3.1. Characterization of Samples
The XRD patterns of g-C
3N
4, ZnO-LDH and
[email protected]3N
4 are shown in
Figure 1. Two diffraction peaks at 13.4° and 28.4° two theta can be seen from the g-C
3N
4 sample, attributing to the diffractions of the (100) and (002) planes of g-C
3N
4. The (100) peak of g-C
3N
4 represents the heptazine unit with an interplanar separation of 0.66 nm and (002) peak represents the graphitic-like layer with an interlayer distance of 0.31 nm [
25]. The XRD pattern of ZnO-LDH showed the presence of both ZnAl-LDH and ZnO [
6]. The strong peaks at 31.2°, 33.8°, and 35.6° two theta found in ZnO-LDH sample are indexed to the diffraction of the (100), (002) and (101) planes of ZnO (JCPDS No. 05-0664), indicating the formation of ZnO in the sample [
26,
27]. The diffraction peaks at 8.6° and 12.1° are ascribed to the diffractions of the (003) and (006) planes of ZnAl-LDH in the ZnO-LDH sample. The interlayer distance of ZnAl-LDH in the ZnO-LDH sample d
(003) was 1.03 nm, which is larger than that of conventional carbonate intercalated LDH (0.73 nm), resulting from the intercalation of EG during solvothermal treatment [
28,
29,
30]. The
[email protected]3N
4 sample exhibited an XRD pattern similar to that of the ZnO-LDH sample. No characteristic diffraction peaks of g-C
3N
4 can be seen, probably due to the relatively weak intensity of the XRD peaks of g-C
3N
4.
Figure 2A shows the XPS survey spectrum of the g-C
3N
4 sample. It can be seen that the g-C
3N
4 sample synthesized in this work consisted of mainly C and N elements (the atomic concentration: 42.6% for C and 55.7% for N) with a small amount of oxygen (1.65%). Based on XPS data, the molar ratio of C to N is 0.76, closer to 0.75, suggesting the formula of g-C
3N
4. The small amount of oxygen in g-C
3N
4 may be caused by incomplete polymerization of urea.
Figure 2B shows the peak deconvolution results of N 1s XPS spectrum of g-C
3N
4. Three peaks at 399.1, 400.4 and 401.6 eV, attributing to sp
2-hybridized aromatic N bonded to carbon atoms (C=N–C), the tertiary N bonded to carbon atoms in the form of N–(C)
3 and N–H side groups, respectively, can be seen [
31,
32]. A weak peak at 404.5 eV ascribing to the π-excitations is also seen [
24]. The high-resolution C 1s XPS spectrum of (
Figure 2C) was deconvoluted into three peaks at 288.3, 286.4, and 284.8 eV, corresponding to sp
2-bonded carbon (N–C=N), C–O and graphitic carbon (C–C), respectively [
24,
32]. The graphitic carbon C–C at 284.8 eV can usually be observed in carbon nitrides [
16,
33]. The XPS survey scan of
[email protected]3N
4 composite is shown in
Figure 2D. The presence of O, C, N, Zn and Al in the
[email protected]3N
4 sample is clearly seen. The atomic concentrations for O/C/N/Zn/Al are 41.59/36.52/4.93/6.67/6.80 (%) from the survey of XPS, which represent the superficial atomic ratios. The high-resolution XPS spectra of N 1s, C 1s, Zn 2p, O 1s, and Al 2p are displayed in
Figure 2E–I, respectively. The binding energies for N 1s and C 1s of
[email protected]3N
4 in
Figure 2E,F showed similar deconvolution peaks as pristine g-C
3N
4 but with lower energies compared to g-C
3N
4, suggesting the strong electrostatic interaction between ZnO-LDH and g-C
3N
4 [
34,
35]. The strong peaks of C–C and C–O were due to the existence of EG and CO
32− in
[email protected]3N
4. The binding energy values of Zn 2p
3/2 and Zn 2p
5/2 were fitted with 1022 and 1045 eV in
Figure 2G. The high-resolution XPS spectra for O 1s and Al 2p in
[email protected]3N
4 composites were also fitted in
Figure 2H,I respectively. The XPS spectra for ZnO-LDH are illustrated in
Figure S1. The binding energy of C 1s in ZnO-LDH only displayed one peak, and no N peaks were detected in ZnO-LDH. Thus, the XPS spectra confirm that layered g-C
3N
4 was loaded on ZnO-LDH successfully.
The surface charges of g-C
3N
4, ZnO-LDH and
[email protected]3N
4 were measured by LSE. As listed in
Table 1, the g-C
3N
4 possessed a positive charge of 20.2 mV attributed to the –NH
2/NH functional groups at the heptazine rings generated from the incomplete polymerization of g-C
3N
4 [
36]. The uncondensed amine functional group and the edge cyano-group bring positive-charge characters for g-C
3N
4 [
37,
38]. During the synthetic process with ultrasonic treatment, g-C
3N
4 was unfolded and mixed well with NaOH in EG solution, and the pH of the mixture was changed to 13. The zeta potential of the g-C
3N
4 was changed into −23.1 mV, attributing to the hydroxyl groups’ adsorption on the g-C
3N
4 surface. Therefore, after adding Zn and Al ions, hydroxides generated and anchored onto g-C
3N
4. Further solvothermal treatment would reduce the aggregation of the
[email protected]3N
4 composite. After the hybridization of g-C
3N
4 and ZnO-LDH (the weight ratio of g-C
3N
4 in
[email protected]3N
4 is 14.6 wt %, calculated from experimental parameter and confirmed by TG-DTA, see
Figure S2), the zeta potential of the composite was shifted to 32.9 mV because of the dominant content of ZnO-LDH (30.4 mV), as illustrated in
Figure S3.
The morphologies of g-C
3N
4, ZnO-LDH, and
[email protected]3N
4 composite are presented in
Figure 3. As shown in
Figure 3A, TEM of g-C
3N
4 exhibits a flake-like morphology with irregular interstitial pores at the edge of the flake. The porous structure was generated due to releasing NH
3 and CO
2 gas during the thermal treatment of urea. The surface area of g-C
3N
4 was 128.6 m
2g
−1, much higher than the literature reports (
Table 1) [
36,
39]. Such a high surface area of g-C
3N
4 might result from the irregular interstitial pore in g-C
3N
4. As shown in
Figure S4a, bulk g-C
3N
4 exhibits overlapped wrinkles and the randomly aggregated g-C
3N
4 sheets. (The SEM of bulk g-C
3N
4 is given in
Figure S4b, which also reveals the layered structure.) After hybridizing ZnO-LDH with C
3N
4, TEM image in
Figure 3B shows the aggregates of the
[email protected]3N
4 composite. Different from pristine ZnO-LDH (
Figure S4c), ZnO-LDH nanoparticles grew on the g-C
3N
4 sheets in
[email protected]3N
4 composite, suggesting the good affinity between ZnO-LDH and g-C
3N
4.
Figure 3C,D are the SEM images of the
[email protected]3N
4 and ZnO-LDH composite, respectively. As seen in
Figure 3C, ZnO nanoparticles were evenly distributed on the surface of LDH in
[email protected]3N
4 composite, which was similar to the characteristic of ZnO-LDH in
Figure 3D. Different from ZnO-LDH, a gray veil can be distinguished from the SEM of
[email protected]3N
4, which represents a covering layer of g-C
3N
4 on ZnO-LDH. The average particle sizes of ZnO in
[email protected]3N
4 and ZnO-LDH are 3 nm and 5 nm respectively, which are coincident with the ZnO crystal sizes (
Table 1) calculated from XRD. In the
[email protected]3N
4 composite, the highly exfoliated g-C
3N
4 wrapped up ZnO-LDH particles tightly, which not only provided an intimate contact and cooperation between g-C
3N
4 and ZnO-LDH but also increased the surface area of the
[email protected]3N
4 photocatalyst (152.5 m
2g
−1). The HRTEM image and the EDX for the
[email protected]3N
4 are provided in
Figure S5. The lattice spaces for the (100) and (101) planes of ZnO were measured to be 0.282 nm and 0.253 nm, in good agreement with d
(100) and d
(101), determined from XRD pattern (0.286 nm and 0.252 nm). The EDX analysis shows the presence of Zn, Al, O, C, and N with a Zn/Al atomic ratio of ~1, matching well with the ratio from XPS. The elemental mapping of
[email protected]3N
4 is presented in
Figure S6, exhibiting a uniform distribution of Zn, Al, O, C, and N in the composite. The textural structure of g-C
3N
4, ZnO-LDH and
[email protected]3N
4 studied by N
2 adsorption-desorption isotherm at 77 K are shown in
Figure S7 and listed in
Table 1.
[email protected]3N
4 shows the highest surface area compared to g-C
3N
4 and ZnO-LDH, which facilitates the high contact between photocatalyst and dyes. Moreover, the intimate two-dimensional nanojunction between ZnO-LDH and g-C
3N
4 favours the photogenerated charge carriers’ transfer between ZnO-LDH and g-C
3N
4, which may be a key factor for photocatalytic activities of the
[email protected]3N
4 photocatalyst.
3.3. Photocatalytic Degradation Performance on MB of [email protected]3N4 under UV and Visible Light
After saturated adsorption of MB, the photocatalytic degradation of MB dyes on commercial ZnO, ZnO-LDH, g-C
3N
4, and
[email protected]3N
4 was undertaken under UV and visible-light irradiation. As shown in
Figure 5a,
[email protected]3N
4 composite removed all of the MB in the water after 1 h UV irradiation. A total of 55.0% of MB was removed by g-C
3N
4, and 21.0% of MB was removed by ZnO-LDH.
[email protected]3N
4 composite showed better photocatalytic performance compared to C
3N
4 and ZnO-LDH. However, the commercial ZnO also removed 100% of MB in the water in 20 min under UV irradiation. Interestingly, as shown in
Figure 5b, 100% of MB was also removed by the
[email protected]3N
4 composite in 4 h under visible-light irradiation; on the contrary, only 27.2% of MB could be degraded on commercial ZnO photocatalyst. The low photocatalytic activity of commercial ZnO under the visible light was ascribed to its wide bandgap, which cannot be excited upon visible-light irradiation. The removal rate of MB over ZnO-LDH and g-C
3N
4 under visible-light irradiation was 15.3% and 64.1% respectively. The better photocatalytic performance of
[email protected]3N
4 composite under UV or visible light was ascribed to its narrower bandgap, high surface area and increased contact surface area between MB and the composite. As mentioned in MB adsorption, a large amount of MB was absorbed on the surface of
[email protected]3N
4; the intimate contact between MB and photocatalyst shortened the mass transfer, which would lead to the dramatic improvement in catalytic performance. Moreover, a combination of ZnO-LDH and g-C
3N
4 reduced the bandgap of the composite, which increased the photon utilization efficiency. Most importantly, the heterojunction between ZnO-LDH and g-C
3N
4 could facilitate the transfer of excited photoelectrons during the photocatalytic process, which increased the photocatalytic effect fundamentally. Moreover, after the reaction, the color of the sediment changed back to the original yellowish (the color after adsorption of MB was blue), suggesting MB has been fully decomposed.
The photocatalytic degradation kinetics of MB on the photocatalysts were investigated and shown in
Figure 5c. The photocatalytic profile of MB on
[email protected]3N
4 followed pseudo-first-order kinetics plot by the equation:
where
k is the pseudo-first-order rate constant, C
0 and C are the MB concentration in solution at times 0 and t, respectively.
After fitting,
k of commercial ZnO, ZnO-LDH, g-C
3N
4 and
[email protected]3N
4 in
Table 1 were 0.0775, 0.0378, 0.185, and 0.487 min
−1, respectively. The value of
k gives an indication of the activity of the photocatalyst [
42].
[email protected]3N
4 had the highest rate constant among all the photocatalysts under visible-light irradiation, almost seven times as high as ZnO. The highest rate constant of
[email protected]3N
4 further demonstrated the better photocatalytic performance of
[email protected]3N
4 than that of commercial ZnO. The reusability of
[email protected]3N
4 was further tested. As shown in
Figure S10, the removal rate can still be 90% at the fifth cycle.
UV-vis diffuse reflection spectroscopy (DRS) was used to test the light-harvesting ability of ZnO, g-C
3N
4, ZnO-LDH and
[email protected]3N
4 samples. As shown in
Figure 5d, sharp absorption edges for all samples were in the range of 380~450 nm. The bandgap can be inferred from the UV-vis absorption measurements. The bandgaps of ZnO, g-C
3N
4, ZnO-LDH and
[email protected]3N
4 were calculated to be 3.20, 2.72, 3.08 and 3.06 eV, respectively. With the existence of g-C
3N
4, we expected that the bandgap of
[email protected]3N
4 would be tuned to the lower bandgap. However, the bandgap of
[email protected]3N
4 was 3.06 eV, slightly lower than that of ZnO-LDH (3.08 eV) instead of being close to that of g-C
3N
4 (2.72 eV). No change in the bandgap of the
[email protected]3N
4 composite may be due to the low content of g-C
3N
4. The other reason should be that during the solvothermal treatment process, the g-C
3N
4 was unfolded and covered on the ZnO-LDH. Therefore, the strong quantum confinement effect (QCE) derived from the highly extended g-C
3N
4 increased the bandgap of g-C
3N
4 simultaneously [
10]. PL spectroscopy was used to investigate the separation efficiency of photoexcited electron-hole pairs [
14,
25]. All the samples were excited at 360 nm, and the emission spectra were recorded in a range between 380 and 550 nm. As shown in
Figure 5e, the g-C
3N
4 had the highest emission peak at 470 nm. The emission peak at 470 nm was ascribed to the band-band PL phenomenon with the energy of light approximately equal to the bandgap energy of g-C
3N
4. This high intensive emission was attributed to the direct recombination of excitons [
43]. Compared to g-C
3N
4, the emission intensity of
[email protected]3N
4 was much lower, suggesting that their e-h
+ pair recombination rate was much lower. The strong separation of charge carriers resulted in the potentially higher photocatalytic activity for
[email protected]3N
4. The charge transport process occurs in photocatalyst under dark condition, which directly reflects its capacity to shuttle and convey charge carriers to the targeted reactive sites [
25]. Thus, to deeply understand the charge transport behaviour of
[email protected]3N
4 in the absence of light excitation, EIS measurements were carried out under dark condition.
Figure 5f displayed the EIS Nyquist plots of all the samples. As already known, the arc radius on the EIS Nyquist plot reflects the reaction rate on the surface of the electrode. The smaller the arc radius, the more effective the separation of photogenerated electron-hole pairs, and the higher the efficiency of charge immigration across the electrode-electrolyte interface [
20,
21,
44]. Among all the samples,
[email protected]3N
4 showed the smallest diameter for arc radius, suggesting its lowest resistance for interfacial charge transfer from the electrode to electrolyte molecules. Therefore, EIS measurements were consistent with PL data, demonstrating that the
[email protected]3N
4 had lower resistance than other samples and made the separation and immigration of photogenerated charges more efficient, indicating the high photocatalytic activity for
[email protected]3N
4.
3.4. Proposed Mechanism under UV and Visible-Light Irradiations
As already known, detection of the main oxidative species in the photocatalytic process is a significant tool to reveal the photocatalytic mechanism. The active species generated during the photocatalytic process can be detected through trapping by tertbutyl alcohol (
t-BuOH) and ethylenediaminetetraacetic acid disodium salt (EDTA-2Na) [
23].
t-BuOH can be the scavenger for radicals like hydroxyl and superoxide; EDTA-2Na is the scavenger for holes. As shown in
Figure 6A, in the
[email protected]3N
4 system, the MB concentration decreased dramatically upon UV irradiation without adding trapping chemicals. However, the addition of
t-BuOH only resulted in a small change in the photocatalytic degradation of MB. On the contrary, the photocatalytic activity of
[email protected]3N
4 was greatly suppressed by the addition of EDTA-2Na. The experiment results indicated that holes were the main oxidative species when the photocatalyst was under UV irradiation. When the photocatalytic reaction was under visible-light irradiation, the situation was reversed. As shown in
Figure 6B, the addition of
t-BuOH suppressed the photocatalytic degradation compared to the addition of scavenger EDTA-2Na. The experiment result suggested that the radicals were the main oxidative species when photocatalytic degradation was processed under visible irradiation.
Through the adsorption and photocatalytic performance results, we proposed the following adsorption and photocatalytic mechanism shown in
Figure 7. As shown in
Figure 7,
[email protected]3N
4 composites were mixed with OrgII solution and then OrgII was absorbed on the surface of
[email protected]3N
4 composites via electrostatic interaction and π-π conjugation adsorption and further intercalated into ZnO-LDH via layered adsorption including ion exchange. For cationic MB dye, MB was first adsorbed on the surface of
[email protected]3N
4 by π-π conjugation. Upon UV irradiation, ZnO-LDH could be excited to produce photogenerated electron-hole pairs. Since the valence band (VB) position of ZnO-LDH is lower than the highest occupied molecular orbital (HOMO) of g-C
3N
4, the photogenerated holes on ZnO-LDH could directly transfer to g-C
3N
4 [
23]. g-C
3N
4 is relatively stable with holding holes. The g-C
3N
4 with holes would accept electrons from MB degradation and then return to the ground state. Upon visible-light irradiation, g-C
3N
4 instead of ZnO-LDH absorbed visible light to induce π-π* transition, transporting the excited-state electrons from HOMO to the lowest unoccupied molecular orbital (LUMO). The LUMO potential of g-C
3N
4 is more negative than the conduction band (CB) edge of ZnO-LDH, due to the comparable energy difference between the CB of ZnO-LDH and g-C
3N
4; there is a strong thermodynamic driving force for electron transfer from excited g-C
3N
4 to ZnO-LDH [
45]. The electrons would subsequently transfer to the surface of
[email protected]3N
4 to react with water and oxygen by generating superoxide and hydroxyl radicals. The radicals can subsequently oxidize the MB into CO
2 and H
2O.