coefficient. Generally, a higher adjusted R2 suggests a better-fitting model; however, this metric alone is not sufficient for confirming the model’s adequacy. From a statistical perspective, model validity is also assessed using P-values and F-statistics. The F-value, which mathematically corresponds to the square of the t-value, is a measure of the overall significance of the regression model. A larger F-statistic indicates a more reliable model. In this research, according to the data in Table 5, the obtained Model F-value of 157.41 signifies strong statistical significance, with only a 0.01% probability of such a value arising due to random variation (noise). Furthermore, the Lack of Fit F-value was found to be 3.24, indicating that the lack of fit is not significant compared to the pure error, with a 10.01% probability of being caused by random error. A non-significant lack of fit is generally considered favorable in regression analysis. Another crucial parameter is the Adequate Precision, which reflects the signal-to-noise ratio in the system. A value exceeding four is generally acceptable. In this case, a ratio of 45.562 was recorded, suggesting a highly adequate signal, thereby confirming the model’s capability to explore the design space effectively. Additionally, the Sum of Squares (SS) values, reported in Table 5, offer insight into the contribution of each variable to the model. A higher SS value represents a stronger influence of the corresponding variable. The SS for the initial concentration was 1128, whereas the contact time had an SS of only 0.96. This disparity implies that the initial concentration exerts a significantly greater effect on the process efficiency, particularly in the context of contaminant removal. Consequently, it can be inferred that this variable plays a dominant role in determining process performance. Finally, considering these statistical indicators, one can also analyze interaction effects among variables to identify which combinations exert the most substantial influence on the system under study. Table 5ANOVA, regression coefficient estimate and test of significance for CIP removal.SourceSum ofsquaresCoefficient estimatedfMean squareF valuep-valueProb > FModel2660.12514190.009100.8302< 0.0001SignificantA-Initial Concentration1128.882− 6.8611128.882599.0528< 0.0001B-Time0.960.210.960.50940.4863C-pH136.3267− 2.381136.326772.3432< 0.0001D-Dose of NPs153.0152.581153.01581.199< 0.0001AB2.1025− 0.3612.10251.11570.3076AC0.04− 0.0510.040.02120.8861AD2.7225− 0.4112.72251.44470.2480BC12.960.9112.966.8770.0192BD3.80250.4913.80252.01780.1759CD0.490.1710.490.26000.6175A^2118.0971–1118.097162.6695< 0.0001B^2111.09–1111.0958.9512< 0.0001C^21141.097–11141.097605.535< 0.0001D^25.5543–15.55432.94740.1066Residual28.26637–151.8843––Lack of fit24.77137–102.47723.5438720.0876Not significantPure error3.495–50.699 ANOVA, regression coefficient estimate and test of significance for CIP removal. Based on the data in Table 5, the highest and lowest interaction effects were related to BC and AC, respectively. In other words, the interaction effect of BC variables, contact time and pH, had the greatest impact on process efficiency (P-value < 0.05). In Table 5, based on the information in column of Coefficient Estimate, it can be seen whether the effect of each variable was direct or inverse, in other words, with an increase in each variable, the efficiency of the process for removing the pollutant under study increased or decreased. For example, the effect of the variables of contact time and photocatalyst dose was direct, and increasing the value of these two variables can increase the removal efficiency. The removal efficiency of CIP was calculated based on the quadratic model using Eq. 9: Diagnostic plots for CIP removal Model evaluation The next step in optimizing the data for this study was to examine the data. To ensure the model’s validity, the data were analyzed in four steps. First, the normal probability plot of the Studentized residuals was used to assess the normality of the residuals. Next, the Studentized residuals versus predicted values were evaluated to check for constant error. The externally Studentized residuals were also analyzed to identify outliers and influential points. Finally, the Box-Cox plot was employed to examine power transformations of the response variable. The diagnostic plots are shown in Fig. 9. Based on the shape of the data distribution in Fig. 9A, the data distribution is normal. This plot provides a guideline for selecting the correct power law transformation. A recommended transformation is listed, based on the best lambda value, which is found at the minimum point of the curve generated by the natural log of the sum of squares of the residuals. If the 95% confidence interval around this lambda includes one, then, the software does not recommend a specific transformation. Figure 9B examines the assumption of equality of variances, which states that the variances are equal based on the distribution of the data. This is a plot of the residuals versus the ascending predicted response values. It tests the assumption of constant variance. Diagnostic plots for CIP removal: A Normal probability plot of the Studentized residuals, B Studentized residuals versus predicted values, C Externally comparison between the actual experimental values and those predicted, and D Box-Cox plot for power transformations. The plot should be a random scatter (constant range of residuals across the graph). Figure 9C presents the comparison between the actual experimental values and those predicted by the fitted model. As illustrated, most of the data points lie close to the 45-degree line, indicating a strong agreement between the observed and predicted responses. This close alignment suggests that the model provides accurate predictions across the experimental range and confirms its adequacy for further statistical analysis and process optimization. Figure 9D provides a guideline for selecting the correct power law transformation. A recommended transformation is listed, based on the best lambda value, which is found at the minimum point of the curve generated by the natural log of the sum of squares of the residuals. If the 95% confidence interval around this lambda includes one then the software does not recommend a specific transformation. This plot is not displayed when either the logit or the arcsine square root transformation has been applied. Model validation The software predicted the optimum conditions for CIP removal as follows: pH = 7.5, initial CIP concentration = 1 mg/L, nanoparticle dosage = 255 mg/L, and contact time = 67.5 min. The predicted removal efficiency under these conditions was 93.7% (Table 6). The optimized run was selected for experimental validation, and the measured responses were subsequently compared with the model-predicted values. The mean experimental removal efficiency for CIP (n = 3) was 91.43%, which shows good agreement with the model-predicted value, confirming that the response model is valid and reliable for process optimization. Table 6Experimental and predicted response values for CIP removal under optimal conditions.Optimum conditionInitial CIP concentration (mg/L)NPs dosage (mg/L)pHContact time (min)Experimental removal (%)Predicted removal (%)Run optimum12557.567.591.4393.70 Experimental and predicted response values for CIP removal under optimal conditions. Effects of operational parameters on the photocatalytic degradation of CIP Effects of pH The results of process optimization are presented in Fig. 10. The optimal conditions for CIP removal were determined as follows: initial concentration 1 mg/L, contact time 67.50 min, pH 7.50, and nanoparticle dose 255 mg/L. The data in Fig. 10a, b indicate that CIP removal efficiency is significantly influenced by pH, with efficiency increasing from 82% to 95% as pH rises from 5.25 to 7. The highest removal rates were observed at pH values between 6.5 and 7.5, consistent with Fig. 10a. Model optimization further confirmed that the optimal pH for CIP removal is 7.50. This finding aligns with previous studies showing that solution pH is a key factor affecting chemical processes, particularly AOPs71. The rate of chemical reactions is dependent on the solution pH and directly or indirectly affects the course of the process72. In the AOPs, the solution pH changes affect the oxidation rate through the production of various radicals. Changes in solution pH lead to positive or negative charging of nanoparticles and consequently affect the disintegrating performance of nanoparticles73. The nature of different antibiotics varies depending on the pH of the solution. The behavior of the antibiotic CIP in the environment is influenced by the solution’s pH: below pH 5.5, it exists in a protonated form, with cations being predominant; between pH 5.5 and 7.7, the hydrogen ion dissociates from the carboxyl group, making the zwitterion (dipolar ion) form dominant; and at pH values above 7.7, CIP loses protons, resulting in the predominance of its anionic form74. 3D surface plot of the pH, contact time, dosage of NPs and initial concentration of CIP in the removal efficiency process. The efficiency of CIP removal in acidic and alkaline environments is largely determined by two critical parameters: the analyte’s pKa and the point of zero charge (pHPZC) of the photocatalyst. In the present study, CIP exhibited a pKa of 5.7, whereas TiO2 displayed a pHPZC of 6.25. The pHPZC denotes the pH at which the surface charge of the nanoparticle becomes zero under specific temperature, pressure, and aqueous conditions35,75. When the solution pH is below these values, both CIP and TiO2 NPS exhibit a positive surface charge, whereas at pH values above these thresholds, the surface charge becomes negative. At or near the pKa and pHPZC values, the molecules remain largely unchanged. These parameters play a critical role in determining the surface charge of TiO2/FeO NPs and the ionization state of CIP, which together govern the adsorption behavior and photocatalytic degradation efficiency76,77. The results clearly demonstrate that TiO2/Fe3O4 NPs under UV-C irradiation efficiently degrade CIP, and that the solution pH has a pronounced impact on the degradation rate. Understanding the interplay between the analyte’s pKa and the photocatalyst’s pHPZC is essential for optimizing photocatalytic performance and provides mechanistic insights into the pH-dependent removal process of target contaminants. Therefore, the observed maximum degradation at near-neutral pH can be explained by the synergistic effect of optimal CIP speciation, favorable surface charge interactions, and enhanced radical-mediated oxidation under UV-C irradiation78. These findings highlight the critical role of pH in regulating both interfacial interactions and photocatalytic efficiency, providing mechanistic insight into the degradation behavior of CIP using TiO2/Fe3O4 nanoparticles. Detailed information on CIP removal efficiency under the influence of various operational variables is presented in Table 3; Fig. 10. Effects of contact time Based on the obtained results, contact time was one of the effective factors in improving the removal efficiency of CIP (Fig. 10a, e, d). At a contact time of 41.25 min, the removal efficiency of CIP was reported to be 87%, indicating the satisfactory performance of the photocatalytic process during this period. With an increase in contact time to 67.5 min, the removal efficiency reached its maximum value of 95%. This significant increase demonstrated the important role of contact time in enhancing the efficiency of CIP removal. Therefore, a contact time of approximately 67.5 min was recognized as the time to achieve the maximum removal efficiency of CIP in this study. Moreover, based on the modeling results in this study, a contact time of 67.50 min was identified as the optimal contact time. In this study, it was found that increasing the contact time significantly improved the removal efficiency of CIP in the TiO2/Fe3O4/UV. This improvement was attributed to the increased opportunity for reactions between CIP molecules and •OH generated by the photocatalytic system. Moreover, the extended contact time led to greater UV light absorption by the photocatalyst and more effective excitation of electrons and holes, which ultimately increased the rate of oxidation reactions. As a result, more complete degradation of the contaminant was achieved. However, after reaching the maximum contact time for CIP removal, further increases in contact time did not significantly enhance the removal efficiency, which was attributed to limiting factors such as catalyst surface saturation and reduced light penetration35,50. Similarly, Dimitrakopoulou et al. 38 reported that the photocatalytic degradation of amoxicillin using the TiO2/UV system achieved a removal efficiency of 93% after 90 min of contact time. Their findings, also, highlighted the critical role of contact time as a key operational parameter influencing the efficiency of photocatalytic antibiotic degradation. The difference in the observed significance of contact time between the present study and a previous study can be attributed to several factors. In the earlier work, the photocatalytic degradation of amoxicillin using a TiO2/UV system was strongly dependent on contact time due to slower reaction kinetics and the involvement of both initial adsorption and photocatalytic degradation processes, requiring a longer period to reach steady-state conditions. In contrast, in the present study, CIP degradation occurred more rapidly, likely due to the higher availability of active sites on the catalyst surface and more efficient generation of reactive species under irradiation, leading to faster photocatalytic reaction kinetics and an earlier attainment of quasi-equilibrium. The dark experiment results also confirmed that adsorption played a negligible role in the overall removal process. Moreover, differences in pollutant structure, catalyst properties, initial concentrations, and light intensity between the two studies can significantly influence the reaction pathway and kinetic behavior. As a result, while contact time was a key operational parameter in the previous study, its effect in the present work was limited within the investigated range and did not reach statistical significance. Effect of the initial concentration of CIP The findings of this study showed that the initial concentration of CIP has a significant impact on its removal process, based on the results presented in Fig. 10b, d, and f, increasing the initial concentration of CIP from 13.25 to 32.5 mg/L resulted in a decrease in removal efficiency from 95% to approximately 60%. Increasing the initial CIP concentration is one of the important parameters in the rate of CIP degradation by the photocatalytic process, which has a significant impact on the removal efficiency. According to the Beer-Lambert law, as the pharmaceutical compound concentration increases, the rate of light and UV radiation photons passing into the solution decreases, and by being trapped and absorbed on the drug surface, it reduces the absorption of photons on the catalyst, reduces the production of active •OH radical, and ultimately reduces the rate of decomposition79,80. Hasani et al. investigated the effect of different concentrations of CIP on its removal efficiency using nanocatalysts. Their results demonstrated that increasing the initial concentration of CIP from 5 to 25 mg/L significantly decreased the removal efficiency, this reduction was attributed to the saturation of active sites on the nanocatalyst surface by adsorbed CIP molecules, which limited the access of the nanocatalyst to reactive radicals. Therefore, at higher concentrations of CIP, a greater production of radicals was required to carry out the effective oxidation process; if this demand was not met, the removal efficiency decreased81. in another study, Jia et al. demonstrated that increasing the initial concentration of CIP from 10 to 45 mg/L significantly decreased the removal efficiency in photocatalytic processes. They attributed the reduction in CIP degradation efficiency to the limited availability of hydroxyl radicals produced relative to the increased number of CIP molecules82. Effect of NPS dose Based on the results obtained from Fig. 10c, e, f, an increase in the dosage of TiO2/Fe3O4 NPS led to a significant improvement in the CIP removal efficiency. At lower nanoparticle doses (below approximately 100 mg/L), the removal efficiency significantly decreased. However, as the NPS dosage was increased to approximately 300 mg/L, the removal efficiency exceeded 94%. This enhancement demonstrates the positive effect of NPS dosage on the CIP removal process. Furthermore, the study demonstrated that once a certain NPS dosage threshold was reached, further increases in the dosage had a negligible effect on the enhancement of removal efficiency, and the process approached a near-saturated plateau. The findings indicate that the use of TiO2/Fe3O4 NPs in the AOPs under UV-C irradiation significantly improves the removal efficiency of CIP. The TiO2 component, due to its photocatalytic properties, and the Fe3O4 component, due to its magnetic properties, can both play a crucial role in this process. TiO2 acts as a photocatalyst under UV-C light, generating electrons and positive holes that lead to the production of free radicals such as •OH, which can degrade organic pollutants. Meanwhile, Fe3O4, due to its magnetic properties, facilitates the easy recovery of NPs after the process, thereby improving the overall removal efficiency of the pollutants. Pishrafti et al. investigated the application of magnetic TiO2/Fe3O4 NPs under UV-C irradiation for the removal of CIP. Their results demonstrated that at a catalyst dosage of 0.6 g/L and a contact time of 60 min, the removal efficiency of CIP reached approximately 99.1% 83. In another study, Gad Allah et al. examined the photocatalytic oxidation of CIP under simulated solar irradiation. Their findings indicated that optimizing the nanocatalyst loading in the reaction medium enhanced the catalyst’s surface area, promoted the generation of reactive free radicals, and increased the availability of active sites for CIP adsorption, thereby improving the overall degradation efficiency84. The results of some studies show that increasing the catalyst beyond a certain amount reduces pollutant decomposition77,85. Generally, increasing the catalyst dosage beyond the optimal level causes higher turbidity and reduces light transmission, which leads to a decrease in photocatalytic process efficiency. Moreover, excessive catalyst consumption is not economically viable77. Kinetic model analysis in the photocatalytic degradation of CIP In this study, kinetic parameters obtained from fitting the experimental data to both the Langmuir–Hinshelwood (Eq. 10) and pseudo-first-order (Eq. 11) models were used to describe CIP degradation in photocatalytic processes. where C0 and Ct (mg/L) represent the initial and final concentrations of CIP at the beginning and end of the contact period, respectively, Kobs (1/min) is the observed reaction rate constant obtained from the pseudo-first-order kinetic model, KC (L/mg) corresponds to the adsorption equilibrium constant describing the affinity of CIP molecules for the catalyst surface, KL−H (mg/L min) is the surface reaction rate constant derived from the Langmuir–Hinshelwood model, accounting for the surface-mediated contribution to the overall photocatalytic degradation process86–88. According to the results presented in Table 7, the pseudo-first-order model revealed a significant decrease in the Kobs, with increasing initial CIP concentration, from 0.0440 1/min at 1 mg/L to 0.0046 1/min at 50 mg/L. This behavior reflects a reduced degradation rate at higher pollutant loadings, likely due to active site limitation and/or mass transfer constraints. High R2 values (0.90–0.985) confirm the excellent agreement between the model and experimental data, indicating that the pseudo-first-order model reliably describes the overall kinetic behavior. In contrast, plotting Kobs versus the initial concentrations of CIP, the Langmuir–Hinshelwood model yielded KC = 0.224 L/mg, KL–H = 0.0825 mg/L.min, and R2=0.913, indicating contributions from both surface adsorption and surface reaction in CIP degradation. Although this model provides mechanistic insight into the reaction pathway, the present study primarily focused on model adequacy rather than mechanistic interpretation. Accordingly, the L–H model is more suitable for interpreting reaction mechanisms, whereas the pseudo-first-order model better represents the overall kinetics of the process. These findings are consistent with previous reports, where Mohamed et al.89 and Namrata et al.90 demonstrated that CIP degradation in aqueous photocatalytic systems follows pseudo-first-order kinetics. Overall, comparison of the two models confirms that the pseudo-first-order model provides a superior fit to the experimental data and is the most appropriate for describing CIP degradation kinetics in these processes. Table 7Pseudo-first-order kinetic parameters for degradation of CIP.Co (mg/L) R 2 Linear equation122.720.985y = 0.0440x + 0.0011556.170.914y = 0.0178x + 0.1002584.030.910y = 0.0119x + 0.15035208.320.905y = 0.0048x + 0.20050217.390.901y = 0.0042x + 0.050 Pseudo-first-order kinetic parameters for degradation of CIP. Comparative review of CIP removal methods from aqueous solutions Based on previous studies summarized in Table 8, the results showed that the removal efficiency achieved in the present study fell within the upper performance range reported for recent advanced treatment processes (94–99%). For instance, the Fe3O4/SiO2/TiO2 nanocomposite with a core-double-shell configuration exhibited a removal efficiency of 94% 91, whereas the combined adsorption-photocatalytic process using nat-HA/TiO2 achieved approximately 98.5% efficiency92. The comparable performance among these treatment processes and the current study suggested that the presence of TiO2 and Fe3O4 phases played a decisive role in enhancing the generation of reactive radicals and facilitating electron-hole separation and transfer. Table 8Comparison of different CIP removal methods from aqueous solutions.MethodsFeaturesRemoval efficiency (%)Referencesphotochemical decomposition(TiO2/Fe3O4/UV-C)Optimal conditions ( pH = 7.50), NPs (255 mg/L), contact time = (67.50 min), initial CIP concentration (1 mg/L), and UV light irradiation (20 mW/cm2)95Present studyFe3O4/SiO2/TiO2Core-2-Layer-Shell NanocompositeContact time = 120 min,CIP concentration = 10 mg/dm394 91 chemically modified bamboo biochar (MBC)MBC dose (0.5 g /L), CIP initial concentration (20 mg/L), pH (7.5), and contact time (46 min)96 57 magnetic Fe3O4 /AC (MAC) nanocomposit with three-dimensional electrocoagulation (EC) system, as well as using persulfate (PS). (EC-MAC-PS process)Optimum conditions: CIP (10 mg/L), MAC (0.9 g/L), pH (7), current density (0.3 mA/cm2) and contact time (35 min)98.21 93 Combination of Adsorption–Photocatalysis Process Using Natural Hydroxyapatite (nat-HA) and TiO2pH (8.1), CIP initial concentration (1 mg/L), contact time (90 min), nat-HA dose (1 g/L)≈ 98.50 92 protein-modified nanosilica (ProMNS)pH (7), contact time (90 min), adsorbent dosage (10 mg/mL)89.86 95 Zn–C battery derived graphene oxide enhanced by hydrogenbonding, electrostatic and π-π interactionpH (5.7), adsorbent dosage (20 mg/L), CIP initial concentration (2 mg/L)99 94 Electro-activated persulfate (EC-PS)oxidation using aluminum electrodesOptimum conditions: pH (7),contact time (40 min), current density (2.75 mA/cm2), CIP initial concentration (2 mg/L)90 27 Magnetic chitosan/Ag/MWCN nanocompositeDosage of nanocomposite (0.6 g/L), contact time (60 min),~ 92.50 96 Sono-nano-chemical/PS (US/ZnO/PS) proceContact time (10 min), pH (7), CIP initial concentration (25 mg/L), and PS concentration of (476.06 mg/L)99 28 Comparison of different CIP removal methods from aqueous solutions. photochemical decomposition (TiO2/Fe3O4/UV-C) Fe3O4/SiO2/TiO2 Core-2-Layer-Shell Nanocomposite Contact time = 120 min, CIP concentration = 10 mg/dm3 Zn–C battery derived graphene oxide enhanced by hydrogen bonding, electrostatic and π-π interaction Electro-activated persulfate (EC-PS) oxidation using aluminum electrodes Optimum conditions: pH (7), contact time (40 min), current density (2.75 mA/cm2), CIP initial concentration (2 mg/L) In comparison with advanced electrochemical and sonochemical oxidation processes such as EC-MAC-PS (98.21%) 93, and US/ZnO/PS (99%) 28, it was observed that although these technologies reported slightly higher removal efficiencies, they required the application of electric current, ultrasonic irradiation, or additional oxidants such as persulfate. These operational demands consequently increased both the cost and the complexity of the treatment processes. In comparison with advanced adsorbents such as modified biochar (MBC) with a 96% removal efficiency57, and graphene oxide derived from Zn-C batteries with a 99% efficiency94, the performance of the present process was comparable. However, a fundamental difference in the removal mechanism was observed. While adsorption-based methods primarily involved the physical or chemical transfer of CIP molecules onto the adsorbent surface, the current process resulted in the complete photodegradation of the molecular structure of CIP. This characteristic provided an environmental advantage, as it minimized the risk of pollutant re-release into the environment. Considerations on CIP degradation intermediates In the structure of CIP, the piperazine ring consists of two nitrogen and four carbon atoms. Due to its higher electron density and reactivity, it is typically the first site attacked by •OH or O2•− radicals in photocatalytic processes. This attack leads to cleavage of the piperazine (C4H10N2) ring and the formation of smaller intermediate compounds97. In the subsequent stages, radicals attack the quinolone core, leading to its degradation, which constitutes the main pathway of core breakdown and molecular fragmentation of CIP. Previous studies have shown that the photocatalytic degradation of CIP is generally accompanied by the formation of defluorinated, decarboxylated, and hydroxylated derivatives; specifically, defluorination reduces antibacterial activity, while decarboxylation decreases molecular weight. These processes indicate that piperazine ring cleavage as the initial attack site and quinolone core degradation at advanced stages represent the primary pathways for the formation of intermediate products97–99. In the present study, due to limitations of the analytical instruments and the inability to directly identify intermediate compounds, the results primarily rely on indirect evidence. Nevertheless, these findings facilitate the prediction of potential transformation products and provide insight into the environmental fate of CIP, as well as the efficiency of the photocatalytic process and the possible toxicity of its degradation products. Assessment of effluent toxicity In this study, the toxicity of the effluent was evaluated using the resazurin reduction test with E. coli bacteria. The resazurin test, known as a rapid, sensitive, and cost-effective method, is widely used to assess environmental toxicities50,100,101. In the presence of this compound, an increase in CIP concentration led to a more pronounced decrease in the metabolic activity of the bacteria. The dye change of the resazurin solution from blue to pink or colorless was used as a qualitative indicator of the metabolic activity of viable E. coli cells under both pre- and post-treatment conditions (Tables 9 and 10). As shown in Table 9, a CIP concentration of 1.30 mg/L (EC50) was sufficient to inhibit 50% of the metabolic activity of E. coli cells. At higher concentrations, up to 2.37 mg/L, complete inhibition (100% mortality) occurred, confirming the strong toxic effect of CIP prior to photocatalytic treatment. Table 9The results of evaluating the toxicity of CIP on E. coli bacteria before the AOPs.ParametersAmount (mg/L)± SDEC101.050.0707EC201.160.2263EC301.210.2970EC401.260.3677EC501.300.4243EC601.380.5374EC701.670.9475EC802.10.1414EC902. 200.3111EC1002.220.5233 The results of evaluating the toxicity of CIP on E. coli bacteria before the AOPs. Table 10The result of bioassay test to determination of effluent assessment in different Runs.Run orderResiduals of CIPin the effluent (mg/L)± SDThe amount of NOEC and EC based on RRT11.370.5233EC5021.690.9758EC59310.941.3294EC100410.951.3435EC10052.270.3818EC10062.120.1697EC8074.330.4667EC10081.781.1031EC6394.711.0041EC1001015.007.0711EC100119.430.6081EC100120.050.0707NOEC1310.200.2828EC1001410.190.2687EC100152.120.1697EC96162.420.5940EC1001712.620.8768EC100181.330.4667EC40190.920.5940NOEC202.320.4525EC100212.781.1031EC100228.680.9617EC100232.801.1314EC100242.550.7778EC100252.240.3394EC95267.650.9192EC100271.270.3818EC36284.080.1131EC1002912.080.1131EC100309.811.1455EC100 The result of bioassay test to determination of effluent assessment in different Runs. Following the photocatalytic process, a significant improvement in microbial viability was observed. Table 10 reveals that the EC values increased markedly, accompanied by a considerable decrease in the residual concentration of CIP in the treated effluent. Statistical analysis using an independent t-test confirmed that the reduction in toxicity after photocatalytic treatment was statistically significant (p < 0.05). These results clearly demonstrate that the photocatalytic process effectively decreases the biological toxicity of CIP by transforming it into less harmful intermediates. Thus, the process provides a dual advantage achieving both chemical degradation and ecological risk reduction in the treated effluent. Such findings underscore the potential of photocatalytic oxidation as a reliable and environmentally sustainable technology for the safe management of pharmaceutical-contaminated effluents. The toxicity assessment of the reactor effluent (Table 10) showed that the EC50 and 100% mortality values were 1.37 and 2.27 mg/L, respectively, which exhibited minor differences compared to the corresponding values reported in Table 9. One of the concerns regarding the degradation of antibiotics in aquatic environments is the formation of intermediate compounds, which can be significantly more toxic than the parent compound. In this study, one of the objectives of assessing effluent toxicity was to determine the toxicity resulting from the degradation of CIP and its intermediate compounds. Based on the results presented in Tables 9 and 10 (comparison of EC₅₀ values), the toxicity after the photocatalytic process decreased by more than 5% compared to before the process. This indicates that the toxicity of the effluent from CIP degradation not only did not increase but also decreased by approximately 5%. Therefore, the TiO2/Fe3O4/UV-C catalytic process can be considered a suitable method for the degradation of CIP in water sources. A study by Toolabi et al. reported the evaluation of toxicity in the effluent from a photocatalytic reactor using the resazurin dye reduction assay in aqueous solutions. According to the results obtained in their study, the EC50, NOEC and 100% mortality concentrations were reported to be 2.34, 0.18 and 4.43 mg/L, respectively102. In another study, Zare et al. employed the Alamar Blue (AB) assay to evaluate the toxicity of effluent from the Sequencing Batch Reactor (SBR) process. Their findings demonstrated a direct correlation between the reduction of AB and bacterial metabolic activity. Consequently, they highlighted the AB reduction test as a novel and reliable method for toxicity assessment of wastewater effluents100. Limitations of the study The present study demonstrated that iron- and titanium-based photocatalysts under UV irradiation can effectively remove CIP. Nevertheless, several important limitations should be acknowledged. The process efficiency was highly sensitive to operational parameters such as pH, contact time, catalyst loading, and light intensity, highlighting the challenge of achieving optimal performance under real effluent conditions with complex matrices. In addition, the requirement for continuous UV irradiation results in considerable energy consumption and heat generation in the reaction medium, potentially affecting temperature-dependent reaction kinetics. The photocatalytic degradation pathway may also generate intermediate compounds whose persistence and toxicity remain uncertain. The present study did not include the identification or characterization of intermediate compounds using advanced analytical techniques such as LC–MS or GC–MS. Consequently, the degradation pathway and the exact nature of the by-products could not be conclusively determined, and pollutant removal does not necessarily correspond to complete detoxification of the treated solution. This represents an important limitation of the current work and should be considered when interpreting the environmental implications of the process. These findings emphasize the need for further studies to identify transformation products and assess their ecological impacts. Furthermore, the study did not include Brunauer-Emmett-Teller (BET) surface area or particle size distribution analyses, which are essential for correlating surface properties with photocatalytic performance. Moreover, the mineralization efficiency (TOC removal) was not measured due to instrumental analysis limitations, which should be considered as another limitation of the present study. These parameters should be examined in future investigations to better understand the role of surface area, particle dispersion, and overall mineralization behavior in determining catalyst activity. Finally, scaling up laboratory results to industrial applications remains challenging due to the high cost of NPs synthesis, significant energy demand, and the complex design of photocatalytic reactors. Therefore, future research should focus on enhancing catalyst stability, reducing operational costs, incorporating comprehensive surface characterization, and ensuring environmental safety before practical implementation. Conclusion The TiO2/Fe3O4 nanocomposite demonstrated efficient performance for CIP degradation under optimized conditions. The CCD-based optimization confirmed that the process behavior can be adequately described by a quadratic model (R2 = 0.9895), achieving a maximum removal efficiency of 95%. Toxicity results (NOEC = 0.92 mg/L, EC₅₀ = 1.37 mg/L, 100% mortality = 2.27 mg/L) indicated that residual toxicity remained in the treated solution despite high removal efficiency. Overall, the results confirm the applicability of the proposed photocatalytic process for CIP degradation. However, the limited reduction in toxicity suggests incomplete mineralization and potential formation of intermediate by-products. This highlights the need for further investigation into transformation products and their toxicological impacts to better evaluate the environmental relevance of the process. Acknowledgements The authors would like to thank Lorestan University of Medical Sciences, Iran, for their financial support of this study. We have used artificial intelligence software to edit and improve the article’s writing. Author contributions M.GH writing, editing, drawing. M.Y writing, original draft. F.A editing, investigation. B.K inves¬tigation, methodology. A.T writing, review, editing, drawing, supervision. Funding This study was financially supported (No. 1397-1-99-3550) by Lorestan University of Medical Sciences (Iran). Data availability All data generated or analyzed during this study are included in this published article. Declarations Competing interests The authors declare no competing interests. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. References Associated Data Data Availability Statement All data generated or analyzed during this study are included in this published article.