Efficiently removing dissolved organic compounds is a major challenge in remediation water treatment [1-3].Photocatalysis is considered as a promising tool to purify aquatic contaminantsbecause it can result in the complete photodestruction of these contaminants [3, 4]. Among them, TiO₂-based photocatalysts are extremely photocatalytic materials due to their excellent advantages, which include strongoxidative power, low cost, and stability [3, 5]. However, the photocatalytic efficiency of TiO₂-based materials is not up to the mark to meet out the practical needs because of their wide band gap and fast electron-hole (e^-/h⁺) pair recombination [3, 5]. To overcome these disadvantages, persistent efforts have beenfocused on changing the composition of TiO₂ by introducing impurities that produce donor or acceptor states in the band gap [5-7]. However, metal-doped TiO₂ usually exhibits low thermal stability and photo-corrosion [5]. Whereas, non-metal doping shows more potential towards improving the visible-light absorptionproperties compared with metal decoration [3, 5]. For example, N-doped TiO₂has been shown to drive the spectral response of TiO₂ into the visible-light region because the N2p states contribute to the band gap [8, 9]. Zeng and co-workers [10] developed N-doped TiO₂, which displayed high visible-light photoactivity for the degradation of phenol and benzene. Wang et al. [11] prepared TiO₂ nanowire arrays via nitrogen implantation, which exhibited significantly enhanced visible light photoactivity. However, multi-step procedures are required to prepare N-doped TiO₂ using traditional methods [12]. In addition, the stability of N-doping is poor under visible-light irradiation because of the low concentration of the nitrogen dopant [13]. On the other hand, carbon-decoration has great potential towards achieving a high level of visible-light performance because carbon materials can act as efficient trapping centers for photo-generated electrons [5, 7, 13]. Moreover, carbon displays strong visible-light absorption properties, which facilitate e^-/h⁺ pair transfer from TiO₂ to the pollutant molecules [5, 14]. Kim et al. [15] prepared mesoporous carbon–TiO₂ via a soft-template-carbonization route, which achieved good visible-light efficiency towards the decomposition of p-nitrophenol. Recently, our group has introduced an in situ synthesis of C-doped TiO₂ photocatalysts and the as-synthesized SiO₂@C-doped TiO₂ showed enhanced visible-light and UV-light performance for the degradation of Rhodamine B (RhB) [5]. However, the performance of single-doped/decorated TiO₂ is still below that required for the efficient use of solar light because the strategies used to prepare single-doped TiO₂ materials containing a high dopant concentration are scarce [13, 16, 17]. Therefore, novel strategies for the preparation of visible-light responsive TiO₂ with high performance are still urgently required.

Compared to single doping/decoration, co-doping has attracted a considerable attention to improve the visible-light performance of TiO₂ photocatalysts due toits synergistic effect[18, 19]. Xu et al. [20] synthesized carbon-decorated and N-doped TiO₂ nanoparticles with high visible-light photoactivity. Vaiano et al. [21] developed a phosphor modified N-doped TiO₂ (NdT/OP) photocatalyst, which can be used to degrade various organic dyes under visible-light irradiation. Li et al. [13] introduced a range of carbon/N-doped TiO₂ photocatalysts and showed that the co-decorated TiO₂ materials have increased visible-light photoactivity for the decomposition of RhB. However, the addition of carbon precursors (e.g., chitosan [13], L-lysine [20], etc.) is necessary in the preparation of carbon-doped TiO₂, as reported in the literature. Specifically, low-surface area TiO₂ nanoparticles/powder are used as the matrix in most of the previously reported studies [10, 13, 16, 19, 20, 22, 23], which significantly impair the photocatalytic efficiency and recovery observed in photocatalytic water decontamination and other applications.

In this study, we introduce a cost-effective technique for the simultaneous synthesis of carbon-modified andN-doped TiO₂ (C/N-TiO₂) hollow spheres. The C/N-TiO₂ photocatalyst not only overcomes the disadvantages of multi-step strategies because it is simultaneously prepared via the pyrolysis of a polystyrene template under an NH₃/Ar atmosphere, but also exhibits a uniform shape, controlled pore size and shell thickness, which improve the photocatalytic performance. Moreover, the band gap of TiO₂ was reduced to ~2.83 eV, whilst the carbon-decoration improves the visible-light absorption properties and efficiently separates the e^‒/h⁺ pairs. In addition, the hollow structure of C/N-TiO₂ with high surface area is able to boost the diffusion of reactants by decreasing the diffusion resistance and enhancing the accessibility, which results in an enhanced visible-light/simulated solar light efficiency.

Tetrabutylorthotitanate (TBT) was obtained from J & K Chemical Company Ltd and used without any further purification. Styrene, K₂S₂O₈, ammonia, ethanol, tetracycline (TC), and tetracycline hydrochloride (TC-HCl)were purchased from Sinopharm Chemical Reagent Co., Ltd (China), in which styrene was purified using NaOH solution (5 wt%). P25 nanoparticles (Degussa) and 2-(methacryloyloxy)ethyltrimethylammonium chloride (DMC) were purchased from Sigma-Aldrich Ltd. Deionized water was used in all our experiments.

Cationic polystyrene spheres (CPS) were synthesized according to our previous works [24-26]. An ethanol solution of CPS (0.1 g/mL, 40 mL) was added to a 100-mL three-necked flask, and an ethanol solution of TBT (0.1 g/mL, 20 mL) was slowly added to the reaction vessel under ice bath. Subsequently, NH₃/H₂O was added to the reaction mixture to give theCPS@TiO₂ precursor. The resulting solid was separated, washed, dried, and calcined under an NH₃/Ar atmosphere at 450 ℃ [10], simultaneously forming the carbon-modified, N-doped TiO₂ (C/N-TiO₂) hollow spheres.

Transmission electron microscopy (TEM), scanning electron microscopy (SEM), and STEM mapping were used to characterize the structure and morphology of the C/N-TiO₂ hollow spheres. In addition, Brunauer-Emmett-Teller (BET) method (TriStar Ⅱ 3020) was employed to obtain the surface area and porosity. The crystal phases of C/N-TiO₂ and commercial P25 were analyzed using X-ray diffraction (XRD). X-ray photoelectron spectroscopy (XPS) was used to analyze the surface properties.

The photocatalytic activity of the C/N-TiO₂ hollow spheres was investigated using the decomposition of TC and TC-HCl. 50 mL of an aqueous solution containing TC and TC-HCl (10 mg/L) and the photocatalyst (0.5 mg/L) was subjected to visible light (λ >420 nm) and simulated solar light (300 W Xe lamp equipped with an AM 1.5 filter) irradiation, respectively. Prior to irradiation, the photocatalytic system was stirred for 30 min in the absence of light to reach equilibrium. At given time intervals, ~4 mL of the suspension was removed from the reaction system, centrifuged, and measured using UV-visible absorption spectroscopy (λ_TC = 357 nm). To measure the recyclability, the used C/N-TiO₂ catalyst was recycled as follows. After the photocatalytic test, the C/N-TiO₂ photocatalyst was recovered via simple filtration, dried at room temperature, and reused in the next cycle.

Scheme 1 illustrates the preparation of the carbon-modified, N-doped hollow TiO₂ spheres (C/N-TiO₂). CPS was prepared via an emulsion-free polymerization reaction [24-26]. TBT was slowly added to the CPS solution followed by the dropwise addition of NH₃/H₂O to produce the CPS@TiO₂ precursor. The precursor was calcined at 450 ℃ under an NH₃/Ar atmosphere to remove the CPS template and produce the carbon-modified, N-doped TiO₂ hollow spheres (C/N-TiO₂).

TEM and SEM were used to characterize the structure and morphology of the as-synthesized C/N-TiO₂ hollow spheres. The TEM images indicate that the C/N-TiO₂ was hollow and porous in nature, as shown in Fig. 1(a) [5]. In particular, the almost uniform shells suggest that this synthesis technique can maintain the porosity and hollow structure during the phase transition of TiO₂, which is undoubtedly a decisive factor that affects the photocatalytic activity of the TiO₂ photocatalyst [27]. The SEM image of several broken spheres confirms the hollow nature of the C/N-TiO₂ spheres, as shown in Fig. 1(b).

EDX analysis of the C/N-TiO₂ sample was used to measure the chemical composition.Clear signals for Ti, O, C, and N were observed in theEDX survey spectrum (Fig. 2(a)), suggesting the existence of C and N species in the TiO₂ [28]. Fig. 2(c) also shows that the N signal is dispersed throughout hollow spheres, whereas the C signal (Fig. 2(f)) is mostly present on the outer layer of the hollow TiO₂ spheres, suggesting the successful integration of the different components in the C/N-TiO₂ hollow spheres. The N₂ adsorption-desorption isotherms show the same condensation steps appear in the relative pressure range of 0.1–1.0, which indicates an ordered porous structure, as illustrated in Fig. 3(a). The BET specific area of the C/N-TiO₂ hollow spheres was found to be high (up to 108.6 m²/g) according to the N₂ adsorption-desorption isotherms, and was more than two-times larger than that observed for commercial P25 (50.2 m²/g, Fig. 3(b)). This enhanced surface area improves the diffusion of the reactants by decreasing the diffusion resistance.

Fig. 4 indicates that the distinctive reflection peaks for TiO₂ observed at 25.4° (101), 37.9° (004), 48.0° (200), 54.9° (211) and 62.8° (204) can be attributed to the anatase form (JCPDS: No 21-1272) [5, 29]. No amorphous or other polymorphs (e.g., rutile or brookite) were observed in the XRD spectrum, suggesting the formation of anatase crystals via the carbonization of CPS at 450 ℃ under an NH₃/Ar atmosphere.

Fig. 5(a) shows that O, Ti, N, and C werepresent according to their binding energies, indicating that the C/N-TiO₂ hollow spheres include N, C, and TiO₂. In the Ti 2p spectrum, the characteristic peak at 464.29 eV was indexed to Ti 2p_1/2 and the other peak observed at 458.41 eV corresponds to Ti 2p_3/2, as shown in Fig. 5(b). Furthermore, the Ti2p3/2 peak of C/N-TiO₂ was positively shifted by 0.17 eV (458.41 eV) when compared to the Ti 2p_3/2 peak observed at 458.24 eV in the spectrum of commercial P25. The Ti 2p_1/2 binding energy of the commercial P25 sample was observed at 463.96 eV, which was again observed at an enhanced binding energy (0.33 eV) in the C/N-TiO₂ hollow spheres. Clearly, the positive shiftcan be ascribed to the strong interactions formed between Ti⁴⁺ and the C and N elements [18], indicating the presence of lattice distortion in the C/N-TiO₂ hollow spheres [5, 18]. In addition, Fig. 5(c) shows that the O 1s peaks were observed at 529.97 and 532.82 eV. Compared with the O 1s spectrum of the commercial P25 sample, a clear positive shift was observed in the O1s spectrum of the C/N-TiO₂ hollow spheres, suggesting the existence of oxygen vacancies [3, 5].

Fig. 5(d) shows the high-resolution N1s spectrum of the C/N-TiO₂ hollow spheres. A broad peak was observed from 396 to 403 eV, which corresponds to the characteristic peak of N-doped TiO₂, as reported in the literature [1, 9, 15]. After fitting the curve, four peaks at 398.4, 399.43, 401.15, and 401.89 eV were observed for N-TiO₂ and N-TiO_2-x, which are consistent with the N1s binding energy region and hollow TiO₂@N-doped carbon [8, 9, 30]. Fig. 5(e) shows the C1s spectrum of the C/N-TiO₂ hollow spheres, which can be fitted into three peaks at 285.24, 286.56, and 289.53 eV, suggesting the existence of carbon in three different chemical environments. The peak observed at 285.24 eV corresponds to residual elemental carbon, which acts as a photo-sensitizer to enhance the visible-light absorptionof the C/N-TiO₂ hollow spheres [31, 32]. The two characteristic peaks observed at 286.56 and 289.53 eV were assigned to the existence of oxygen bound species[9, 30, 31]. Therefore, the XPS measurements confirm the formation of carbon-modified, N-doped TiO₂ hollow spheres via the pyrolysis of CPS under an NH₃/Ar atmosphere.

As shown in Fig. 6(a), it is clear that commercial P25 only absorbs UV light because of its wide band gap [1, 5]. In contrast, the C/N-TiO₂ photocatalyst not only displays a strong UV light response, but also exhibits a wide visible-light response. Obviously, the visible-light absorption can be ascribed to the co-decoration of TiO₂ by N and C, which results in a dramatic enhancement in its visible-light absorption properties [1, 5]. The band gap energy (E_g) values for commercial P25 and the C/N-TiO₂ hollow spheres were calculated according to Kubelka-Munk method, as shown in Fig. 6(b) [13]. Excitedly, C/N-co-doping was able to narrow the band gap energy (2.83 eV) of the C/N-TiO₂ photocatalyst, which facilitates the excitation of the electrons in C/N-TiO₂ under visible-light irradiation and thus, leads to its enhanced photocatalytic efficiency.

The shell thickness of the C/N-TiO₂ hollow spheres was easily controlled byvarying the amount of TBTprecursor used [33], as shown in Fig. 7. When 2 g of TBT was used, the C/N-TiO₂hollow spheres have a shell thickness of 15 nm determined from from the cross-sectional TEM images of the sample obtained using digital micrograph software (Fig. 7(a)). When 3 g of TBT was added, an enhanced shell thickness (~22 nm) was obtained, as shown in Fig. 7(b). When 4 g of TBT was added, the shell thickness of the C/N-TiO₂hollow sphereswas ~29 nm. Moreover, we further measuredthe effect on the surface area when increasing the amount of TBT used. Figs. 3 and 8 showthat the surface area of the three C/N-TiO₂ samples was 108.6, 70.6, and 53.3 m²/g when using 2, 3, and 4 g of TBT, respectively. In addition, the BET isotherms demonstrate that the shell thickness does not significantly change the adsorption type, pore structure and size of the C/N-TiO₂ hollow spheres. Therefore, the shell thickness (15–29 nm) of the C/N-TiO₂ hollow spheres can be easily controlled by changing the amount of TBT added during their synthesis.

Simultaneous modification/doping can significantly enhance the photocatalytic activity of TiO₂ because of its synergistic effect. In this work, we have demonstrated its applicability for the decomposition of antibiotics under visible-light and simulated solar light irradiation. Single-decorated TiO₂ (e.g., carbon-modified hollow TiO₂ (C/TiO₂)) was used as a reference photocatalyst under the same experimental conditions.

Fig. 9 shows the visible-light catalytic performance of the photocatalysts (C/TiO₂ and C/N-TiO₂) for the degradation of tetracycline (TC). As shown in Fig. 9(a), with the increase of illumination time, the characteristic absorption band of TC (~357 nm) does not exhibit any obvious change in the absence of a photocatalyst. After the addition of the C/TiO₂ photocatalyst, the characteristic absorption band of TC decreases gradually upon visible-light irradiation (Fig. 9(b)), indicating that the C/TiO₂ photocatalyst can effectively speed up the degradation of TC. In contrast, a faster decrease in the characteristic absorption band of TC was observed for the as-synthesized C/N-TiO₂ photocatalyst, as shown in Fig. 9(c). The time course for the decrease in the TC concentration using the photocatalysts is shown in Fig. 9(d). Tetracycline (TC) itself cannot be degraded under visible-light irradiation. In the presence of a photocatalyst, the as-synthesized C/N-TiO₂ photocatalyst can degrade 86.3% of TC, which is beyond the TC degradation value (76.4%) of the C/TiO₂photocatalystwithin 30 min after irradiation.The enhanced visible-light photocatalytic performance was also further confirmed by degrading tetracycline hydrochloride (TC-HCl) using the two photocatalysts (C/TiO₂ and C/N-TiO₂) under the same photocatalytic conditions, as shown in Fig. 10.

The degradation of TC was further investigated using the two photocatalysts under simulated solar light irradiation, as shown in Fig. 11. Compared with the visible-light removal rate of TC, the degradation value of TC was increased to 23.1% after 30 min of irradiation. In contrast, the removal rate of TC using C/TiO₂ was increased to 98.3% under the same photocatalytic conditions. However, the as-synthesized C/N-TiO₂ photocatalyst exhibited an TC removal efficiency of 99.6%, indicating that the TC was almost completely degraded under simulated solar light irradiation. Based on the previously reported studies [34, 35], the degradation of pollutants can be attributed to a pseudo-first-order reaction using the simplified Langmuir-Hinshelwood model when C₀ is very low.

where, k and C are the apparent first-order rate constant and concentration of TC in solution at a certain time, respectively. The degradation of TC is slow using C/TiO₂ with a rate constant (K_a) of 0.1311 min^-1, whereas the as-synthesized C/N-TiO₂ hollow spheres further improve the degradation rate of TC to K_a = 0.1812 min^-1, which approaches a 1.4-fold increase than C/TiO₂. Obviously, N-doping plays an important role in improving the degradation rate of TC. The enhanced photocatalytic efficiency was further confirmed by the decomposition of tetracycline hydrochloride (TC-HCl), as demonstrated in Fig. 11(c) and (d). All the data strongly confirm that co-decoration improves the photocatalytic performances of TiO₂ due to the synergistic effect [18, 19], as expected.

In order to achieve the optimal photocatalytic activity, the most suitable shell thickness needs to be determined. Fig. 12 indicates the photocatalytic performance of the three representative C/N-TiO₂ hollow spheres towards the degradation of TC under visible light and simulated solar light irradiation. Specifically, when the shell thickness was ~15 nm (C/N-TiO₂-2TBT), the highest photocatalytic activity was observed. When the thickness was increased to ~22 nm (C/N-TiO₂-3TBT), a slightly decreased degradation of TC was observed. When shell thickness was further increased to ~29 nm (C/N-TiO₂-4TBT), the degradation rate of TC further decreases accordingly. This trend was mainly attributed to the decreased surface area (Fig. 8) indicating that an optimal shell thickness was important for the C/N-TiO₂ hollow spheres because it offers more space and active sites for the adsorption of TC.

To elaborate the increased photocatalytic performance of the C/N-TiO₂ hollow spheres, we have proposed a possible mechanism for the degradation process. When the oxygen atoms are replaced with nitrogen atoms, a new impurity level is formed between the valence and conduction band of TiO₂. Under visible-light irradiation, the electrons are excited to the conduction band of TiO₂ from the impurity level (~2.83 eV) [1, 36]. In addition, the carbon-coating can exhibit many advantages including the enhanced adsorption of TC or TC-HCl, efficient electron trapping, and decreased charge transfer resistance [2, 7]. Therefore, the excellent photocatalytic behavior can be ascribed to the synergistic effect of the nitrogen and carbon species in the as-synthesized C/N-TiO₂ hollow spheres.

The photodegradation of TC exhibits a typical ·OH oxidation mechanism and pathway because larger intermediates with an m/z = 461 produced via the hydroxylation reaction are detected in this study [37, 38], as shown in Fig. 13.The product with m/z = 445 was ascribed to the molecular weight of TC [39]. After TC was attacked by ·OH, intermediate 1 (m/z = 461) was produced. Subsequently, the N-methyl groups are gradually cleaved to generate intermediates 2 (m/z = 430) and 3 (m/z = 416). Because the amino group and carbocyclic ring in intermediate 3 was further lost to produce intermediate 4 (m/z = 345), intermediate 4 could be further oxidized to form intermediates 5 (m/z = 303) [40] and 6 (m/z = 276).Intermediate 7 (m/z = 136) is also detected because of the cleavage of the carbocyclic ring. CO₂, H₂O and inorganic ions are ultimately formed in the presence of the C/N-TiO₂ photocatalyst.

A series of recycling experiments were carried out for the decomposition of TC under visible-light irradiation using the C/N-TiO₂ photocatalyst in order to evaluate its potential application [2, 41, 42]. Four consecutive recycling runs shows that the as-synthesized C/N-TiO₂ photocatalyst exhibits an almost unchanged degradation rate during the four consecutive runs (Fig. 14), indicating its high recyclability for the treatment of wastewater [2, 41]. Clearly, the good durability is able to be responsible for the unique synergistic effect of C and N species in the TiO₂ hollow spheres.

In summary, we have devised a novel strategy to prepare a carbon-modified and N-doped TiO₂ hollow photocatalyst. The pore size and shell thickness C/N-TiO₂ are readily controlled by the CPS template and tuning the amount of TBT precursor used. After carbonization of the CPS template under an NH₃/Ar atmosphere at 450 ℃, the nitrogen element is doped in the place of oxygen, simultaneously forming the C-modified and N-doped anatase TiO₂ hollow spheres. The results confirm that the N-doping has been formed and the carbon species were decorated on the surface of the photocatalyst, narrowing the band gap of TiO₂, and enhancing the visible-light absorption of TiO₂. Therefore, when the C/N-TiO₂ hollow spheres were employed as a photocatalyst for the decomposition of TC and TC-HCl, and the C/N-TiO₂ photocatalyst showed an enhanced efficiency when compared to C/TiO₂. Specifically, the C/N-TiO₂ photocatalyst showed excellent recyclability for water decontamination, which considerably increased its potential for actual application in industry.