In the past few decades, water pollution has become a global problem, which seriously threatened the ecological environment and human health. Among numerous pollutants existed in wastewater, pathogenic bacteria [1], benzene based organic dyes [2], and endocrine-disrupting chemicals (EDCs) [3] are three kinds of major contaminants that arouse wide public concerns. Pathogenic bacteria could exert an influence on aquatic ecosystem and pose an epidemical risk to human health [4]. Benzene based organic dyes, one kind of potential carcinogens, could lead to high chromaticity and toxicity of water body [5]. EDCs could mimic the bioactivity and disturb the metabolic process of natural hormones, which would damage the endocrine function of humans and even cause the fetal malformation [6]. Consequently, it is a significant task to seek an efficient technology to remove pathogenic bacteria, organic dyes, and EDCs from wastewater. Compared with traditional treatments, photocatalysis technology shows a great potential in environmental purification because of its mild reaction conditions, strong oxidation ability, and no harmful by-products [7]. Graphitic carbon nitride (g-C₃N₄), an organic semiconductor, shows an outstanding potential in the fields of environment remediation due to its low cost, high stability, visible-light response, and appropriate electronic band structure [[8], [9], [10], [11], [12]]. However, poor light utilization, low specific surface area, and rapid charge recombination have severely limited the further photocatalytic application of g-C₃N₄ [13,14]. Since the polymeric framework of g-C₃N₄ make it feasible to tune its textural, chemical, and electronic structures, varieties of strategies have been adopted to overcome the above problems and improve the photocatalytic activity of g-C₃N₄, including hierarchical structure construction [15,16], defect engineering [17,18], morphology manipulation [[19], [20], [21]], elemental doping [22,23], heterojunction designing [24,25], and noble metals decoration [26,27].

Among these approaches, hierarchical structure construction is regarded as a promising way to optimize g-C₃N₄ as they can strengthen the light harvesting, increase the exposed surface, promote the reactant diffusion, and accelerate the charge transfer [28]. Tubular and porous structures are two kinds of popular morphology for g-C₃N₄ that have attracted extensive attention. Tubular g-C₃N₄ possesses the superiorities of both 1D and hollow structures, which bring about several unique advantages, such as facilitating the light reflection/scattering, providing internal and external active sites, and endowing the electron with oriented transfer channels [29,30]. Meanwhile, the well-developed porosity of porous g-C₃N₄ can lead to large specific surface area, abundant exposed active sites, and low mass transfer resistance [31,32]. Inspired by these respective structural merits, the fabrication of microtubular nanoporous g-C₃N₄ with hierarchical structure should be an efficient pathway to enhance the photocatalytic performance by taking advantage of both tubular and porous structures. The hierarchical g-C₃N₄ with microtubular exposed edges and easily accessible nanoporous networks can not only gain higher adsorption capacity for pollutants and offer more active sites for charge transfer, but also avoid the aggregation of catalyst particles and facilitate the separation of catalysts from the reaction system [33]. At the same time, the introduction of special point defects into g-C₃N₄ framework, such as carbon or nitrogen vacancies, can modify the surface property, tune the electronic structure, and serve as shallow charge trap sites to promote charge separation, and finally enhance the photocatalytic quantum efficiency of g-C₃N₄ [34,35]. Therefore, it is highly desirable and challenging to achieve g-C₃N₄ photocatalyst with novel hierarchical structure and point defects simultaneously.

Generally, to obtain the especial architecture, hard/soft-templating approach is commonly adopted [36,37]. However, this kind of preparation method is neither cost-efficient nor environmental-friendly, as it usually requires complex procedures of template modification and the involvement of poisonous etchants [38]. Additionally, hard-templating often restrain the subsequent functionalization while soft-templating may lead to the lattice disorder in the matrix, seriously restricting the widespread application [39]. Thus, it is an urgent desire for developing an effective, economical, and green approach to synthesize the microtubular nanoporous g-C₃N₄. In recent years, molecular self-assembly has emerged as a versatile self-templating method for preparing hierarchical materials with unique morphology under mild conditions [40]. It enables two or more kinds of organic molecules to regularly assemble into large-sized and ordered supramolecular aggregates by noncovalent interactions including hydrogen bonding and π-π stacking due to their strong direction and saturation properties [41]. To date, specially-shaped g-C₃N₄ has been obtained using hydrogen bonding and π-π stacking induced supramolecular intermediate as the precursor by controlling the synthesis temperature and adjusting the pH value of aqueous solution [42]. Therefore, the design of ordered supramolecular precursor is very crucial to the fabrication of g-C₃N₄ with micro-nanostructure.

In this work, microtubular nanoporous g-C₃N₄ (TPCN) with hierarchical structure and nitrogen defects was prepared via a molecular self-assembly approach. As melamine can hydrolyze into cyanuric acid at an appropriate pH value, it was utilized as the raw material to in situ produce hexagonal melamine-cyanuric acid supramolecular precursor under acetic acid-assisted hydrothermal process. The acetic acid could not only adjust the pH value to prompt the hydrolysis process, but also fine tune the skeletal structure of precursor to construct defect engineered TPCN. The formation mechanism of the prism-like supramolecular intermediate and TPCN with micro-nanostructure was discussed in detail. The morphology, structure, and properties of TPCN were investigated by various techniques. The disinfection and degradation activities of TPCN photocatalyst were carefully evaluated under visible light irradiation towards typical pathogenic bacteria, organic dyes, and EDCs, which were remarkably enhanced in comparison with bulk g-C₃N₄ (BCN). Moreover, the important roles of the hierarchical structure and nitrogen defects played in the photocatalytic process of TPCN system were also systematically elucidated to understand the enhanced activities.

Firstly, 2 g of melamine was put into a flask with 90 mL of acetic acid solution in a specific concentration (2%, 5%, 8% (v/v)) and refluxed in a heating mantle at 100 °C for 30 min. Then the solution was transferred into a stainless autoclave with a Teflon-inner-liner. The autoclave was sealed and put into an oven, kept at 180 °C for 10 h. After the hydrothermal process, the suspension was centrifuged, washed with deionized water, and dried at 60 °C. The obtained solids were calcinated at 520 °C for 4 h under N₂ atmosphere with a heating rate of 3 °C min^-1. The final products were designated as TPCN-X photocatalysts, X labeled as the concentration of acetic acid in preparation, which were 2, 5, and 8, respectively. BCN, prepared by heating melamine at 520 °C for 4 h with a rate of 3 °C min^-1 in N₂, was selected as the reference g-C₃N₄.

The morphological images of samples were taken using scanning electron microscopy (SEM, FEI Quanta 200) and transmission electron microscopy (TEM, JEOL- 2100). The Brunauer-Emmett-Teller (BET) specific surface area, pore volume, and pore size distribution of the samples were estimated by N₂ adsorption and desorption isotherms with TriStar II 3020 instrument (Micromeritics) at 77 K. X-ray diffraction (XRD) patterns of the powder samples were characterized on a Bruker D2-phaser X-ray diffractometer at room temperature with a monochromatized Cu Kα radiation. Fourier transformed infrared (FTIR) spectra were performed on a Nicolet iS10 spectrometer (Thermo Fisher Scientific) in the frequency range of 4000 and 400 cm^-1. UV-vis diffuse reflectance spectra (DRS) of the powder samples were recorded between 200 and 800 nm on a UV-3600 plus spectrometer (Shimadzu). Photoluminescence (PL) spectra were obtained on a fluorescence spectrometer (Hitachi F-7000) with an excitation of 363 nm incident light at room temperature. The time-resolved fluorescence decay spectra were recorded at room temperature on an Edinburgh FLS920 spectrophotometer using a 375 nm nanosecond pulse laser as the excitation source. Elemental analysis results were collected by a vario Micro cube (Elementar). X-ray photoelectron spectroscopy (XPS) analysis was performed on a Thermo Fisher ESCALAB 250Xi system with Al Kα radiation, and C 1s at 284.8 eV was used to internally calibrate the binding energies. The total organic carbon (TOC) was measured with a Multi N/C 2100S TOC analyzer. ESR signals of radicals were recorded on a Bruker EMXplus spectrometer with a modulation frequency of 100 kHz and a microwave power of 20 mW.

The synthetic process of TPCN photocatalyst including four steps was illustrated in Scheme 1. Firstly, melamine was dissolved in acetic acid solution after heating, and then partially in situ hydrolyzed into cyanuric acid at a rather slow rate during the acetic acid-assisted hydrothermal process. Secondly, the generated cyanuric acid and the remaining melamine immediately self-assembled into melamine-cyanuric acid supramolecular hexamer in the same plane through multiple hydrogen-bond interactions. Thirdly, different planar supramolecular sheets stacked in a vertical direction via π-π interaction to form the hexagonal prism-like supramolecular precursor (Fig. S1 in Supplementary Material) [40]. Finally, the precursor acted as the self-template for the synthesis of hexagonal tubular TPCN. As quantities of gases, such as NH₃, NO_x, and CO₂, were released during this pyrolysis process under N₂ atmosphere, nanoporosity and nitrogen defects were also formed to be embedded into g-C₃N₄ framework [39].

The morphology transformation from supramolecular precursor to TPCN was studied by SEM and TEM analysis. As shown in Fig. 1(a), the supramolecular precursor for TPCN-5 exhibits a hexagonal rod-like microstructure with a length of 100-400 μm and a diameter of 20-50 μm. Fig. 1(b) shows that the rods are composed of two-dimensional sheets stacking in a perpendicular direction via π-π interaction. As shown in Fig. 1(c-e), the hexagonal prisms transform into hexagonal tubes after calcination, indicating that typical sample TPCN-5 well maintained the external hexagonal morphology of precursor during the polycondensation process. Similar hexagonal tubular microstructures are also observed for TPCN-2 or TPCN-8 samples (Fig. S2 in Supplementary Material). The tubular morphology of TPCN-5 is further confirmed by TEM (Fig. 1(f)), which is quite different from that of BCN with irregular particle shapes (Fig. S3 in Supplementary Material). From the magnified images (Fig. 1(g, h)), it is found that the tube walls of TPCN-5 consist of extensive nanosheets embedded with numerous nanopores of tens of nanometers. This hierarchical micro-nanostructure is expected to provide multiple light reflection/scattering channels and more exposed active edges that contribute to the photocatalytic activity of TPCN [30].

To measure the nanoporous structure in detail, the BET specific surface area and pore size distribution were analyzed by N₂ adsorption-desorption isotherms (Fig. 2(a)). As shown in Table 1, all TPCN samples possess increased BET specific surface area and pore volume in contrast to BCN, resulting in more active sites. Among them, TPCN-5 exhibits the largest surface area (72.3 m² g^-1) and pore volume (0.30 cm³ g^-1), which are about 21.3 and 14.2 times as high as those of BCN (3.4 m² g^-1 and 0.021 cm³ g^-1), respectively. The nitrogen adsorption-desorption isotherms of three TPCN all exhibit type IV curves with hysteresis loops, which are indicative of the mesoporous structures. Compared with BCN, increasingly apparent pore size distributions of TPCN samples can be observed (Fig. 2(b)), which all exhibit a broad peak at 5-60 nm [31].

The crystal and chemical structures of as-prepared TPCN were characterized by XRD and FTIR spectra, respectively. The XRD pattern of BCN shows two characteristic diffraction peaks at 12.8° and 27.4° related to in-planar packing of heptazine units and inter-planar stacking of conjugated C—N heterocycles, respectively (Fig. 3(a)) [19]. By contrast, the (100) peak becomes broader while the (002) peak turns weaker with the increasing concentration of acetic acid, which could be attributed to the strong quantum size effect due to the microtubular nanoporous structure of TPCN [41]. The FTIR spectrum of BCN display three characteristic absorption bands at 808, 1200-1650, and 2900-3500 cm^-1, assigned to the breathing vibration of the heptazine units, the stretching vibration of aromatic heterocycles, and the stretching vibrations of N—H and O—H bonds, respectively (Fig. 3(b)) [27]. In the case of TPCN, a small peak emerges around 3350 cm^-1, which should be attributed to the increased amount of NH_x groups within the TPCN skeleton. Meanwhile, the stretching vibration of aromatic heterocycles of TPCN strengthens obviously compared to that of BCN, indicating that there might be π-π interaction between the different planar sheets of TPCN to preserve the hexagonal tubular structure from the prism-like supramolecular precursor after the pyrolysis process.

The light absorption properties and band structures of as-prepared photocatalysts were investigated by DRS and Mott-Schottky (MS) analysis. As shown in Fig. 4(a), compared with BCN, three TPCN all possess slightly red shifted absorption edges and improved light harvesting in longer wavelength region (λ > 450 nm), which should be mainly owing to the formation of the nitrogen vacancies (V_N) and the multiple reflection/scattering channels of incident light within the tubular structure of TPCN [42]. However, three TPCN exhibit slightly depressed light absorption in the range of 200-450 nm, which could be attributed to the quantum confinement effect induced by the nanoporous structure [39]. Additionally, the band gap of TPCN-5 calculated from Tauc-Plot (Fig. S4(a)) is 2.71 eV, which is smaller than 2.76 eV of BCN. According to the MS plots with the positive slope (Fig. S4(b)), BCN and TPCN-5 are assigned to n-type semiconductor whose conduction band edge potential (V_CB) is approximately equal to the flat band potential (V_fb) [43]. According to the intercept values of linear potential curves, the V_fb of BCN and TPCN-5 are about -1.10 and -1.00 V (vs. SCE), equivalent to -1.06 and -0.96 V (vs. NHE), respectively. The corresponding V_CB of TPCN-5 positively shifts by 0.10 eV in comparison of BCN, indicating that the electronic structure has been altered due to V_N existed in TPCN framework according to the previous reports [35]. Based on the V_CB and band gap values, the calculated valence band edge potential (V_VB) of TPCN-5 (1.75 V vs. NHE) is more positive than that of BCN (1.70 V vs. NHE), indicating that the generated holes (h⁺) of TPCN-5 possess more powerful driving force during the photooxidation process [27,44]. The electronic band structures of BCN and TPCN-5 are illustrated in Fig. 4(b).

During the pyrolysis process of supramolecular precursor, quantities of gases were released as by-products, resulting in the nitrogen defects formed in the framework of TPCN. The existence of V_N was further confirmed by elemental analysis, ESR, and XPS measurements. As shown in Fig. 5(a) and Table S1 in Supplementary Material, the elemental analysis results reveal the detailed compositions of C, N, and H in BCN and TPCN. The C/N atomic ratio is 0.6779 for BCN, while it gradually increases for TPCN. The loss of N atom could be corresponding to the increasing number of V_N [45]. As TPCN-5 shows the highest C/N ratio (0.6843), it should contain the largest amount of V_N among three TPCN photocatalysts. Since the generation of V_N could result in the delocalized electrons of carbon atoms within heptazine rings of g-C₃N₄, ESR spectra was recorded to prove the presence of unpaired electrons in the localized π-conjugated structure (Fig. 5(b)) [35]. Although both samples exhibit an asymmetric Lorentzian signal line centered at g = 2.003, the ESR signal intensity of TPCN-5 is much stronger than that of BCN, suggesting that more V_N are existed in the framework of TPCN-5 [46]. XPS analysis was performed to further investigate the chemical states and the specific location of V_N in the heptazine rings of TPCN (Fig. S5 in Supplementary Material). The XPS N 1s spectra comprise N (C - N—C) in heptazine rings (N_2C, 398.8 eV), N in N-(C)₃ (N_3C, 400.1 eV), and N in -NH_x (401.2 eV) (Fig. 5(c)) [18]. Compared with BCN, the intensity of N_2C peak slightly increases while that of N_3C peak decreases. As displayed in Table 2, the N_3C/N_2C ratio of peak area decreases from 0.229 (BCN) to 0.176 (TPCN-5), suggesting there should be V_N generated in the N_3C sites of TPCN-5 (Fig. 5(d)) [47]. Additionally, the NH_x/N_3C ratio for TPCN-5 (0.745) is bigger than that for BCN (0.535). It has been reported that the generated V_N in g-C₃N₄ can act as shallow charge traps to accelerate charge transfer efficiency [35], while NH_x groups can induce hydrogen-bonding interaction between C₃N₄ layers to facilitate the interlayer charge transport [48]. As TPCN-5 shows lower N_3C/N_2C and higher NH_x/N_3C ratios, it should contain a greater number of V_N and NH_x groups to facilitate its photocatalytic activity as discussed below.

The potential of TPCN in photocatalysis application was evaluated by the antimicrobial experiments. Escherichia coli (E. coli), one of the most common pathogenic bacteria, was chosen as the probe microbe. As displayed in Fig. 6, almost all bacteria can survive in the light control treatment, while about 90.1 % of bacteria remain after 4 h in the dark control treatment, suggesting neither visible light illumination nor TPCN-5 itself can cause the inactivation of E. coli effectively. After 4 h of visible light irradiation, it can be found that all TPCN photocatalysts present significantly enhanced disinfection activity in contrast with BCN. Among three TPCN samples, TPCN-5 possesses the highest bactericidal efficiency. It can make about 99.2 % of E. coli cells inactivated, whereas about 74.4 % of E. coli cells are killed by BCN. Additionally, the XPS analysis of the used TPCN-5 photocatalyst was carried out to clarify the variation of nitrogen defects in the photoreaction. As shown in Fig. S6 in Supplementary Material, the N 1s spectrum of the used TPCN-5 is almost similar to the original one, indicating that the nitrogen defects in the framework of TPCN has not changed after photocatalytic reaction.

The photocatalytic inactivation performance of TPCN toward E. coli cells was further verified by fluorescent-based cell live/dead tests. Fluorescein isothiocyanate (FITC) and propidium iodide (PI) were used to stain the DNA of E. coli cells. FITC is a cell-permeable green-fluorescent stain that labels both live and dead bacteria, whereas PI is a cell-impermeable red-fluorescent stain that only labels dead bacteria [49]. As shown in Fig. 7, there are few dead cells observed in the light control and dark control groups. Meanwhile, only partial amount of E. coli cells is labeled with red fluorescent in BCN group under visible light irradiation, indicating that a number of bacteria can still survive. In contrast, after irradiation treatment with TPCN photocatalysts, the bacterial cells exhibit stronger red fluorescence, suggesting that more E. coli cells become dead due to the ruptured cellular membranes. Among three TPCN treatment groups, E. coli cells treated with TPCN-5 under exposure of light display the strongest red fluorescence, almost all the E. coli cells are stained by PI. Thus, TPCN-5 shows the best bactericidal efficiency. The fluorescent staining results are in accordance with those of the CFU counting method, demonstrating that TPCN photocatalyst possesses an excellent disinfection activity via damaging the cellular membranes of bacteria.

Moreover, SEM imaging was performed to demonstrate the morphological changes of the treated E. coli. The initial E. coli cells show a micrometer-sized rod-like morphology with intact cell membranes (Fig. 8(a)). After mixing with TPCN-5, E. coli cells are firmly adsorbed on the surface of the tubular matrix, which could accelerate the antibacterial process (Fig. 8(b)). As shown in Fig. 8(c, d), after 4 h of visible light illumination, most of E. coli cells become malformed with obviously wrinkled cell walls as pointed out by the green arrow, and some cavities are formed on the cell surfaces as pointed out by the red arrow. Thus, we can suppose that E. coli cells should be inactivated by the reactive species generated from TPCN that could attack the cellular membranes and result in the leakage of intracellular contents [50].

Subsequently, to evaluate the photocatalytic degradation performance of TPCN under visible light irradiation, cationic dye (methylene blue, MB), anionic azo pigment (amaranth) and colorless EDC (bisphenol A, BPA), were chosen as three contaminant models. The apparent rate constant (k) of the photodegradation reaction can be calculated by the pseudo first-order kinetic equation [2]. As shown in Fig. S7 in Supplementary Material and Fig. 9(a), compared with BCN, TPCN exhibits improved photocatalytic degradation activities, which is consistent with the disinfection result. The determined apparent k of TPCN-5 against MB, amaranth, and BPA are almost 3.1, 2.5, and 1.6 times as fast as than those of BCN, respectively. Accordingly, TPCN photocatalyst has a vast potential in the field of pollutants degradation.

To illustrate the mineralization efficiency after photodegradation, the TOC test was conducted in the photodegradation process of MB by BCN and TPCN-5 (Fig. S8 in Supplementary Material). The TOC removal percentage is 18.2 % and 44.2 % for BCN and TPCN-5 after 5 h of reaction, respectively. Thus, the mineralization ability of TPCN has been evidently enhanced. Moreover, the high performance liquid chromatography (HPLC) analysis results indicate that the strong oxidative ability of TPCN-5 can thoroughly mineralize not only MB but also its intermediates to CO₂ and H₂O after 8 h irradiation (Fig. S9). The adsorptivity of TPCN-5 towards the target pollutant is also improved due to the remarkably increased BET surface area and the well-developed nanoporosity (Fig. S10 in Supplementary Material), which should be an important factor for the enhanced photocatalytic activity of TPCN. The cycling test for the photodegradation of MB was further performed (Fig. 9(b)). It can be found that the photocatalytic activity of TPCN-5 has not decreased apparently after four runs under visible light, indicating the excellent photostability of TPCN photocatalyst.

The highly efficient photocatalytic disinfection and degradation activities of TPCN should be attributed to the well-organized hierarchical structure and nitrogen defects, which can endow TPCN with large specific surface area, abundant exposed active sites, strengthened light harvesting ability, and accelerated charge transfer efficiency.

3.5.1. Photogenerated charge behavior studies

To prove the important roles of hierarchical structure and nitrogen defects played for the enhanced activities of TPCN photocatalyst, steady state PL spectra, time-resolved fluorescence decay spectra, and photoelectrochemical experiments were performed to study the dynamic process of photogenerated charge carriers in TPCN.

As shown in Fig. 10(a), BCN shows a strong PL emission peak centered at about 464 nm, while three TPCN samples all exhibit the evident PL quenching phenomena and the peak positions are red shifted to 480 nm. The decreased peak intensity indicates a lower recombination of charge carriers for TPCN comparing with BCN [26]. The charge transfer kinetics of TPCN was confirmed by time-resolved fluorescence decay spectra. It could be seen in Fig. 10(b) that all TPCN photocatalysts exhibit slower decay kinetics than BCN. The decay profiles of four samples were fitted by biexponential functions with shorter and longer lifetimes of charge carriers (τ₁ and τ₂, respectively). As shown in Table 3, the lifetimes of three TPCN samples are longer than those of BCN, and TPCN-5 shows the longest ones. τ₁, τ₂, and τ_av are 1.55, 8.94, and 6.92 ns for TPCN-5, while 1.39, 5.42, and 3.51 ns for BCN, respectively. Notably, TPCN possesses longer τ_av than BCN, indicating that more charge carriers could participate in the photocatalytic redox reaction. The quenched PL emission and prolonged lifetimes of charge carriers can be ascribed to the microtubular nanoporous structure with appropriate amount of V_N and NH_x groups, which is favorable for the charge transfer and finally enhance the photocatalytic activities of TPCN [48].

The accelerated charge transfer efficiency in TPCN was further verified by the photocurrent and electrochemical impedance spectroscopy (EIS) measurements. As shown in Fig. 11(a), all TPCN samples show a reinforced visible-light-driven photocurrent in comparison with BCN. Among them, TPCN-5 has the strongest current density, which is nearly 2.0 times as high as that of BCN, illustrating the mobility of electron-hole pairs for TPCN has been effectively improved [24]. The consistent result was also found in EIS Nyquist plots (Fig. 11(b)). The arc radii of TPCN-5 are smaller than those of BCN under both dark and visible light conditions. As a smaller radius represents a low charge transfer resistance on the surface of the working electrode, TPCN-5 should possess faster charge migration rate and higher charge separation potency due to the defects-rich hierarchical structure [20].

The above results certify that the hierarchical structure and nitrogen defects both play an important role to improve the photocatalytic activities of TPCN by affecting its dynamic process of charge carriers as follows. The hexagonal microtubes can provide oriented long-distance channels for charge transport, and the well-developed nanoporosity can induce accessible porous channels for charge migration. V_N can serve as the shallow charge trap sites to prompt the charge separation, and the interaction generated between NH_x groups can facilitate the interlayer charge transport.

3.5.2. Reactive species generated in photocatalytic process

To identify the reactive species generated in the photocatalytic process of TPCN, ESR analysis was conducted and 5,5-dimethyl-1-pirroline-N-oxide (DMPO) was used as the spin-trapping probe. Hydroxyl radical (•OH) and superoxide radical (•O₂^-) were measured in H₂O and dimethyl sulfoxide (DMSO), respectively. As shown in Fig. 12(a), the signal of DMPO-•OH adducts cannot be found under both dark and visible-light conditions, implying that there should be no •OH generated in TPCN-5 system. This is because the V_VB of TPCN-5 (1.75 V vs. NHE) is lower than E^θ (OH^-/•OH) (2.40 V vs NHE), which makes the photogenerated holes (h⁺) has no ability to oxidize OH^- into •OH [27]. By contrast, the characteristic signal for DMPO-•O₂^- adducts is detected after the visible light irradiation, demonstrating that •O₂^- can be successfully generated by reducing the adsorbed O₂ under the effect of photogenerated electrons (e^-) [26]. Besides, the DMPO-•O₂^- signal intensity of three TPCN samples is stronger than that of BCN (Fig. S11 in Supplementary Material), suggesting that the e^- concentration in the localized π-conjugated structure of TPCN is higher than that of BCN under visible light [46]. Thus, larger amount of •O₂^- can be produced in TPCN system. Furthermore, trapping experiments were carried out to evaluate the contribution of different reactive species to the photocatalytic activities of TPCN. Herein, t-BuOH, formic acid, and N₂ were employed as the scavenger of •OH, h⁺, and •O₂^-, respectively [39]. As observed in Fig. 12b, the photodegradation activity of TPCN-5 changes slightly after the addition of t-BuOH but decreases significantly in the presence of formic acid and N₂, suggesting that h⁺ and •O₂^- should be the dominant reactive species in the photocatalytic process of TPCN-5.

3.5.3. Charge transfer and photocatalytic processes

Based on the above experimental results, the charge transfer and photocatalytic processes in TPCN system under visible light irradiation was illustrated in Scheme 2. Compared with BCN, TPCN exhibits dramatically enhanced visible-light-induced photocatalytic disinfection and degradation performances, which could be attributed to several reasons as follows. Firstly, the huge BET surface area of TPCN can provide abundant exposed active sites and excellent adsorptivity to speed up the photocatalytic reaction rate. Secondly, the tubular structure and V_N can improve the light-harvesting ability in longer wavelength region (λ > 450 nm) of TPCN, which would lead to more h⁺ and e^- participating in the photocatalytic reaction. Thirdly, the hierarchical structure of TPCN is conducive to charge transfer. The hexagonal microtubes can induce fast and long-distance charge transport, and well-developed nanoporosity can promote the charge carriers to migrate from the birthplace to surface. Fourthly, appropriate amount of nitrogen defects within the TPCN framework is beneficial to charge transfer. The existence of V_N can act as the shallow charge traps to prompt charge separation, and more NH_x groups can lead to the hydrogen-bonding interaction generated between C₃N₄ layers to facilitate the interlayer charge transport. Finally, since more charge carriers can survive after the separation process and then migrate to the surface, larger number of reactive species (h⁺ and •O₂^-) would be formed in TPCN system, which have strong oxidation ability to effectively inactivate E. coli cells and thoroughly mineralize organic pollutants.

In conclusion, TPCN with hierarchical structure and nitrogen defects was successfully prepared through a facile self-templating approach. The 1D hollow structure and well-developed porosity of TPCN photocatalyst could increase the specific surface area and exposed active sites, provide multiple light reflection and scattering channels, and facilitate the fast and long-distance charge migration. Additionally, the introduction of nitrogen defects in g-C₃N₄ skeleton improved the light harvesting (λ > 450 nm), tuned the electronic structure, and prompt the charge transfer of TPCN. Consequently, TPCN exhibited highly efficient visible-light photocatalytic disinfection and degradation activities. Furthermore, h⁺ and •O₂^- were the reactive species of TPCN photocatalyst. The combination of hierarchical structure designing and in-situ defect engineering provides a new insight of fabricating powerful g-C₃N₄ based photocatalyst for environmental remediation.

This work was financially supported by the National Natural Science Foundation of China (No. 21707052), the Jiangsu Agriculture Science and Technology Innovation Fund (No. CX(18)2025), the Fundamental Research Funds for the Central Universities (Nos. JUSRP11905 and JUSRP51714B), and the Key Research and Development Program of Jiangsu Province (No. BE2017623).

Supplementary material related to this article can be found, inthe online version, at doi:https://doi.org/10.1016/j.jmst.2020.02.024.