Effect of Contact Mode of TiO₂/g-C₃N₄ Heterojunction on Photocatalytic Performance for Dye Degradation

Abstract

Two types of TiO₂/g-C₃N₄ heterojunctions, physically and chemically contacted samples, were synthesized to investigate the effect of their contact modes on the photocatalytic activity for dye degradation. The physically contacted TiO₂/g-C₃N₄ heterojunction (TCN_PHY) was prepared by an electrostatic assembly process. In addition, the chemically connected heterojunction (TCN_CHE) was synthesized by a hydrothermal method accompanied by the formation of Ti–O–C covalent bonds. Fourier-transform infrared and X-ray photoelectron spectroscopies confirmed the formation of Ti–O–C covalent bonds in TCN_CHE. Subsequently, their photocatalytic activity for dye degradation was evaluated as a model reaction. The results showed that TCN_CHE exhibited higher degradation efficiency than TCN_PHY because of its higher UV light absorbance and lower recombination rate than those of the physically contacted sample. These results indicate that the hydrothermal method gives unique advantages in chemically contacted heterojunction construction, which can lead to the improvement of photocatalytic activity.

1. Introduction

Recently, world energy demand has been largely dependent on fossil fuels such as petroleum, coal, and natural gas, which are rapidly being depleted. Therefore, novel discoveries in materials science and engineering have been desired to overcome the obstacles to effective energy conversion and environmental protection. Among various renewable energy projects, semiconductor-based photocatalysts that can utilize inexhaustible and clean solar energy have received much attention as a promising strategy. In general, heterojunctions constructed from two different semiconductors provide improvement in photocatalytic performance owing to raising the carrier separation efficiency, widening the range of absorption wavelength and tuning the oxidation-reduction properties.^1,2)

Graphitic carbon nitride (g-C₃N₄), an n-type semiconductor of great thermal/chemical stability with a facile synthesis procedure, has attracted wide attention as a photocatalyst.^3–5) g-C₃N₄ is used as a photocatalyst,⁶⁾ which is often combined with other semiconductors such as metal oxides^7–9) and halides^10–12) to construct heterojunctions. Staggered gap catalysts have been actively studied by many research groups. Among them, TiO₂/g-C₃N₄ heterojunctions (TCN) are among of the most popular systems due to their chemical stability, non-toxicity and low cost.^3,13–15) Photocatalytic processes can be explained by different reaction mechanisms according to the transfer routes of carriers (electrons and holes), such as type II,^7,16) Z-scheme^14,17,18) and S-scheme.^19,20) In the case of the type II mechanism, carriers transfer occurs in opposite directions; for the Z-scheme^21–23) and S-scheme,²⁴⁾ carriers are partially recombined at the interface and the carriers with high redox ability can be left for the redox reaction. Thanks to these investigations, heterojunction construction and the reaction mechanisms are well understood.

From an overview of these valuable studies, a systematic comparison of the factors influencing the photocatalytic activity of TiO₂/g-C₃N₄ heterojunctions has become imperative. Although efforts have been made to illustrate the effect of contact type and morphology on catalytic performance,^25,26) detailed investigation is still needed to clarify the effects on the photocatalytic performance and reaction mechanism. It would be helpful to further improve the strategies for heterojunction construction. In this study, two different contact modes, namely physical and chemical connection, are regarded as the two main ways to construct TiO₂/g-C₃N₄ heterojunctions. Especially, g-C₃N₄ nanosheets and TiO₂ nanoparticles with high specific surface area were selected as primary materials. Specifically, g-C₃N₄ nanosheets were prepared by exfoliation of bulk g-C₃N₄, and TiO₂ nanoparticles (NPs) were synthesized by a hydrothermal method. Due to their opposite surface zeta potentials (§) at pH = 7.56, physically contacted heterojunctions (TCN_PHY) were prepared by electrostatic attraction. Using the hydrothermal method, which provided us an effective way to combine 0D TiO₂ NPs and 2D g-C₃N₄ nanosheets together, chemically connected TCN heterojunctions (TCN_CHE) were also synthesized. Relevant characterization methods such as Fourier transform infrared (FT-IR) spectroscopy and X-ray photoelectron spectroscopy (XPS) results revealed the differences in their surface chemical structures. Dye degradation is a well-known model reaction to investigate the potential of materials as photocatalysts. Therefore, the dye degradation activity of the two heterojunction catalysts was evaluated to understand the effect of the contact mode of heterojunctions on photocatalysis, and the results illustrated their differences in photocatalytic activity.

2. Experimental Procedure

2.1 Materials

All chemicals, including melamine (C₃H₆N₆), titanium isopropoxide (Ti(OCH(CH₃)₂)₄, TIPO), ethylene glycol ((CH₂OH)₂, EG), 3,4-dihydro-2,2-dimethyl-2H-pyrrole 1-oxide agent (C₆H₁₁NO, DMPO) and methylene blue (C₁₆H₁₈N₃SCl·3H₂O, 98.5%, MB) were purchased from FUJIFILM Wako Pure Chemical Corporation. Deionized water (DW) was used throughout this study.

2.2 Synthesis of photocatalysts: g-C₃N₄ nanosheets and TiO₂ NPs

Firstly, bulk g-C₃N₄ was prepared by a pyrolysis process. Specifically, 3.0 g of melamine was calcined in a sealed crucible at 520 and 550°C for 2 h for each step in an N₂ atmosphere with ramp rate of 5°C/min for heating process. Yellow g-C₃N₄ bulk was thus achieved (Fig. A1(a)). After this process, the bulk sample was ground to form a powder sample for the following exfoliation. Then 300 mg of g-C₃N₄ powder was added into 20 ml of 6 M HNO₃ aqueous solution and refluxed at 138°C for 12 h. After that, the obtained product was washed with distilled water three times and filtered to obtain the solid fraction. The solid sample was then dried at 60°C overnight. Finally, sonication treatment was conducted for 8 h to exfoliate the bulk g-C₃N₄. The obtained light-yellow g-C₃N₄ nanosheet was filtered and dried for heterojunction preparation.

TiO₂ nanoparticles were synthesized via a hydrothermal process.²⁷⁾ In detail, 0.1 ml of TIPO was added dropwise into 10 ml of EG in an Ar atmosphere. After stirring for 30 min, 10 ml solution was mixed with the same volume of distilled water. The mixture was then moved into an autoclave and heated at 180°C for 12 h. Finally, the obtained white product was purified and filtered, and washed with ethanol and water three times. The size of the synthesized TiO₂ nanoparticles was about 7–8 nm (Fig. A1(b)).

2.3 Preparation of TCN_PHY and TCN_CHE heterojunctions

6.0 mg of TiO₂ NPs and 14.0 mg of g-C₃N₄ nanosheets were first dispersed in 20 ml of water with a weight ratio of 3:7, denoted as the TCN_mix dispersion (Fig. A2(a)). Then, the pH of the TCN_mix dispersion was tuned to 7.56 by adding 0.01 M NaOH and 0.01 M HClO₄ aqueous solutions. At this pH, the TiO₂ NPs and g-C₃N₄ nanosheets in the dispersion were positively and negatively charged, respectively. Under the effect of electrostatic attraction, the two materials were physically adsorbed together and formed a physically contacted heterojunction. Thus, the prepared sample was designated as TCN_PHY.

The chemically contacted heterojunction was constructed by a hydrothermal method. First, 100 mg of g-C₃N₄ nanosheets was dispersed in water (20 ml) and pretreated at 200°C for 6 h to form surface OH groups.²⁸⁾ Then, 0.1 ml of TIPO was dissolved in 10 ml of EG. The resulting solution was then added to 20 ml of the above-mentioned g-C₃N₄ suspension dropwise under stirring for 1 h. Next, the final suspension was transferred into a Teflon-lined autoclave and aged at 180°C for 12 h. Finally, the obtained product was centrifuged and washed with ethanol and distilled water three times and is hereinafter referred to as TCN_CHE.

2.4 Characterization

Structural analysis was performed by X-ray diffraction (XRD, Rigaku Ultima IV, Cu Ka irradiation, 40 kV, 40 mA). The surface morphologies were characterized by transmission electron microscopy (TEM, HITACHI H-7650). FT-IR spectra were recorded using an IRAffinity-1S system (SHIMADZU). The surface zeta potential of TiO₂ NPs and g-C₃N₄ nanosheet was measured by a zeta-potential analyzer (Otsuka, ELSZ-2000). The chemical state of the samples was investigated by XPS (SHIMADZU ESCA-3400). The C 1s peak (284.8 eV) of the carbon tape was used for calibration of the binding energy of each element. UV-vis and photoluminescence (PL) spectra were obtained using HITACHI U-4100 and U-3900H spectrometers, respectively. To fairly compare light absorption and recombination property of sample, the concentration of the dispersion was set as 1 mg/10 ml for UV-vis and PL measurements; the excitation wavelength for PL measurements was set as 300 nm. The TiO₂:g-C₃N₄ weight ratio for both samples was estimated by thermogravimetric analysis (TGA, Rigaku Thermo Plus EVO 2) over the range of 25–800°C. To confirm the photocatalytic mechanism, the h⁺ transport behavior was investigated by electron paramagnetic resonance spectroscopy (EPR, Bruker BioSpin E580) with a modulation frequency of 100 kHz. In detail, 2.0 mg of the TCN heterojunction samples was dispersed in 20 ml of distilled water with the addition of 1.0 mg of DMPO, and EPR spectra were measured after UV light (λ = 254 nm) irradiation for 5 min.

2.5 Evaluation of photocatalytic activity

The photocatalytic activity of each sample for dye degradation was evaluated under UV light irradiation (ASONE, SLUV-8. λ = 254 nm). Firstly, 10 mg of catalyst was dispersed in 400 ml of MB (10 mg/L) solution via sonication for 5 min. The entire catalytic reaction system (Fig. A3), including the ultraviolet water sterilizer and the water recirculation system, was wrapped with aluminum foil to avoid natural light irradiation. Before the experiment, the dispersion was stirred for 30 min to obtain absorption-desorption equilibrium. At fixed time intervals, 2 ml of heterojunction samples were taken from the solutions to measure the dye concentration by UV-vis spectroscopy. Degradation experiments using the primary g-C₃N₄ and TiO₂ materials were also performed to compare with the heterojunction groups (Fig. A2(b)). The concentration of MB was determined based on the intensity of the absorption at 663 nm (Fig. A4(a)). By referring to a standard UV absorption curve series of different concentrations (1–10 mg/l), the concentration of MB in the solution before and after the reaction was estimated (Fig. A4(b)).

3. Results and Discussions

3.1 Preparation of TiO₂ NPs and g-C₃N₄ nanosheets

In this work, the conventional hydrothermal method was utilized to prepare the TiO₂ sample. Figure 1(a) shows an XRD pattern of the as-prepared TiO₂ NPs. Diffraction peaks observed at 25.3°, 37.7° and 48.0° were assigned to the (101), (004) and (200) planes of the anatase phase (ICSD: 71-1167), respectively. Nanoparticles with a size of 7–8 nm were observed in the TEM image (Fig. 1(b)). These results indicate that TiO₂ NPs with the anatase phase were successfully synthesized.

Fig. 1

Synthesized pristine TiO₂ nanoparticles: (a) XRD pattern and (b) TEM image. Prepared g-C₃N₄ nanosheet: (c) XRD pattern and (d) TEM image.

In the XRD pattern for the pristine g-C₃N₄ bulk sample (Fig. 1(c)), two characteristic peaks were observed at 13.2° and 26.5°, which can be assigned to tri-s-triazine periodic stacking (100) and graphitic layers (002), respectively.^29,30) After the sonication process, the bulk g-C₃N₄ was exfoliated to form nanosheets, as shown in Fig. 1(d). The peak shift of the (002) reflection implied a change in the interlayer distance, and the disappearance of the (100) reflection indicates a decrease in the size and order of the layers.³¹⁾ The TEM image showed a sheet-like morphology with an average lateral size of more than 400 nm, which is much larger than the size of the TiO₂ NPs.

3.2 Construction of heterojunctions

The obtained surface zeta potentials for TiO₂ and g-C₃N₄ (Fig. 2(a)) reflect their different surface charge properties. At pH 7.56, the isoelectric point for g-C₃N₄ nanosheets was about +21.4 mV, which means that the outer surface is electropositive. In contrast, the zeta potential for TiO₂ NPs showed evidence of a negatively charged surface (ca. −30 mV). Owing to the electrostatic attractive force between TiO₂ NPs and g-C₃N₄ nanosheets at this pH, a physically contacted TCN heterojunction was prepared, where uniformly supported TiO₂ NPs were observed in Fig. 2(b). In previous studies, the hydrothermal method was applied to prepare g-C₃N₄-metal oxide heterojunctions.^32–36) As reported, a covalently bonded interface was formed during the hydrothermal process.³⁷⁾ The water processing strategy causes the formation of surface hydroxyl groups on the g-C₃N₄ nanosheets that tend to react with Ti(OH)₄ molecules and form Ti–O–C covalent bonds at the heterojunction interface.²⁸⁾ The following Fig. 3 was proposed as the route for covalent bond formation. Herein, water plays an important role in forming the heterojunction. During the pretreatment process for g-C₃N₄, the terminal amino groups are converted to hydroxyl groups and unsaturated C=O bonds.^28,38) In addition, the hydrolysis process for TIPO generates Ti(OH)₄ as an intermediate. It can react with the surface C–OH and unsaturated C=O bonds on g-C₃N₄, resulting in the formation of Ti–O–C bonds.^32,39,40) After the following hydrothermal process, a covalently bonded heterojunction is formed by dehydration condensation. TEM images (Figs. 2(c) and (d)) reflected the morphology of TCN_CHE, where the g-C₃N₄ nanosheets became thinner after the hydrothermal process, and some holes were observed in the surface of the g-C₃N₄ nanosheets. This phenomenon was probably due to the high temperature and pressure of the hydrothermal treatment, which was accompanied by the generation of hydroxyl radicals.⁴¹⁾ The size of TiO₂ particles in the prepared TCN_CHE sample was about 8–9 nm (Fig. 2(d)), which is nearly the same as that for pristine TiO₂ NPs. Thus, the effect of particle size on the catalytic activity can be considered negligible.

Fig. 2

(a) Zeta potentials for TiO₂ NPs and g-C₃N₄ nanosheets. (b) TEM image of TCN_PHY. (c) TEM image (d. magnified image) of TCN_CHE.

Fig. 3

Synthetic route for chemically contacted TCN heterojunction.

3.3 Structure and chemical composition

XRD patterns for TCN_PHY and TCN_CHE are shown in Fig. 4(a). In the case of TCN_PHY, both TiO₂ and g-C₃N₄ pristine samples basically retained their primary structure after the assembly process. Specifically, the characteristic peaks for TCN_PHY were the same as those for the pristine g-C₃N₄ and TiO₂, which excludes the possibility of structural effects on photocatalytic activity. FT-IR spectra were measured to investigate the presence of functional groups on each sample. The peaks observed at 1600–1200 cm⁻¹ are attributed to in-plane vibrations of tri-s-triazine species,^42,43) which were observed in both TCN_PHY and TCN_CHE samples. The peak detected at 810 cm⁻¹ represents the out-of-plane C–N bending vibration of triazine in the g-C₃N₄ framework. This peak became less prominent for TCN_CHE, which was due to the out-of-plane chemical structure being partly destroyed by the hydrothermal process, as observed in other works.³⁸⁾ Moreover, TCN_CHE showed a distinct peak at 1084 cm⁻¹. This peak is regarded as the C–O stretching vibration on TiO₂ (Fig. 4(b)) and was not observed for TCN_PHY. These results indicate that the whole spectrum of TCN_PHY can be basically regarded as a combination of the primary TiO₂ and g-C₃N₄ spectra, which is essentially different from the spectrum of TCN_CHE.

Fig. 4

(a) XRD patterns and (b) FT-IR spectra for TCN_PHY and TCN_CHE.

XPS is an effective approach for investigating the chemical state of surface elements. As shown in Fig. 5(a), all the elements, C, N, Ti and O, were detected in the survey spectra at 285, 399, 454 and 531 eV, respectively. In Fig. 5(b), two prominent C 1s peaks were observed at 288.1 and 284.8 eV for the pristine g-C₃N₄ nanosheets, which are attributed to the (N)₂–C=N group and abundant C–OH groups on the surface generated by exfoliation using nitric acid. For the C 1s XPS spectrum of TCN_PHY, the peaks detected at 284.8, 286.4 and 288.1 eV can be assigned to the hydroxyl groups on g-C₃N₄, N=CH–N and (N)₂–C=N in the g-C₃N₄ components, respectively.^34,44,45) These peaks were also observed for TCN_CHE. On the other hand, the intensity of the peak detected at around 286.3 eV in TCN_CHE was higher than that for TCN_PHY. If Ti–O–C bonds exist in TCN_CHE, the peak should be observed at 286.3 eV.³²⁾ This is close to the peak energy for N=CH–N (286.4 eV). Thus, the peak could be recognized as the superposition of N=CH–N and Ti–O–C, which causes higher intensity for the peak at 286.3 eV in TCN_CHE, suggesting the presence of Ti–O–C bonds. Figures 5(c) and 5(d) show the N 1s and Ti 2p XPS spectra of both heterojunctions. In Fig. 5(c), the N 1s spectra of two heterojunctions are almost the same—the peaks observed at 398.6 and 400.0 eV correspond to C=N–C groups of triazine rings and N–(C)₃.⁴⁶⁾ No extra N 1s peak was found for either of the samples, indicating that neither Ti–N nor Ti–C bonds were formed in TCN_CHE or TCN_PHY. For the Ti 2p spectrum of TCN_PHY, the peaks observed at 459.1 and 464.9 eV are attributed to Ti 2p_1/2 and Ti 2p_2/3 in TiO₂.⁴⁷⁾ For both heterojunction samples, the entire spectra showed no peak shifts or the formation of any Ti-related bond. This result implied that the new Ti-linked chemical bond was not generated during the construction of the heterojunction. Herein, the main difference is the change in the peak distribution, which is evidence for the chemical state change for TCN_CHE. Therefore, it could be deduced that the chemically connected heterojunction was generated by the formation of C–O–Ti covalent bonds.

Fig. 5

XPS spectra of TCN_PHY and TCN_CHE. (a) Survey. (b) C 1s. (c) N 1s and (d) Ti 2p.

Surface area is a critical factor affecting photocatalytic activity. Therefore, the BET surface area and pore size were estimated by a N₂ physisorption experiment. Figure A5 shows N₂ adsorption-desorption isotherms for TCN_PHY and TCN_CHE, where type IV isotherms were observed for both samples. This result suggests the presence of mesopores in the g-C₃N₄-based samples. During the thermal polymerization process of melamine, defect sites were formed as pores in the g-C₃N₄ bulk. The calculated BET surface areas and pore volumes are summarized in Table 1. The results show that the surface area of the g-C₃N₄ nanosheet changed from the 9.1 to 98.5 m²/g after the exfoliation process, accompanied by smaller pore size and larger pore volume compared with the bulk sample. After the construction of heterojunctions with TiO₂ NPs, the surface area decreased to 68.48 and 30.64 m²/g for TCN_PHY and TCN_CHE, respectively, and the pore size and volume were also decreased to some extent. It should be noted that the pore size for the primary g-C₃N₄ nanosheet is larger than that for TiO₂ NPs; thus, the smaller pore size and lower pore volume could be attributed to TiO₂ NPs filling the pore sites of g-C₃N₄.

Table 1 Porosities of primary g-C₃N₄ and g-C₃N₄ nanosheets and TCN heterojunctions.

3.4 Comparison of photocatalytic activity

To fairly compare the photocatalytic activities of the TCN heterojunction samples, the weight ratio of TiO₂ to g-C₃N₄ was set to 3:7 (Fig. A6). Figure 6(a) shows the MB degradation curve for the primary TCN_PHY, TCN_CHE and contrasting TCN_mix sample. At the beginning of the degradation process, TCN_PHY showed similar tendencies to TCN_mix. After 2.0 h, the degradation process for TCN_PHY began to accelerate and showed higher efficiency than the mixture group. Herein, the electrostatic attraction-based preparation was proved to be an effective method to improve the photocatalytic activity.^24,48) Moreover, the degradation results also implied that TCN_CHE, which could decompose over 60% MB pollutant within 1.5 h, showed the highest catalytic activity at the beginning. This result implied the covalent bonded interface could help to gain more effective charge carriers than the electrostatic force-based heterojunction.

Fig. 6

(a) MB degradation performance of TCN_mix, TCN_PHY and TCN_CHE. (b) UV-vis diffuse reflectance and (c) Photoluminescence spectra of TCN heterojunction samples.

The differences in these results could be ascribed to their different light absorbance and recombination rate. In the UV-vis diffuse reflectance (UV-Vis DRS) spectra (Fig. 6(b)), TCN_CHE showed higher absorbance than TCN_PHY at 254 nm. The PL spectra also implied lower PL intensity for TCN_CHE, which reflected the lower photo-generated carrier recombination rate in the covalent-bonded heterojunction (Fig. 6(c)). For these reasons, the preparation of TCN_CHE by the hydrothermal process was a favorable strategy. In terms of electronic mobility improvement in the covalently bonded structure, mechanisms such as an extended π-conjugated covalent organic framework^49,50) and orbital reconstruction of metal oxide/sulfide surfaces^51–55) were investigated. For g-C₃N₄ in the heterojunction sample, the π-conjugated structure could contribute unpaired electrons in the 2p_z orbital from N=C–N₂ and C–N=C bonds.⁵⁶⁾ The proportion of these bonds in TCN_CHE was higher than that in TCN_PHY, which probably implied the presence of more effective carriers in TCN_CHE. Also, for TiO₂ NPs, the hydrothermal method with the addition of g-C₃N₄ might also cause reconstruction of surface Ti–O–Ti covalent bonds. Whether it would contribute to higher carrier mobility or not needs more investigation.

In summary, construction of heterojunctions proved to be an effective strategy to increase photocatalytic activity; moreover, the covalently bonded TCN heterojunction showed further improvement in photocatalytic activity due to its higher light absorbance and lower carrier recombination rate. To explore the reaction mechanism of dye degradation over the TCN catalysts, the bandgap structures were investigated according to the primary TiO₂ and g-C₃N₄ bandgaps via UV-DRS spectra and Tauc plots. Figures 7(a) and 7(b) show UV-Vis DRS spectra of TiO₂ and g-C₃N₄, and the bandgap energies were calculated by the following formula:

  

\begin{equation} (\alpha h\nu)^{\frac{1}{2}} = A(h\upsilon - E_{g}) \end{equation} (1)

where α, h, ν and A are the absorption efficient, Planck constant, light frequency and a constant, respectively.

Fig. 7

UV-vis DRS spectra of (a) TiO₂ and (b) g-C₃N₄. Tauc plots for primary (c) TiO₂ and (d) g-C₃N₄. (e) Conceptual bandgap scheme for TiO₂ and g-C₃N₄.

The calculated band gaps for the synthesized g-C₃N₄ and TiO₂ were 2.52 and 3.23 eV, as shown in Figs. 7(c) and 7(d). The valence band (VB) and conduction band (CB) potentials were calculated with the formulas,^57,58)

  

\begin{equation} E_{\textit{CB}} = \chi - E_{C} - \frac{1}{2}E_{g} \end{equation} (2)

  

\begin{equation} E_{\textit{VB}} = E_{\textit{CB}} + E_{g} \end{equation} (3)

where χ is the Sanderson electronegativity of semiconductors, which is 4.73 and 5.81 eV for g-C₃N₄ and TiO₂.^13,59) E_c, the scaling factor related to the difference between AVS and NHE, is 4.5 eV.⁵⁴⁾ Therefore, the calculated E_CB/E_VB results for g-C₃N₄ and TiO₂ were −1.03 eV/1.49 eV and −0.29 eV/2.91 eV, respectively (Fig. 7(e)).

To illustrate the degradation mechanism of TCN heterojunctions, the investigation of hydroxyl radicals (OH·), the key species to decomposing the MB structure,⁶⁰⁾ was performed using EPR measurement. Both generated electrons and holes could further generate OH· in aqueous solution, which follows the decomposition reactions listed in Supporting information (eq. (A1)).⁶¹⁾ Herein, it should be noted that the OH·/OH⁻ redox potential (1.99 eV) is between the valence bands of g-C₃N₄ (1.49 eV) and TiO₂ (2.91 eV); therefore, a hydroxyl radical scavenger can help to detect for the existence of O·. Theoretically, only photo-generated holes in TiO₂ will take part in the hydroxyl radical redox reaction. By using DMPO as a radical scavenger, the lifetime-extended OH· was examined by EPR measurements of the dispersion. For example, the strong resonance signal of TiO₂ NPs was examined after UV light irradiation for 5 min, which was due to the highly oxidizing nature of h⁺ in the valence band (Fig. 8(a)). For g-C₃N₄, by contrast, only a weak DMPO-·OH signal could be observed, which can be explained by the weaker oxidation activity of h⁺ in its valence band. After formation of heterojunctions, both types exhibited ·OH radical signals as strong as that for the TiO₂ NPs sample. This result showed the high oxidation activity of the photogenerated holes, which implied accumulation of holes on the TiO₂ side rather than the g-C₃N₄ side. Then the holes would participate in the OH⁻/·OH redox reaction, as shown in Fig. 8(b). Such behavior for transport of carriers did not follow the rules of type II mode, where holes transfer from the VB of TiO₂ to the VB of g-C₃N₄ and the signal is weak due to the more negative redox potential of g-C₃N₄ (ca. +1.5 mV). Therefore, the two types TCN heterojunctions cause the same carrier transport routes.

Fig. 8

(a) EPR signals from DMPO–·OH adducts in distilled water after illumination for 5 min. (b) Possible photocatalytic mechanism for two types of heterojunctions.

4. Conclusion

Two types of TiO₂/g-C₃N₄ heterojunctions, namely physically and chemically connected heterojunctions (TCN_PHY and TCN_CHE), were prepared. TCN_PHY was constructed by electrostatic attraction by adjusting the pH of a dispersion of the two materials, while TCN_CHE was synthesized by a hydrothermal process with the formation of covalent bonds. The photocatalytic performance of the catalysts was evaluated by MB degradation experiments, and TCN_CHE showed higher photocatalytic performance than TCN_PHY due to its higher light absorbance and lower recombination rate. The formation of chemical bonds was proved to be a better strategy for TCN heterojunction construction. These results will contribute to the development of highly active photocatalysts with heterojunctions.

Acknowledgments

This work was supported by a Grant-in-Aid for Scientific Research (S) (KAKENHI, 21H05011) from the Japan Society for the Promotion of Science (JSPS) and JST SPRING, Grant Number JPMJSP2114.

REFERENCES

• 1)   J.  Low,  J.  Yu,  M.  Jaroniec,  S.  Wageh and  A.A.  Al-Ghamdi: Adv. Mater. 29 (2017) 1601694. doi:10.1002/adma.201601694
• 2)   H.  Wang,  L.  Zhang,  Z.  Chen,  J.  Hu,  S.  Li,  Z.  Wang,  J.  Liu and  X.  Wang: Chem. Soc. Rev. 43 (2014) 5234–5244. doi:10.1039/C4CS00126E
• 3)   J.  Fu,  J.  Yu,  C.  Jiang and  B.  Cheng: Adv. Energy Mater. 8 (2018) 1701503. doi:10.1002/aenm.201701503
• 4)   X.  Wang,  K.  Maeda,  A.  Thomas,  K.  Takanabe,  G.  Xin,  J.M.  Carlsson,  K.  Domen and  M.  Antonietti: Nat. Mater. 8 (2009) 76–80. doi:10.1038/nmat2317
• 5)   J.  Wen,  J.  Xie,  X.  Chen and  X.  Li: Appl. Surf. Sci. 391 (2017) 72–123. doi:10.1016/j.apsusc.2016.07.030
• 6)   B.  Zhao,  D.  Gao,  Y.  Liu,  J.  Fan and  H.  Yu: J. Colloid Interface Sci. 608 (2022) 1268–1277. doi:10.1016/j.jcis.2021.10.108
• 7)   K.  Sridharan,  E.  Jang and  T.J.  Park: Appl. Catal. B 142–143 (2013) 718–728. doi:10.1016/j.apcatb.2013.05.077
• 8)   J.X.  Sun,  Y.P.  Yuan,  L.G.  Qiu,  X.  Jiang,  A.J.  Xie,  Y.H.  Shen and  J.F.  Zhu: Dalton Trans. 41 (2012) 6756–6763. doi:10.1039/c2dt12474b
• 9)   L.  Huang,  H.  Xu,  Y.  Li,  H.  Li,  X.  Cheng,  J.  Xia,  Y.  Xu and  G.  Cai: Dalton Trans. 42 (2013) 8606–8616. doi:10.1039/c3dt00115f
• 10)   J.  Di,  J.  Xia,  S.  Yin,  H.  Xu,  L.  Xu,  Y.  Xu,  M.  He and  H.  Li: J. Mater. Chem. A 2 (2014) 5340–5351. doi:10.1039/c3ta14617k
• 11)   J.  Fu,  Y.  Tian,  B.  Chang,  F.  Xi and  X.  Dong: J. Mater. Chem. 22 (2012) 21159–21166. doi:10.1039/c2jm34778d
• 12)   W.  Zhang,  L.  Zhou,  J.  Shi and  H.  Deng: J. Colloid Interface Sci. 496 (2017) 167–176. doi:10.1016/j.jcis.2017.02.022
• 13)   N.  Boonprakob,  N.  Wetchakun,  S.  Phanichphant,  D.  Waxler,  P.  Sherrell,  A.  Nattestad,  J.  Chen and  B.  Inceesungvorn: J. Colloid Interface Sci. 417 (2014) 402–409. doi:10.1016/j.jcis.2013.11.072
• 14)   J.  Yu,  S.  Wang,  J.  Low and  W.  Xiao: Phys. Chem. Chem. Phys. 15 (2013) 16883–16890. doi:10.1039/c3cp53131g
• 15)   T.  Ueda,  R.  Koizumi,  K.  Ueda,  K.  Ito,  K.  Ogasawara,  H.  Kanetaka and  T.  Narushima: Mater. Trans. 64 (2023) 155–164. doi:10.2320/matertrans.MT-MLA2022006
• 16)   H.  Yan and  H.  Yang: J. Alloy. Compd. 509 (2011) L26–L29. doi:10.1016/j.jallcom.2010.09.201
• 17)   J.  Li,  M.  Zhang,  Q.  Li and  J.  Yang: Appl. Surf. Sci. 391 (2017) 184–193. doi:10.1016/j.apsusc.2016.06.145
• 18)   M.  Liu,  S.  Wei,  W.  Chen,  L.  Gao,  X.  Li,  L.  Mao and  H.  Dang: J. Chin. Chem. Soc. 67 (2020) 246–252. doi:10.1002/jccs.201900135
• 19)   J.  Wang,  G.  Wang,  B.  Cheng,  J.  Yu and  J.  Fan: Chin. J. Catal. 42 (2021) 56–68. doi:10.1016/S1872-2067(20)63634-8
• 20)   F.  He,  B.  Zhu,  B.  Cheng,  J.  Yu,  W.  Ho and  W.  Macyk: Appl. Catal. B 272 (2020) 119006. doi:10.1016/j.apcatb.2020.119006
• 21)   D.  Zheng,  C.  Pang and  X.  Wang: Chem. Commun. 51 (2015) 17467–17470. doi:10.1039/C5CC07867A
• 22)   H.  Cheng,  J.  Hou,  O.  Takeda,  X.-M.  Guo and  H.  Zhu: J. Mater. Chem. A 3 (2015) 11006–11013. doi:10.1039/C5TA01864A
• 23)   S.  Chen,  Y.  Hu,  S.  Meng and  X.  Fu: Appl. Catal. B 150–151 (2014) 564–573. doi:10.1016/j.apcatb.2013.12.053
• 24)   J.  Fu,  Q.  Xu,  J.  Low,  C.  Jiang and  J.  Yu: Appl. Catal. B 243 (2019) 556–565. doi:10.1016/j.apcatb.2018.11.011
• 25)   W.-K.  Jo and  T.S.  Natarajan: Chem. Eng. J. 281 (2015) 549–565. doi:10.1016/j.cej.2015.06.120
• 26)   J.  Low,  S.  Cao,  J.  Yu and  S.  Wageh: Chem. Commun. 50 (2014) 10768–10777. doi:10.1039/C4CC02553A
• 27)   X.L.  Zhang,  C.  Zheng,  S.S.  Guo,  J.  Li,  H.H.  Yang and  G.  Chen: Anal. Chem. 86 (2014) 3426–3434. doi:10.1021/ac500336f
• 28)   L.  Wang,  G.  Zhou,  Y.  Tian,  L.  Yan,  M.  Deng,  B.  Yang,  Z.  Kang and  H.  Sun: Appl. Catal. B 244 (2019) 262–271. doi:10.1016/j.apcatb.2018.11.054
• 29)   S.C.  Yan,  Z.S.  Li and  Z.G.  Zou: Langmuir 25 (2009) 10397–10401. doi:10.1021/la900923z
• 30)   Q.  Liang,  Z.  Li,  X.  Yu,  Z.H.  Huang,  F.  Kang and  Q.H.  Yang: Adv. Mater. 27 (2015) 4634–4639. doi:10.1002/adma.201502057
• 31)   I.  Papailias,  N.  Todorova,  T.  Giannakopoulou,  N.  Ioannidis,  N.  Boukos,  C.P.  Athanasekou,  D.  Dimotikali and  C.  Trapalis: Appl. Catal. B 239 (2018) 16–26. doi:10.1016/j.apcatb.2018.07.078
• 32)   Y.  Chen,  H.  Gao,  J.  Xiang,  X.  Dong and  Y.  Cao: Mater. Res. Bull. 99 (2018) 29–36. doi:10.1016/j.materresbull.2017.08.054
• 33)   R.  Hao,  G.  Wang,  C.  Jiang,  H.  Tang and  Q.  Xu: Appl. Surf. Sci. 411 (2017) 400–410. doi:10.1016/j.apsusc.2017.03.197
• 34)   J.  Wang,  W.  Lin,  H.  Hu,  C.  Liu,  Q.  Cai,  S.  Zhou and  Y.  Kong: New J. Chem. 45 (2021) 3067–3078. doi:10.1039/D0NJ05500J
• 35)   F.T.  Li,  Y.  Zhao,  Q.  Wang,  X.J.  Wang,  Y.J.  Hao,  R.H.  Liu and  D.  Zhao: J. Hazard. Mater. 283 (2015) 371–381. doi:10.1016/j.jhazmat.2014.09.035
• 36)   A.  Khan,  U.  Alam,  W.  Raza,  D.  Bahnemann and  M.  Muneer: J. Phys. Chem. Solids 115 (2018) 59–68. doi:10.1016/j.jpcs.2017.10.032
• 37)   A.  Thomas,  A.  Fischer,  F.  Goettmann,  M.  Antonietti,  J.-O.  Müller,  R.  Schlögl and  J.M.  Carlsson: J. Mater. Chem. 18 (2008) 4893–4908. doi:10.1039/b800274f
• 38)   X.  Wu,  F.  Chen,  X.  Wang and  H.  Yu: Appl. Surf. Sci. 427 (2018) 645–653. doi:10.1016/j.apsusc.2017.08.050
• 39)   J.  Li,  Q.  Li,  B.  Gupta and  J.  Yang: Mater. Res. Bull. 140 (2021) 111292. doi:10.1016/j.materresbull.2021.111292
• 40)   Y.  Chen,  J.  Qian,  N.  Wang,  J.  Xing and  L.  Liu: Inorg. Chem. Commun. 119 (2020) 108071. doi:10.1016/j.inoche.2020.108071
• 41)   J.  Xiao,  Q.  Han,  Y.  Xie,  J.  Yang,  Q.  Su,  Y.  Chen and  H.  Cao: Environ. Sci. Technol. 51 (2017) 13380–13387. doi:10.1021/acs.est.7b04215
• 42)   X.  She,  L.  Liu,  H.  Ji,  Z.  Mo,  Y.  Li,  L.  Huang,  D.  Du,  H.  Xu and  H.  Li: Appl. Catal. B 187 (2016) 144–153. doi:10.1016/j.apcatb.2015.12.046
• 43)   Z.  Lu,  L.  Zeng,  W.  Song,  Z.  Qin,  D.  Zeng and  C.  Xie: Appl. Catal. B 202 (2017) 489–499. doi:10.1016/j.apcatb.2016.09.052
• 44)   J.  Gao,  Y.  Wang,  S.  Zhou,  W.  Lin and  Y.  Kong: ChemCatChem 9 (2017) 1708–1715. doi:10.1002/cctc.201700492
• 45)   M.  Zhang,  Y.  Duan,  H.  Jia,  F.  Wang,  L.  Wang,  Z.  Su and  C.  Wang: Catal. Sci. Technol. 7 (2017) 452–458. doi:10.1039/C6CY02318E
• 46)   S.  Guo,  Y.  Tang,  Y.  Xie,  C.  Tian,  Q.  Feng,  W.  Zhou and  B.  Jiang: Appl. Catal. B 218 (2017) 664–671. doi:10.1016/j.apcatb.2017.07.022
• 47)   B.M.  Reddy,  P.  Lakshmanan and  A.  Khan: J. Phys. Chem. B 108 (2004) 16855–16863. doi:10.1021/jp047217q
• 48)   M.  Dang,  G.  Tan,  M.  Wang,  B.  Zhang,  Y.  Wang,  L.  Lv,  H.  Ren and  A.  Xia: Mater. Lett. 285 (2021) 129131. doi:10.1016/j.matlet.2020.129131
• 49)   W.  Zhao,  N.  Zhang,  J.  Sun and  J.  Zou: J. Hydrometeorol. 15 (2014) 1078–1090. doi:10.1175/JHM-D-13-0129.1
• 50)   S.  Wang,  L.  Da,  J.  Hao,  J.  Li,  M.  Wang,  Y.  Huang,  Z.  Li,  Z.  Liu and  D.  Cao: Angew. Chem. Int. Ed. 60 (2021) 9321–9325. doi:10.1002/anie.202100464
• 51)   J.  Chakhalian,  J.  Freeland,  H.-U.  Habermeier,  G.  Cristiani,  G.  Khaliullin,  M.  Van Veenendaal and  B.  Keimer: Science 318 (2007) 1114–1117. doi:10.1126/science.1149338
• 52)   J.W.  Spalenka,  P.  Gopalan,  H.E.  Katz and  P.G.  Evans: Appl. Phys. Lett. 102 (2013) 041602. doi:10.1063/1.4790155
• 53)   K.  Li,  Z.  Huang,  X.  Zeng,  B.  Huang,  S.  Gao and  J.  Lu: ACS Appl. Mater. Interfaces 9 (2017) 11577–11586. doi:10.1021/acsami.6b16191
• 54)   S.  Yang,  Y.  Lin,  X.  Song,  P.  Zhang and  L.  Gao: ACS Appl. Mater. Interfaces 7 (2015) 17884–17892. doi:10.1021/acsami.5b04368
• 55)   D.  Gao,  P.  Deng,  J.  Zhang,  L.  Zhang,  X.  Wang,  H.  Yu and  J.  Yu: Angew. Chem. Int. Ed. 62 (2023) e202304559. doi:10.1002/anie.202304559
• 56)   P.  Xia,  B.  Cheng,  J.  Jiang and  H.  Tang: Appl. Surf. Sci. 487 (2019) 335–342. doi:10.1016/j.apsusc.2019.05.064
• 57)   Y.  Xu and  M.A.  Schoonen: Am. Mineral. 85 (2000) 543–556. doi:10.2138/am-2000-0416
• 58)   W.J.  Ong,  L.L.  Tan,  Y.H.  Ng,  S.T.  Yong and  S.P.  Chai: Chem. Rev. 116 (2016) 7159–7329. doi:10.1021/acs.chemrev.6b00075
• 59)   L.  Ge,  C.  Han,  J.  Liu and  Y.  Li: Appl. Catal. A 409–410 (2011) 215–222. doi:10.1016/j.apcata.2011.10.006
• 60)   Z.  Yu and  S.S.  Chuang: J. Phys. Chem. C 111 (2007) 13813–13820. doi:10.1021/jp0715474
• 61)   A.  Houas,  H.  Lachheb,  M.  Ksibi,  E.  Elaloui,  C.  Guillard and  J.-M.  Herrmann: Appl. Catal. B 31 (2001) 145–157. doi:10.1016/S0926-3373(00)00276-9

Appendix

A.1. Preparation of g-C₃N₄ Bulk and TiO₂ Nanoparticles (NPs)

Fig. A1

(a) Photograph of prepared g-C₃N₄ bulk. (b) TEM image of synthesized TiO₂ nanoparticles.

A.2. MB Degradation Results for TCN Heterojunction with Primary g-C₃N₄ and TiO₂

Fig. A2

(a) TEM image of TiO₂/g-C₃N₄ mixture without pH adjustment. (b) MB degradation curve for physical assembled TCN heterojunction and mixture.

A.3. Reaction System for MB Degradation

Fig. A3

Schematic illustration of equipment configuration.

The facility includes a mechanic pump, a UV light source (λ = 253.7 nm) and a 500 ml glass bottle. The pump transported solutions with a speed of 70 ml/min under light irradiation. At intervals, 3 ml of the dispersion sample was taken from the system. The whole equipment was covered with aluminum foil to avoid sunlight interference.

A.4. Standard Methylene Blue Solution Concentration Calibration Reaction System for MB Degradation

Fig. A4

(a) Methylene blue UV-vis absorption spectra with standard concentration from 10 mg/l to 0.5 mg/l. (b) Calibration curve for MB peak intensity at 663 nm wavelength.

The linear relationship between C_MB and I_ABS follows the formula,

  

\begin{equation*} A = 0.159\times C + 0.038 \end{equation*}

where A is the absorption intensity and C is the concentration of methylene blue.

A.5. BET Surface Area and Pore Structure Measurement

Fig. A5

N₂ adsorption-desorption isotherms for (a) g-C₃N₄ nanosheet. (b) TCN_PHY and (c) TCN_CHE.

A.6. TGA Analysis Results and XRD Patterns for TCN Heterojunction after TGA Measurement (800°C)

Fig. A6

(a) Thermogravimetric analysis (TGA) results for TCN heterojunction. (b) XRD patterns for TCN heterojunction after TGA measurement.

A.7. Photocatalytic Reactions

Equation (A1):

  

\begin{equation} \mathrm{TCN}+ \text{h}\nu \to e_{\textit{CB}}^{-} + h_{\textit{VB}}^{+} \end{equation} (A1)

  

\begin{equation} H_{2}O_{ad} \to \textit{OH}^{-} + H^{+} \end{equation} (A2)

  

\begin{equation} \textit{OH}^{-} + h_{\textit{VB}}^{+} \to \cdot \textit{OH} \end{equation} (A3)

  

\begin{equation} O_{2ad} + e^{-} \to O_{2}^{-} \end{equation} (A4)

  

\begin{equation} O_{2}^{-} + H_{2}O \to \cdot \textit{OH} + \textit{OH}^{-} + 1/2\ O_{2} \end{equation} (A5)

  

\begin{equation} (\textit{CH}_{3})_{2} - N - R + \cdot \textit{OH} \xrightarrow{\textit{demethylation}+O_{ad}} \textit{CH}_{3} - \cdot N - R + \textit{HCHO} + H_{2}O \xrightarrow{\textit{ring breaking}} \textit{degradation} \end{equation} (A6)