*2.4. Photocatalytic Study*

The photocatalytic performance of g-C3N4, TiO2, and composites was evaluated for IMI degradation under UV-Vis light irradiation in Figure 5. Photolysis of IMI degradation was carried out under the same conditions, as can be seen from Figure S2. It was found that the photolysis is not the main cause of effective degradation of IMI. On the other hand, the treatments of IMI with catalysts are less effective in dark conditions. From this result, the g-C3N4 system exhibited low photocatalytic efficiency in the degradation of IMI. This could be because of the fast recombination of electron–hole pairs, as evidenced by PL spectra (Figure 4c). However, it was found that exfoliated g-C3N4 showed higher photocatalytic activity than bulk-g-C3N4. It could be explained as the effects of a larger specific surface area, narrow band gap, and nitrogen defects which improved photogenerated charge separation and transfer [56]. In addition, the incorporation of TiO2 clearly enhances the photocatalytic activity of g-C3N4. Specifically, 0.5CNS/TiO2 and 1CNS/TiO2 exhibited excellent photocatalytic activity, and photocatalysts were able to degrade 93.1% and 88.3% of IMI within 150 min, while pure TiO2 and g-C3N4 only degraded IMI by 79.7% and 51.8%, respectively. It is reasonable that there might have been a synergetic effect between TiO2 and g-C3N4. g-C3N4 can narrow the band gap energy and increase solar absorption efficiency. In addition, Ti3+ and oxygen vacancies (Ov) in TiO2 can suppress the recombination of photogenerated electron–hole pairs and promote charge separation [57] which led to high photocatalytic degradation of IMI. As seen in Figure 4c, CNS gave a very broad PL spectrum having a very high intensity. In addition, the PL intensity from 0.5CNS/TiO2 was found just slightly higher than that from the TiO2 material (much weaker than that of CNS). Several works [58,59] related the intensity of PL spectra to the oxidation–reduction potential between the conduction band and the valence band. PL spectra with lower intensity described a low probability of photogenerated electron–hole recombination. Although PL results suggest a slightly faster recombination rate on the 0.5CNS/TiO2, its relatively narrow band gap (compared with TiO2) promoted superior IMI removal efficiencies (Figure 5). It should be noted that the loading level of g-C3N4 played an important role in improving the IMI photodegradation. It was found that the photocatalytic rate of activity slightly decreased after 90 min irradiation time when the loading of g-C3N4 was increased from 4% to 15%. This might be due to the fast recombination of the electron–hole pair in g-C3N4. Furthermore, from Figure 5, 0.5CNS/TiO2 gave a higher IMI degradation rate than that of 1CNS/TiO2, as seen from the slope. However, the IMI removal efficiencies obtained from the 4CNS/TiO2 treatment were higher during 0–120 min. Effective photodegradation of organic compounds requires a suitable amount of stable radical species in the aqueous media. Too-high concentrations of radical species may cause termination of the radical reaction pathway, while insufficient radical concentrations resulted in slow degradation rates and low removal efficiencies. We could explain the removal efficiencies by the varied concentrations of radicals over time. Hence, after 150 min, three samples (i.e., 0.5CNS/TiO2, 1CNS/TiO2, and TiO2) gave % IMI removal efficiencies of 80% and above, likely due to the suitable amount of stable radical species in the aqueous media through the prolonged degradation process.

**Figure 5.** (**a**) Photocatalytic degradation of imidacloprid when treated with g-C3N4, TiO2, and g-C3N4/TiO2 composites under UV-Vis light irradiation (10 ppm of pesticide and 1g/L of catalyst loading), and (**b**) the first-order kinetic fitting curve of the photocatalytic IMI degradation during 30 min irradiation time.

The initial rate constants, derived from the first-order kinetic fitting curve (Figure 5b), for the photodegradation of IMI from the highest to the lowest, are given in the order of 4CN/TiO2 (1.50 × <sup>10</sup>−<sup>2</sup> min−1), 10CNS/TiO2 (9.96 × <sup>10</sup>−<sup>3</sup> min−1), 0.5CNS/TiO2 (9.70 × <sup>10</sup>−<sup>3</sup> min−1), 15CNS/TiO2 (8.26 × <sup>10</sup>−3), CNS (8.00 × <sup>10</sup>−<sup>3</sup> min−1), TiO2 (7.60 × <sup>10</sup>−<sup>3</sup> min−1), bulk-CN (6.13 × <sup>10</sup>−<sup>3</sup> min−1), and 1CNS/TiO2 (1.88 × <sup>10</sup>−<sup>3</sup> min−1) catalysts. As a result, the initial rate constants are poorly correlated with the IMP removal efficiencies after 180 min of irradiation time, possibly due to the stability of radical species as a function of time discussed earlier.

### *2.5. Reusability and Regeneration*

The stability of the photocatalysts was evaluated over multiple cycles of IMI degradation. As shown in Figure 6a, the IMI removal efficiency of 0.5CN/TiO2 decreased significantly in the fourth cycle. The SEM image (Figure 6b,c) shows that the sheet-like morphology of the photocatalyst remained. However, the surface of the catalysts could be covered either by reactants or products that hindered photocatalytic performance. A regeneration experiment was carried out. After the photocatalysis experiment, the catalyst was separated from the reaction mixture by centrifugation. The used photocatalyst was regenerated by stirring in water (dark) for 1 h and irradiated for 2.5 h before using it in the next cycle. It was found that the 0.5CN/TiO2 composite still kept ~91% regeneration efficiency at the end of the fourth cycle, indicating a relatively high regeneration potential of the nanocomposite.

From this work, the bulk carbon nitride is less suitable than the exfoliated material to be incorporated with TiO2 for photocatalytic applications. The IMI removal efficiencies obtained from the 4CNS/TiO2 treatment are significantly higher (ca. 30%) than those obtained from 4CN/TiO2 (Figure S1, Supplementary Data). The photocatalytic performance of several carbon nitride based composites in the degradation of imidacloprid is given in Table 1.

**Figure 6.** (**a**) Reusability and regeneration performance test of 0.5CN/TiO2 for imidacloprid degradation, SEM images of (**b**) fresh and (**c**) spent 0.5CN/TiO2 photocatalyst



As seen in Table 1, a quite prolonged reaction time (5 h) was required in order to achieve high IMI removal efficiencies in the photocatalytic treatments of IMI (aq) over the g-C3N4 materials, and the photocatalytic performance of g-C3N4 is precursor-dependent. Direct comparison of the catalytic performance of the reported photocatalysts and those developed in this work could not be entirely appropriate as each report utilized specific performance testing setups and conditions (initial concentration, catalyst loading, and reaction time). Nevertheless, a greater number of steps and expensive chemicals would be required to prepare several functional photocatalysts (0.04C60/PCN, Ag-Bi2O3/g-C3N4, Bi2WO6: NH2-MOF, Ag4V2O7/g-C3N4), compared to this work.

### *2.6. Photocatalytic Mechanism*

To find out the major active species for the photocatalytic oxidation, several scavengers were added to the photocatalytic system individually to trap and remove active species (Figure 7). Ammonium oxalate (AO), isopropanol (IPA), and benzoquinone (BQ) act as scavengers to holes (h+), hydroxyl radical (•OH), and superoxide radical (•O2 −), respectively. The addition of p-benzoquinone had a little effect on the photocatalytic degradation of IMI, implying that •O2 − has a minor role in the reaction as an oxidative species. In contrast, the photodegradation activity of the 0.5CNS/TiO2 had a dramatic decrease with the addition of IPA and AO, suggesting that both OH− and holes are the main oxidative species in this system.

**Figure 7.** Effects of different scavengers on the photocatalytic degradation of IMI.

In order to describe the photocatalytic mechanism of 0.5CNS/TiO2 for the degradation of IMI, the CB and VB edge potentials of g-C3N4 and TiO2 were calculated from Equations (2) and (3) [66].

$$\rm{E\_{CB}} = \rm{\chi} - \rm{E\_c} - 1/2\rm{E\_g} \tag{2}$$

$$\mathbf{E\_{VB}} = \mathbf{E\_{CB}} + \mathbf{E\_{g}} \tag{3}$$

where X is the absolute electronegativity of the atom semiconductor, and the X values of TiO2 and g-C3N4 are 5.8 eV and 4.73 eV, respectively [66]. Ec is the energy of free electrons of the hydrogen scale (4.5 Ev). Eg is the band gap of the semiconductor which is 2.93 and 3.20 eV for g-C3N4 and TiO2, respectively. Therefore, the reductive potentials of the conduction band (CB) are −0.30 and −1.23 V for TiO2 and g-C3N4, and the oxidizing potentials of the valence band (VB) of TiO2 and g-C3N4 are +2.90 and +1.70 V, respectively.

Based on the above results, the possible Z-scheme photocatalytic mechanism of g-C3N4/TiO2 was proposed as shown in Figure 8. Under UV-Vis irradiation, TiO2 absorbed photon energy, and then electrons were excited from the VB to the CB. The photogenerated holes tended to stay in the VB of TiO2, whereas photogenerated electrons on the CB of TiO2 can be directly transferred into the VB of g-C3N4 due to their proximity to each other. Then, the electrons in the VB of g-C3N4 are further excited into the CB. This resulted in an efficient charge separation of the photo-induced electron–hole pair and an enhancement in their oxidation–reduction ability. Specifically, the presence of Ti3+ and oxygen vacancy could be an important reason for the hindrance of the electron–hole recombination. It was found that the photogenerated holes (h+) in the VB of TiO2 (EVB = 2.90 V vs. NHE) have the ability to oxidize H2O or hydroxyl ions (OH−) to hydroxyl radicals (•OH), while the photogenerated h+ in the VB of g-C3N4 (EVB = 1.70 V vs. NHE) is not sufficient for the oxidation of H2O to hydroxyl radicals. In addition, the photogenerated electron in the CB of g-C3N4 was

trapped on the surface to form reactive superoxide radical ions (•O2 −). The photocatalytic mechanism was consistent with the scavenger experiments in which the hydroxyl radical and holes were the principal reactive species for the IMI degradation, whereas the superoxide radical had a minor role. The Z-scheme photocatalyst was suggested since the photogenerated h+ on the TiO2/g-C3N4 composite has a sufficient oxidation potential for producing •OH radicals [67]. Evaluated by using Equations (2) and (3), the reduction potential of g-C3N4 (+1.70 V) is less positive to oxidize H2O to •OH (+1.99 V). Thus, the holes in the VB of g-C3N4 cannot adsorb water molecules near the surface of g-C3N4 to generate hydroxyl radicals (•OH). Note that •OH radicals can be produced on semiconductors with an oxidation potential of 2.4 V (and above) versus NHE. The scavenging testing indicated that •OH radicals are the key radicals promoting effective IMI degradation. The Z-scheme g-C3N4/TiO2 composites showed better photocatalytic performance than TiO2 or g-C3N4 alone. However, with the content of g-C3N4 in g-C3N4/TiO2 being in excess, numerous photo-induced electrons and holes would recombine easily. Therefore, the 0.5CNS/TiO2 sample displayed the best photocatalytic performance among these different g-C3N4/TiO2 photocatalysts.

**Figure 8.** Photocatalytic mechanism of 0.5CNS/TiO2 for degradation of imidacloprid.
