Visible-Light-Driven Photocatalytic Water Disinfection Toward Escherichia coli by Nanowired g-C3N4 Film

Graphitic carbon nitride (g-C3N4) as metal-free visible light photocatalyst has recently emerged as a promising candidate for water disinfection. Herein, a nanowire-rich superhydrophilic g-C3N4 film was prepared by a vapor-assisted confined deposition method. With a disinfection efficiency of over 99.99% in 4 h under visible light irradiation, this nanowire-rich g-C3N4 film was found to perform better than conventional g-C3N4 film. Control experiments showed that the disinfection performance of the g-C3N4 film reduced significantly after hydrophobic treatment. The potential disinfection mechanism was investigated through scavenger-quenching experiments, which indicate that H2O2 was the main active specie and played an important role in bacteria inactivation. Due to the metal-free composition and excellent performance, photocatalytic disinfection by nanowire-rich g-C3N4 film would be a promising and cost-effective way for safe drinking water production.


INTRODUCTION
The COVID-19 pandemic has demonstrated the critical importance of sanitation, hygiene and adequate access to clean water for preventing and containing diseases. Developing economical and effective water disinfection methods is of great significance for human health as well as alleviating clean water scarcity issues Zhang et al., 2018). Conventional water disinfection approaches, including chlorine-based chemical disinfectants, ozone, and ultraviolet (UV), are capable of eliminating a majority of pathogens in water (Xia et al., 2017). However, the disinfection by-products produced by chlorine and chlorine-based disinfectants may cause potential health problems Huang et al., 2018). Ozone and UV inactivation methods are much safer, but require more energy consumption. Recently, metal-based photocatalysts, such as titanium dioxide (Fernández-Ibáñez et al., 2015;Cavalcante et al., 2016) and zinc oxide (Masoumbaigi et al., 2014), have been extensively investigated as new disinfectants (Cavalcante et al., 2016;Zhang et al., 2018). However, most metal-containing photocatalysts are activated by UV light irradiation, which is only 4% of the solar spectrum reaching earth surface (McGuigan et al., 2012;Ng et al., 2020). Not to mention, secondary pollution may occur due to the inevitable release of metal ions during the disinfection process (Hayden et al., 2010;Bai et al., 2011;Wang et al., 2012;Valsami-Jones and Lynch, 2015;Liu et al., 2016a;Wang and Mi, 2017;Zeng et al., 2017). Hence, it is of great importance to seek a non-toxic, visiblelight-driven and cost-effective method for future water disinfection (Ng et al., 2020).
Due to the metal-free composition, low-cost, suitable bandgap and beneficial energy levels, graphitic carbon nitride (g-C 3 N 4 ) has recently attracted much attention as photocatalyst for water splitting, CO 2 reduction, pollutant degradation, organic syntheses, and bacterial disinfection under visible light irradiation in the past decade Cao et al., 2015;Liu et al., 2015;Liu et al., 2016b;Zhang et al., 2017;Jia et al., 2019;Zhang et al., 2019;Giusto et al., 2020b;Lim et al., 2020;Zhang et al., 2020a;Zhang et al., 2020b). Especially, for bacterial disinfection, most of the previously reported studies focused mainly on the bulk particulate material Ran et al., 2019;Gan et al., 2020;Ni et al., 2020;Zhao et al., 2021), which makes recycling difficult and would inevitably lead to secondary pollution of fine nanoparticles (Wang et al., 2019a). Recently, breakthroughs in fabrication methods of g-C 3 N 4 films including vapor deposition polymerization, thermal vapor deposition and chemical vapor deposition have led to innovative and inspiring applications in many fields (Thomas et al., 2008;Bian et al., 2015b;Xu et al., 2015a;Xu et al., 2015b;Peng et al., 2018a). It is promising to further improve the performance and stability of g-C 3 N 4 materials for water disinfection by employing thin films instead of the common bulk materials (Thomas et al., 2008;Wang et al., 2019c;Chen et al., 2020;Jia et al., 2020;Villalobos et al., 2020). However, the nanoscale topography on such film is usually absent, which would enhance the capture of bacteria from the solution and thereby the disinfection efficiency (Arazoe et al., 2016;Peng et al., 2018b;Jia et al., 2019).
Herein, we report a simple route to synthesize the g-C 3 N 4 film featuring numerous nanowires on the top through a vaporassisted confined deposition onto the ground glass (Bian et al., 2015b;Jia et al., 2019). The adoption of high-roughness ground glass could prevent the easily peeling off of the g-C 3 N 4 film, which was usually the case in the smooth counterpart. By adjusting the precursor compositions, the g-C 3 N 4 films could be switched from flat surface to nanowired film with superhydrophilic property. Specifically, with the optimal precursor amount and the enhanced vapor pressure at elevated temperature, the formation of nanowire could be expected, which was speculated to be one dimensional growth on the convex bump of the on the flat g-C 3 N 4 film (Zhao et al., 2008;. The presence of nanowire on the top of the film could enhance the light harvesting and promote the adhesion of bacterial as well (Liu et al., 2009;Liu et al., 2010). Under visible light irradiation, the nanowire-rich g-C 3 N 4 film shows significantly better disinfection performances than the nanowire-free g-C 3 N 4 film. By employing the scavengerquenching experiments, a disinfection mechanism with H 2 O 2 as the reactive oxygen species for the g-C 3 N 4 film was proposed. A further hydrophobic modification of the pristine nanowire-rich g-C 3 N 4 film resulted in diminished disinfection performance, demonstrating the necessity of the beneficial interfacial property. The work reported herein provides a practical and efficient step forward for water disinfection exploiting hierarchical g-C 3 N 4 film materials.

EXPERIMENTAL SECTION
All chemicals from commercial sources were used as received without further purification, unless otherwise stated.
A simple vapor-assisted confined deposition method was performed to obtain the nanowire-rich and superhydrophilic g-C 3 N 4 films as previously reported (Jia et al., 2019). As shown in Supplementary Figure S1, 50 mg of the mixture of melamine and urea (mass ratio of 3:1) was uniformly placed into a ceramic crucible (60 ml) with the cleaned ground glass (2.5 cm × 4 cm) covering the top. Then the entire setup was sealed by aluminum foil. Then the sealed setup was placed in a tubular furnace and maintained at 550°C for 4 h in N 2 atmosphere with a heating rate of 2.3°C/min. The samples were let cool down naturally and collected at room temperature.
The photocatalytic disinfection performance of g-C 3 N 4 film was examined by the bacteria inactivation rate of E. coli as an indicator. E. coli stain was cultured in liquid medium for 18 h, and harvested via centrifuge at 8,000 rpm (5 min). The E. coli solution (2 ml, 1 × 10 4 colony forming unit (CFU) mL −1 ) was then added to a reactor containing a 10 cm 2 g-C 3 N 4 film. A 300 W xenon lamp equipped with a 420 nm cut-off filter (CEL-HXF300, Zhongjiao Jinyuan Technology Co., Ltd. China) was used as a visible light source. Circulating cooling water bath was used to stabilize the temperature of system (25°C). Each batch test lasted for 4 h with samples taken at a 1 h interval. All collected samples were resuspended in phosphate buffer solution and transferred to coated plate. The number of viable cells was determined by plate count method after incubation at 37°C for 18 h. For comparison, control experiments were also conducted in dark condition and in absence of g-C 3 N 4 . Superhydrophobic samples for control were prepared by modifying the surface of g-C 3 N 4 film using polytetrafluoroethylene. For each condition, the experiment was repeated three times. All glass apparatuses and photocatalysts used in the experiments were autoclaved at 121°C for 30 min prior to use and all experiments were operated under sterile conditions.
In order to provide insights on the disinfection mechanism of g-C 3 N 4 film, several functional scavengers were added to the system to remove selectively the different active species and depict a preferential disinfection mechanism (Wang et al., 2019b). L-histidine (0.5 mM) was employed for photogenerated singlet oxygen ( 1 O 2 ) removal, catalase (200 U/mL) for H 2 O 2 elimination, sodium oxalate (0.5 mM) for photogenerated holes removal, isopropanol (2.5 mM) for photogenerated OH removal, Cr (VI) (2 mM) for quenching photogenerated electrons, and 4-hydroxy-2,2,6,6-tetramethylpiperidinyloxy (TEMPOL) (1 mM) to neutralize photogenerated superoxide radicals (·O 2 − ). The disinfection performance of was tested via cell density quantification. For each condition, the experiment was repeated three times.
The surface morphologies of the as-prepared g-C 3 N 4 films were characterized by field-emission scanning electron microscopy (SEM, JSM-7500F). The crystal structure of samples was characterized by X-ray diffraction (XRD) with Cu Kα source radiation (D-MAX 2500/PC diffractometer) in the range from 10 to 50°, with scanning step of 0.02°. The absorption spectra of samples were recorded by UV-vis diffuse reflectance spectra (Hitachi U-3900). Contact angle measurements were performed with a Surface Analyzer (Lauda LSA 60). Visible light irradiation using a 300 W Xenon lamp (CEAULIGHT, Beijing). Figures 1A,B show the typical SEM images of the obtained g-C 3 N 4 films. The nanowires were randomly distributed on the top of the underlayer of the film with the length range about few micrometers. In a typical synthesis of such nanowire-rich g-C 3 N 4 film on the ground glass substrate, the mixture of urea and melamine was employed as precursor for thermal polycondensation in a vapor-assisted confined deposition setup. For comparison, either melamine or urea was employed as precursor alone for synthesizing the g-C 3 N 4 film and the optical images were shown in Supplementary Figure  S2. A homogeneous flat film without nanowire formation could be obtained when using either melamine or urea (see SEM images in Supplementary Figure S3). This morphology variation could be partly attributed to the formation of a hydrogen-bonded network between melamine and urea during the film growth. Thus, such nanowired thin films could be easily obtained by thermal condensation of melamine and urea mixture in a confined deposition setup (Thomas et al., 2008;Xiong et al., 2020). It's also worth to mention the current method is very economical in terms of the consumed precursor amount, which required only 50 mg of nitrogen-rich precursor to prepare the g-C 3 N 4 films over an area of 12.5 m 2 (2.5 cm × 5 cm) (Bian et al., 2015a;Lv et al., 2017;Sheng et al., 2017;Peng et al., 2018b;Lin et al., 2019;Xiao et al., 2019;Giusto et al., 2020b). Previously, only few works report on the wettability of g-C 3 N 4 film, which is an important parameter for water disinfection (Liang et al., 2015;Lin et al., 2019;Wu et al., 2019;Giusto et al., 2020c). As displayed in Figure 1C, the water wettability on such film was investigated by contact angel measurements. The asprepared g-C 3 N 4 films are proven to be superhydrophilic. When a water droplet (2 µL) was deposited on the film, it quickly spread on the surface. Finally, the droplet spread out completely and the contact angle decreased to 0°within 2 s. So, the g-C 3 N 4 film prepared by this method exhibits superhydrophilicity, which is conducive to improving disinfection performance. As shown, the as-prepared films present the typical bright yellow color of g-C 3 N 4 , with an intense blue homogeneous fluorescence when irradiated with UV light, confirming that the deposition occurred homogeneously over the whole substrate surface ( Figure 1D). The optical properties of the as-prepared g-C 3 N 4 films were further studied by photoluminescence (PL) at an excitation wavelength of 365 nm (Supplementary Figure S4). In addition to the film prepared by urea, the other g-C 3 N 4 films showed a strong PL emission centered at about 466 nm, which is corresponding to the absorption edges in UVvis diffuse reflectance spectra. And the emission intensity of g-C 3 N 4 film prepared by the mixture of melamine and urea is stronger than the g-C 3 N 4 film prepared by melamine alone, indicating the low charge separation efficiency of photogenerated electron-hole pairs of the former.

RESULTS AND DISCUSSION
The UV-Vis diffuse reflectance absorbance spectra of the bulk g-C 3 N 4 and g-C 3 N 4 film are shown in Figure 2A. The bulk and the nanowired film show similar optical properties with their absorption edges locating at 471 and 456 nm, respectively. Correspondingly, their optical band gaps are 2.52 and 2.58 eV as identified by the Kubelka-Munk method ( Figure 2B), which is in good agreement with previous reports (Giusto et al., 2020a). At the same time, we measured the transmittance of these three kinds of g-C 3 N 4 films, and found that the g-C 3 N 4 film prepared by the mixture of melamine and urea is more opaque, which will be more conducive to light absorption (Supplementary Figure S5).
The crystallinity of the g-C 3 N 4 film and powder were analyzed by XRD and the corresponding results are shown in Figure 2C. The diffraction peaks of bulk powder with strong intensity located at 27.5°and 13.1°could be ascribed to the typical graphitic interlayer (002) plane and the in-plane structural repeating unit of the aromatic systems, respectively (Thomas et al., 2008). The g-C 3 N 4 film shows a peak at 27°while the peak of (100) plan disappeared. Due to the enhanced roughness of the ground glass, the obtained film could not be easily peeled off from the substrate. So we have to prepare a lot of g-C 3 N 4 film to scrape the powder for the measure. For comparison purpose, nearly all the g-C 3 N 4 film grown on the smooth glass surface could be easily peeled off from the substrate and could get wrinkled and crushed to particles in a sustained reaction under aqueous environment (Arazoe et al., 2016;Xiao et al., 2018;Jia et al., 2019). FIGURE 2 | (A) UV-Vis absorption spectra of blank glass (black line), bulk g-C 3 N 4 (red line) and g-C 3 N 4 film (blue line), respectively. (B) Tauc plots of blank glass (black line), bulk g-C 3 N 4 (red line) and g-C 3 N 4 film (blue line), respectively. (C) XRD diffraction patterns of bulk g-C 3 N 4 (black line) and g-C 3 N 4 film (red line). (D) Infrared spectra of bulk g-C 3 N 4 (black line) and g-C 3 N 4 film (red line). (E) Typical N 1s and (F) C 1s XPS spectra of g-C 3 N 4 film.
Frontiers in Nanotechnology | www.frontiersin.org May 2021 | Volume 3 | Article 684788 The typical features of g-C 3 N 4 materials can be depicted by means of Fourier transform infrared spectroscopy (FTIR) ( Figure 2D). The peak at 807 cm −1 corresponds to the out-of plane ring bending of the triazine unit with the characteristic stretching modes of the CN heterocycles was observed ranging from 1,040 to 1,640 cm −1 . And the broad absorption peak from 2,900 to 3,600 cm −1 is attributed to the stretching vibration of terminal -NH 2 group.
XPS spectra of the film give more insights on its chemical features ( Figure 2E). The N 1s spectrum can be deconvoluted into four peaks corresponding to nitrogen atoms in different functional groups: sp  hybridized aromatic N bonded to carbon atoms of the s-triazine ring at 398.4 eV, tertiary N bonded to carbon atoms in the form of N-(C) 3 at 400.1 eV, amino functional groups with hydrogen (C-N-H) at 400.9 eV, and π-excitations at 404.2 eV. Three fitting peaks located at 284.4, 286.2, and 287.9 eV in Figure 2F corresponded to the graphitic carbon (C-C), the sp )-type C N bonds and sp  hybridized carbon in tri-s-triazine rings (N 2 -C N), respectively. The XPS spectra demonstrate the presence of the tri-s-triazine units, which is in good agreement with FTIR spectra.
In this work, E. coli, one of the most commonly available bacteria, was selected as a model microbe to evaluate the deactivation performances. The photocatalytic disinfection performances can be clearly seen even by naked eyes in the digital images from the agar plates containing the bacterial culture of E. coli collected at 0-, 2-and 4-hours steps ( Figures  3A,I, ii,iii are 0, 2 and 4 h respectively), while no difference can be seen in absence of the photocatalyst or visible light ( Figure 3A iv-ix). The high disinfection efficiency and the synergistic effect of g-C 3 N 4 and visible light is confirmed with the residual bacteria counting ( Figure 3B): indeed, most of the E. coli was inactivated already in 2 h, and almost complete inactivation (>99.99%) was reached after 4 h. Negligible variation of the bacteria activities was obtained in the reference cases (<1%), using only the g-C 3 N 4 film in dark ( Figure 3B, blue line) or in absence of the photocatalyst ( Figure 3B, red line). The obtained results confirms unequivocally that high disinfection performances are obtained only by the synergistic effect of g-C 3 N 4 and visible light. It is worth mentioning that the g-C 3 N 4 film is still strongly bonded to the substrate after the sterilization experiment, which provides the possibility for the subsequent repeated experiments (Supplementary Figure S6). The stability of g-C 3 N 4 membrane was investigated by repeating E. coli inactivation experiments with recycled g-C 3 N 4 membrane. After three times' recycling, disinfection performance of g-C 3 N 4 film has not decreased significantly (Supplementary Figure S7).
We further investigated the effect of the precursor mass on the disinfection performances. As shown in the following ( Figure 4A), the optimum of disinfection was achieved by using 50 mg of precursor, with respect to 30 and 100 mg. It is worth noting that by using 50 mg precursor, the resulting g-C 3 N 4 film showed a higher density of nanowires per unit area, which resulted in a better disinfection activity compared to the other two cases, as confirmed by SEM (Supplementary Figure S8). Therefore, we attributed the enhanced disinfection ability to the nanowired structure which allows for an intimate contact between the g-C 3 N 4 film and the bacterial disinfection with efficient separation of electrons and holes. The influence of nanowires on the disinfection performance was investigated in an experiment by using different precursor. The results showed that the g-C 3 N 4 film with nanowired structure has a better disinfection performance (Supplementary Figure S9). Since the E. coli concentration after 4 h was almost four order of magnitude lower, we further investigated on the performances of the film prepared from 50 mg precursor.
To further prove that the surface wettability influences the effective contact between the photocatalyst and bacteria, the hydrophilic g-C 3 N 4 film was treated with  Figure S10) and study how that influences the performances (Sheng et al., 2017). The effect of hydrophobic treatment on photocatalytic disinfection was investigated over time and compared to the bare sample ( Figure 4B), confirming that the superhydrophilic properties are of paramount importance for improving the disinfection process. Superhydrophilicity guarantees the close contact between bacteria and photocatalyst, which leads to the quite high catalytic activity. During the photocatalytic disinfection, active species, such as H 2 O 2 , e − , · OH, · O 2 − , 1 O 2 and h + , are directly involved in the disinfection step, as shown in Scheme 1 (Liang et al., 2015). In order to study the mechanism of g-C 3 N 4 film, we performed further tests using selective scavengers. Several scavengers were added to the solution containing E. coli to eliminate the corresponding active species, and the photocatalytic reaction without scavenger was used as the control group ( Figure 5).
The addition of L-histidine exhibited negligible influence on the disinfection performance ( Figure 5A). Addition of catalase significantly reduced the photocatalytic disinfection efficiency, suggesting that H 2 O 2 play an important role in bacterial inactivation ( Figure 5B). The photo-generated holes were much less likely to come into direct contact with bacteria. Moreover, h + is more difficult to diffuse than H 2 O 2 , and the photo-generated lifetime is also extremely short. Therefore, H 2 O 2 served as the disinfectant has a more important role than h + ( Figure 5C). The valence band position of g-C 3 N 4 is more negative than the redox potential of OH/OH − , it is difficult for OH − to be oxidized to generate OH. Thus, the role of OH in the photocatalytic disinfection is only minor ( Figure 5D). As reported, ·O 2 − and e − were also critical active species in the photocatalytic disinfection process Liang et al., 2015), because the photocatalytic disinfection could occur through oxidation by active oxidants and reduction by electrons ( Figure 5E). Indeed, the addition of TEMPOL reduces significantly the photocatalytic efficiency ( Figure 5F). As the superoxide anion (·O 2 − ) is an active intermediate in the production of H 2 O 2 , the more O 2 − is produced, the more H 2 O 2 will be generated. Based on these results we can then confirm that the disinfection mechanism occurs via the generation of superoxide and water peroxide, and thus is related to the strong oxidation properties of g-C 3 N 4 (Mazzanti and Savateev, 2020).

CONCLUSION
In this work, a superhydrophilic nanowire-rich g-C 3 N 4 film was synthesized and applied for disinfection to remove E. coli under visible light irradiation. The disinfection efficiency could reach over 99.99% in 4 h under visible light irradiation when the mass of the precursor was 50 mg. Surface wettability investigation demonstrated that the superhydrophilic property enhances the photocatalytic disinfection performance greatly, while hydrophobic treatments to the surface lead to an increase of the contact angle and a decrease of the disinfection. Mechanism for disinfection was further explored with H 2 O 2 being the crucial active specie. Furthermore, the convenient use of the g-C 3 N 4 as a surface-immobilized film allows FIGURE 4 | (A) Effect of precursor mass on disinfection performance of g-C 3 N 4 film. (B) Effect of wettability on photocatalytic disinfection. SCHEME 1 | The proposed disinfection mechanism of nanowire-rich g-C 3 N 4 film.
Frontiers in Nanotechnology | www.frontiersin.org May 2021 | Volume 3 | Article 684788 for simple and complete removal of the photocatalyst after the disinfection process. We expect therefore that these results will encourage further investigation and development of metal-free carbon materials for water disinfection applications.

DATA AVAILABILITY STATEMENT
The original contributions presented in the study are included in the article/Supplementary Material, further inquiries can be directed to the corresponding authors.

AUTHOR CONTRIBUTIONS
JL conceived the idea. YZ and SS performed the experiments and drafted the manuscript. YZ, XZ, PG, XH, contributed to the discussion of the experimental data and the writing of the manuscript. PG, XH, JL contributed to the writing, reviewing and editing of the manuscript. JL contributed to the funding acquisition and supervised the project.

FUNDING
This work was financially supported by the Natural Science Foundation of Shandong Province (ZR2018MB018).