Norfloxacin is a second generation synthetic fluoroquinolone antibiotic that has been widely detected in aqueous environments. In Chesapeake Bay, the largest estuary in the United States, the highest aqueous phase concentration of norfloxacin was recorded at 94.1 ng/L [1]. In China, it has been reported that norfloxacin is the most frequently detected antibiotic in coastal waters with a maximum concentration of 1990 ng/L [2]. Furthermore, in an artificial drinking water reservoir in the Yangtze River delta of East China, norfloxacin was detected at a frequency of 97.85%. With the deliberate release of antibiotics into the aqueous environment and lack of effective treatments for antibiotic-contaminated wastewater, antibiotic pollution continues to pose a serious environmental threat to aquatic and terrestrial ecosystems [3]. Therefore, the removal of norfloxacin from aquatic environments is becoming an important issue [4, 5].

Different technologies are available for the remediation of norfloxacin, including biological degradation, adsorption, and AOPs, to minimize its damage in aquatic environment [6-8]. Among the emerging treatment approaches, AOPs are considered to be effective and are currently gaining significant attention in the water treatment industry [9-13]. The pH is considered to be an important factor for photocatalytic performance, as demonstrated by many previous studies [14, 15]. Generally, the optimum pH for semiconductor photocatalysts are confined within a narrow range [14, 16]. For example, the optimum pH of ZnO photocatalysts was shown to be 11 for the degradation of amoxicillin, ampicillin, and cloxacillin in aqueous solution [17]. The optimum working pH of TiO₂ was reported to be 9.0 for chlorinated aniline degradation [18]. In addition to TiO₂, Bi-based semiconductors have been proven effective for organic pollutant removal [19-22]. Fu et al. [23] showed that B-doped Bi₂WO₆ exhibited the highest degradation efficiency toward RhB removal at a pH of approximately 7.0. Outside of this narrow pH range, the catalytic performance significantly decreased. To increase the decay rate of various pollutants, the initial pH of wastewater was adjusted to an optimum level by addition of acid or base [24, 25]. However, Hu et al. [26] reported that the pH decreased during photocatalytic reaction, falling outside of the optimal range, slowing pollutant decay.

Previously, we demonstrated that the pH cannot be maintained throughout the SSL/Bi₂WO₆ process due to the formation of small molecule organic acids and carbon dioxide during ciprofloxacin degradation. The pH continuously decreased during the reaction and eventually the solution pH dropped to 3.0, departing drastically from the optimal pH and was accompanied by the retardation of CIP decay [27]. Rao et al. [28] reported a WO₃ photocatalytic system for monuron degradation, achieving the best performance at pH = 6.0. Furthermore, during the developed process, the monuron molecule was transformed to a simple aliphatic acid, resulting in decreased solution pH and retardation of total organic carbon (TOC) removal. In the UV/TiO₂ degradation simazine process, the best degradation efficiency was obtained at an initial pH of approximately 9, and the photocatalytic reaction was retarded under strongly acidic and extreme basic conditions [29]. Meanwhile, the stable cyanuric acid was formed upon the oxidative decomposition of simazine, resulting in acidification of the aqueous solution. Bi₂O₃/Bi₂O₂CO₃ was proven to be effective for catalyzing ciprofloxacin decay at neutral pH, but the pH of the final aqueous solution became acidic [15]. Despite the solution pH decrease and inhibited efficiency for the photocatalytic degradation of organic pollutants, few studies have investigated possible solutions to this problem to maintain high catalytic performance at different solution pH values.

Bismuth tungstate (Bi₂WO₆) is the most studied visible light-driven photocatalyst due to its wide spectrum light response and lack of secondary contamination after utilization [30, 31]. Moreover, Bi₂WO₆ is composed of perovskite-like [WO₄]²⁻ layers sandwiched between [Bi₂O₂]²⁺ layers. This structure is beneficial for separating photoexcited electron-hole pairs and forming internal electric fields between slabs, which enhances photocatalytic performance [32]. In addition, Bi₂WO₆ is stable in aqueous solution without dissolving or element leaching [27, 33].

In this study, innovative approaches were proposed to expand the operating pH range for simulated solar light-mediated Bi₂WO₆ photocatalysis (SSL/Bi₂WO₆). Norfloxacin was selected as the target antibiotic compound. The performance for norfloxacin degradation at different pH values were studied in the proposed SSL/Bi₂WO₆ process. To broaden the operating pH range, efforts were undertaken to achieve high removal efficiency under extremely basic and acidic pH conditions. At an extremely basic pH, a buffer system was developed using an NH₄⁺ salt and at extremely acidic pH, the SSL/Bi₂WO₆ process was improved by adding ferric salts.

All chemicals were of analytical reagent grade and used as-received without further purification. Norfloxacin (C₁₆H₁₈FN₃O₃) was purchased from Wako Pure Chemical Industries, Ltd. and used as the target molecule for degradation. Bi₂WO₆ was synthesized using tungsten acid (H₂WO₄), citric bismuth (C₆H₅BiO₇), and ammonia (NH₃·H₂O) as precursors. The synthesis of bismuth tungstate was reported in detail in the literature [33, 34], where the characterizations of Bi₂WO₆ via X-ray diffraction (XRD), scanning electron microscopy (SEM), transmission electron microscopy (TEM), N₂ adsorption-desorption, ultraviolet-visible diffuse reflectance spectroscopy (UV-Vis DRS) have been well documented. The phase of the obtained product was evaluated by XRD and all diffraction peaks were indexed to pure orthorhombic Bi₂WO₆ with JCPDF No. 39-256. The point of zero charge was tested using a mass titration method [35]. Other chemicals used in this study including Mg(NO₃)₂, Ca(NO₃)₂, Fe(NO₃)₃·9H₂O, NaOH, HCl, and H₃PO₄ were obtained from Sigma Aldrich Inc. The mobile phase solvent for HPLC analysis (acetonitrile) was purchased from Tedia Company (USA). Distilled-deionized water with a resistivity of 18.2 MΩ of was used for preparing the mobile phase and stock solutions and was generated using a Barnstead NANOpure water treatment system (Thermo Fisher Scientific Inc., USA).

Norfloxacin degradation was conducted in a simulated solar light photochemical reactor (Beijing Changtuo Company with a Xenon lamp of 300 W). Prior to the reaction, a predetermined amount of Bi₂WO₆ was added to 100 mL of the norfloxacin solution in a 200-mL quartz beaker/cylinder and stirred under darkness for 30 min to reach adsorption equilibrium (pre-adsorption step). After reaching equilibrium, degradation was started by turning on the pre-heated light source. At preset intervals, 1 mL aliquots of the sample were withdrawn and filtered using a 0.2-µm PTFE membrane for further analysis. All experiments were performed in duplicate at room temperature (air-conditioned) at 23±1 ℃.

The amount of norfloxacin remaining in the solution was quantified using high-performance liquid chromatography (HPLC) system with a Waters 515 pump, Waters 2489 dual λ absorbance detector, and Waters 717 plus autosampler. The norfloxacin in the solution was separated from its related intermediates using a Restek pinnacle Ⅱ C18 column (5.0 μm, 4.6 × 250 mm) and quantified via adsorption at 280 nm. A mixture of 50% acetonitrile and 50% 0.025 mmol/L orthophosphoric acid was selected as the mobile phase at a flow rate of 0.8 mL/min, and 10 µL samples were injected for analysis. Parts of the experiment were run in triplicate with relative standard deviations never exceeding 10%. The pH was measured using a digital pH meter (HANNA instrument, B417). Unless otherwise specified, the pH was measured from the solution just prior to reaction. The TOC was analyzed using a Shimadzu TOC-5000A analyzer equipped with an ASI-5000A autosampler to determine the mineralization of norfloxacin using this process.

Norfloxacin degradation was performed at different pH values ranging from 3.0 to 13.0, and the decay curves are shown in Fig. 1. The decay was significantly reduced under extremely acidic (pH = 3.0) and alkaline (pH 13.0) conditions due to the electrostatic repulsion between the norfloxacin molecule and Bi₂WO₆ particle surface, arising from the pKa of norfloxacin (pK_a1 = 6.34, pK_a2 = 8.75) and the point of zero charge of Bi₂WO₆ (pH = 5.0). Three species of norfloxacin can exist in solution (cationic, zwitterionic, and anionic) and two phases of Bi₂WO₆ (cationic and anionic at a pH of 5.0) with variable pH. Under acidic (pH < 5.0) and basic pH conditions (pH > 8.75), both norfloxacin and Bi₂WO₆ mostly exist as like charges, preventing attraction between them and inhibiting norfloxacin decay.

On the other hand, a moderate pH range (i.e. pH = 5.0 to 10.8) is favorable for norfloxacin decay, where all tests at moderate pH showed ˃90% norfloxacin removal in 150 min. However, norfloxacin decay in the moderate pH range still showed certain differences as a function of pH. In particular, the optimal performance was observed at pH = 8.6, showing 93% norfloxacin removal in 60 min. From pH = 5.0 to 8.6, the norfloxacin removal efficiency gradually increased, indicating that higher pH values improved norfloxacin decomposition as long as the Bi₂WO₆ and norfloxacin remained electrostatically attractive. For pH values slightly higher than that of pKa₂ (i.e. pH at 9.0 and 10.8), inhibition of norfloxacin decay was observed, which can be attributed to the repulsive forces between norfloxacin and Bi₂WO₆ at this pH that restrict interaction between norfloxacin and the Bi₂WO₆ surface. Despite the repulsive force preventing norfloxacin removal, it was interesting that the removal efficiency at pH = 9.0 (repulsive force) was superior to that at pH = 7.5 (attractive force), suggesting the pH itself has a significant effect on norfloxacin decay, and can even be considered more critical than the electrostatic effect in some cases.

The pH variations during the pre-adsorption and reaction steps were monitored at three pH values, and the results are shown in Fig. 2. All three runs showed a continuous pH decrease, regardless of initial pH value. During the pre-adsorption stage, the pH quickly dropped from 10.0, 9.3, and 8.1 to 8.6, 7.8, and 7.5, respectively. This significant pH decrease in the pre-adsorption step indicated that hydroxyl ions in the solution were adsorbed (or at least attracted) to the Bi₂WO₆ surface. This was critical for the SSL/Bi₂WO₆ process as it provided a reservoir of precursors for the formation of hydroxyl radicals in the subsequent (reaction) step, where the hydroxyl radical was the dominating active species, as determined by quenching experiments [27]. An OH^– enriched model was proposed to describe this property of Bi₂WO₆ and is shown in Scheme 1, where the hydroxyl ions adhered and concentrated on the Bi₂WO₆ surface and its surroundings. After initiation of the photocatalytic reaction, the pH continuously decreased at a lower rate, suggesting that the hydroxyl ions were gradually consumed in the degradation process, likely due to the formation of low molecular weight organic acids and carbon dioxide.

Considering the pH decrease during the SSL/Bi₂WO₆ process, it is clear that the norfloxacin decay rate may be gradually hindered if the pH is shifted to an unfavorable range. It is well-known that only the hydroxyl ions close to the Bi₂WO₆ surface can be the source of active hydroxyl radicals due to the limited lifetime of the photo-generated holes [36]. Therefore, the maintenance of higher hydroxyl ion concentrations nearby or on the Bi₂WO₆ surface is critical for improving or maintaining rapid target decomposition.

To verify the proposed OH^– enriched model, norfloxacin decay in the presence of calcium and magnesium salts was investigated due to their similar chemical properties and large differences hydroxyl ion capture abilities. The K_sp (solubility product constant) of Mg(OH)₂ and Ca(OH)₂ are 1.5×10^-11 and 0.166, respectively. Thus, it can be deduced that the magnesium ions more effectively deplete hydroxyl ions and generate magnesium hydroxide at lower [OH^–] compared to that of calcium ions. As described in the OH^– enriched model in Scheme 1, the hydroxyl ions were adsorbed onto the Bi₂WO₆ surface to form a thin layer of concentrated OH^–. It is presumed that once the metal salt dosages are properly administered, the hydroxyl ions originally concentrated on the Bi₂WO₆ surface are removed by the magnesium ion, resulting in reduced hydroxyl radical generation. However, this procedure is ineffective when using calcium salts, due to their higher K_sp values.

Detailed investigations were performed by varying the [Ca²⁺] and [Mg²⁺] from 0 to 40 mmol/L, and the results are shown in Fig. 3. The calcium ions showed no effect on norfloxacin degradation, even when the [Ca²⁺] was increased to 40 mmol/L. However, magnesium significantly retarded norfloxacin decay and higher [Mg²⁺] resulted in a slower norfloxacin decay rates. The norfloxacin remaining after 20 min of degradation was 22.9%, 29.0%, 31.9%, 38.3%, and 42.1% at [Mg²⁺] of 0, 10, 20, 30, and 40 mmol/L, respectively. These experimental results are quite encouraging and support the assumptions presented in Scheme 1, where the concentration of hydroxyl ions surrounding Bi₂WO₆ is sufficient for the production of Mg(OH)₂, but not for Ca(OH)₂ formation, This also validates the proposed OH^– enriched model in Scheme 1. Therefore the hydroxyl ion concentration in the thin layer surrounding Bi₂WO₆ was calculated to be between 3.87 × 10^–4 and 2.03 mol/L in the tests based on the K_sp values and concentrations of magnesium and calcium salts (10 and 40 mmol/L, respectively). The calculation method is as follows, i.e. when [Mg²⁺] = 10 mmol/L, the formation of Mg(OH)₂ suggested [Mg²⁺]·[OH^–]² < 1.5 × 10^–11 (Ksp of Mg(OH)₂), thus [OH]^– was calculated as greater than 3.87 × 10^–4 mol/L; when [Ca²⁺] = 40 mmol/L, the non-formation of Ca(OH)₂ suggested [Ca²⁺]·[OH^–]² > 0.166 (Ksp of Ca(OH)₂), thus [OH]^–_- was calculated to be less than 2.03 mol/L.

The experiments showed that the pre-adsorption capacity of Bi₂WO₆ for norfloxacin was decreased in the presence of Mg²⁺, and further decreased with increasing [Mg²⁺]. When [Mg²⁺] was 0, 10, 20, 30, and 40 mmol/L, the pre-adsorption efficiency of norfloxacin was 7.1%, 6.2%, 4.8%, and 2.7%, respectively. These variations in pre-adsorption efficiency can also be ascribed to the formation of magnesium hydroxides near the Bi₂WO₆ surface, which left fewer adsorption sites available for norfloxacin and blocked the diffusion pathway for norfloxacin to the Bi₂WO₆ surface. However, in the presence of Ca²⁺ ions no pre-adsorption variation was observed because the surface properties of Bi₂WO₆, as shown in the OH^– enriched model, remained largely unchanged, even at a high [Ca²⁺] of 40 mmol/L. The pre-adsorption variation in the presence of Mg²⁺, and its insensitivity toward Ca²⁺ further justify the proposed OH^– enrichment model.

Norfloxacin decay under extremely basic pH conditions was significantly slowed, as shown in Fig. 4. At pH = 13.0, only 22.0% of the norfloxacin was removed after 150 min. Considering that the hydroxyl ion is the source of the active radical ·OH, an ammonium salt (NH₄Cl) was gradually added to the reaction solution to decrease the pH to 9.0 by forming a buffer system, as shown in Fig. 4. The ammonium salt was chosen because it is a natural degradation product of nitrogenous organic matter, and significant sources of enrichment include industrial waste, municipal wastewater treatment plants, and agricultural runoff (including animal wastes and chemical fertilizers) [37]. From Fig. 4, it is clear that after dosing ammonium, norfloxacin decay efficiency dramatically increased to 97.0% in 60 min. This observation indicates that the pH adjustment by forming a buffer system with simple addition of an NH₄⁺ salt under extremely basic condition is a practical method for the SSL/Bi₂WO₆ process.

Furthermore, a comparison experiment was performed using a non-buffered system, where the solution pH was initially set at 13.0, followed by adjustment to 9.0 using HCl. As shown in Fig. 4, norfloxacin decay in the non-buffered system was slightly slower than that of the NH₄Cl buffer system. This is because in the non-buffered system the pH gradually decreased from 9.0 to 8.8 (10 min), 8.3 (20 min), and 7.0 (150 min), which was far from the optimum working pH of 8.6. In contrast, a constant pH was maintained at 9.0 in the buffered system. In addition, TOC removal was investigated, as shown in Fig. 5. Mineralization at pH = 13.0 was negligible, whereas the buffered and non-buffered systems at pH = 9.0 showed 52.0% and 30.0% TOC removal, respectively. Comparing the buffered and non-buffered systems, it was clear that the difference in TOC removal was approximately 22.0%, higher than the difference in norfloxacin removal efficiency (i.e. 3.8% calculated from Fig. 4). This demonstrated that the mineralization of intermediates/byproduct is more efficient in the buffered system.

When the aqueous solution is acidic, ferric salts were added to the solution to form the SSL/Fe³⁺/Bi₂WO₆ process. Fe³⁺ was used because SSL/Fe³⁺ (simulated solar-light-irradiated ferric salt) is efficient at acidic pH for the evaluation of norfloxacin decay. In Fig. 6, the SSL/Fe³⁺ process showed a certain capacity for norfloxacin decay at pH = 4.0. Its mechanism can be rationalized by the oxidative properties of the ferric ions under acidic conditions when irradiated with UV light from the simulated solar light (200–400 nm) [38]. In the acidic solution, Fe³⁺ undergoes spontaneous hydrolysis with water to form a species of Fe^ⅢOH²⁺, which is a photosensitive species, and this complex can produce hydroxyl radicals directly via a photo-sensitization reaction:

(1)

Fig. 6 showed that the removal efficiencies at 20 min in the SSL/Fe³⁺/Bi₂WO₆, SSL/Fe³⁺, and SSL/Bi₂WO₆ processes were 75.3%, 44.2%, and 42.4%, respectively. Among these processes, the fastest norfloxacin decay was observed in the SSL/Fe³⁺/Bi₂WO₆ process. This superior performance at acidic pH can be ascribed to the combination of the advantages of homogeneous SSL/Fe³⁺ and heterogeneous SSL/Bi₂WO₆ processes.

A series of parallel experiments were performed at pH = 7.2 and 9.0 for the above-mentioned three processes and the results are listed in Table 1. The SSL/Fe³⁺ process was largely inert (i.e., no norfloxacin removal was observed) at both pH = 7.2 and 9.0 because the reaction described by Eq. (1) could not proceed at non-acidic pH. In addition, the SSL/Fe³⁺/Bi₂WO₆ process was slightly superior compared to the SSL/Bi₂WO₆ process at pH = 7.2 (by 5.9%) and 9.0 (by 4.0%) after 20 min of reaction. This improvement can be ascribed to the electron transfer carrier role of Fe³⁺, as shown in Eq. (2), resulting in decreased photogenerated electron-hole combination, which prolonged the lifetime of photo-generated holes and promoted the generation of ·OH radicals [39].

(2)

This observation suggests that Fe³⁺ as an electron transfer carrier also benefitted the SSL/Fe³⁺/Bi₂WO₆ process under acidic pH conditions.

To determine the optimum pH level for norfloxacin degradation by the SSL/Fe³⁺/Bi₂WO₆ process, the decay performance at different acidic pH values (i.e. 2.0, 3.0, and 4.0) was investigated and the results are shown in Fig. 7. The SSL/Fe³⁺/Bi₂WO₆ process at pH = 3.0 showed the most rapid norfloxacin decay, where 82.0%, 86.0%, and 78.3% of the norfloxacin was decomposed at a pH of 2.0, 3.0, and 4.0, respectively, in 20 min. The pH variations during all three runs were negligible and both norfloxacin and Bi₂WO₆ were positively charged with the repulsive force between the two materials preventing norfloxacin decay. Therefore, the difference in performance with pH variation was not a result of the electrostatic effect between norfloxacin and Bi₂WO₆.

It is well-known that Fe³⁺ undergoes spontaneous hydrolysis with water to form four species of Fe(Ⅲ)-hydroxo complexes, Fe^ⅢOH²⁺, Fe^Ⅲ(OH)₂⁺, Fe₂^Ⅲ(OH)₂⁴⁺, and Fe^Ⅲ(OH)₃⁰, in an aqueous solution [40]. Moreover, Flynn at el. [41] Reported that the monohydroxy complex Fe^ⅢOH²⁺, which is the most photosensitive and photoactive species, was predominant at pH = 3.0. The slowing of norfloxacin decay at pH 4.0 can be attributed to the formation of Fe^Ⅲ(OH)₃⁰ [42], which results in a decreased [Fe^ⅢOH²⁺] in the solution. On the other hand, the norfloxacin removal performance also decreased when the initial pH decreased to 2.0. This is because the hydrolysis of ferric ions at pH 2.0 was inhibited leading to slower generation of Fe^ⅢOH²⁺ [41].

Fig. 8 shows the effect of [Fe³⁺] on norfloxacin decay in the SSL/Fe³⁺/Bi₂WO₆ process at the optimum pH of 3.0. After 20 min, the removal efficiency at [Fe³⁺] of 0.1, 0.2, 0.3, and 0.5 mmol/L was 23.6%, 71.0%, 88.0%, and 86.2%, respectively. As shown in the inset of Fig. 8, [Fe³⁺] significantly changed the decay rate of norfloxacin, increasing with increased [Fe³⁺], which plateaued after the critical concentration of 0.3 mmol/L. As described previously, the Fe^ⅢOH²⁺ species is essential for norfloxacin decay in the SSL/Fe³⁺/Bi₂WO₆ process. Higher [Fe³⁺] provided a larger amount of Fe^ⅢOH²⁺ for hydroxyl radical generation. However, if Fe³⁺ is present in excess, too many ·OH radicals will be produced, resulting in the self-reaction of ·OH radicals [43]. Therefore, excess [Fe³⁺] is not suggested in practical operation.

In this study, the SSL/Bi₂WO₆ process was used to decompose the antibiotic norfloxacin. The pH significantly influenced the performance in the SSL/Bi₂WO₆ process, which generally exhibited good performance over a narrow pH range. To describing the reaction mechanism of the SSL/Bi₂WO₆ process, an OH^– enriched Bi₂WO₆ model was proposed and verified experimentally by adding Mg²⁺ and Ca²⁺. Moreover, the hydroxyl ion concentrations in the thin layer surrounding Bi₂WO₆ were determined to range between 3.87×10^-4 and 2.03 mol/L. Under acidic and alkaline pH conditions, the SSL/Bi₂WO₆ process was significantly inhibited. To broaden the operating pH range of the SSL/Bi₂WO₆ process, different modifications of the original approach were explored. At extremely basic pH, a buffer was successfully introduced into the SSL/Bi₂WO₆ process by simply adding an NH₄⁺ salt, which showed excellent performance for both norfloxacin decay and TOC removal. At the extremely acidic pH, the SSL/Bi₂WO₆ process was significantly improved by adding ferric salts (i.e. forming an alternative SSL/Fe³⁺/Bi₂WO₆ process). The improvement can be ascribed to the homogeneous photo-sensitization mechanism offered by the SSL/Fe³⁺, whereas the electron transfer carrier role of Fe³⁺ in the SSL/Bi₂WO₆ process was relatively minor. The SSL/Fe³⁺/Bi₂WO₆ process can be used over a wide acidic pH range (2–4); however, a pH of 3.0 showed slightly better performance. The optimal [Fe³⁺] for the SSL/Fe³⁺/Bi₂WO₆ process was 0.3 mmol/L, and the decay rate increased with increasing [Fe³⁺], then plateaued when ferric ion where in excess.