Wen-Jun Zhang,
Feng-Jue Wang,
Han Liu,
Yue Wang and
Meng Chen*
Department of Materials Science, Fudan University, Shanghai 200433, China. E-mail: chenmeng@fudan.edu.cn
First published on 27th February 2023
In most of the research about graphitic carbon nitride (g-C3N4), g-C3N4 is prepared through the calcination of nitrogen-rich precursors. However, such a preparation method is time-consuming, and the photocatalytic performance of pristine g-C3N4 is lackluster due to the unreacted amino groups on the surface of g-C3N4. Therefore, a modified preparation method, calcination through residual heating, was developed to achieve rapid preparation and thermal exfoliation of g-C3N4 simultaneously. Compared with pristine g-C3N4, the samples prepared by residual heating had fewer residual amino groups, a thinner 2D structure, and higher crystallinity, which led to a better photocatalytic performance. The photocatalytic degradation rate of the optimal sample for rhodamine B could reach 7.8 times higher than that of pristine g-C3N4.
In recent decades, numerous efforts have been made to synthesize metal-free photocatalysts. At present, graphitic carbon nitride (g-C3N4) is one of the most popular metal-free photocatalysts. g-C3N4 is an n-type semiconductor consisting of only carbon and nitrogen elements, making its raw materials easy to obtain. Typically, pristine g-C3N4 can be prepared through the thermal polycondensation reaction of N-rich precursors, such as urea, dicyandiamide, and melamine.5–9 The band gap of g-C3N4 is ∼2.7 eV, and the band structure is suitable, therefore, it can be applied to degrade organic contaminants and split water under visible irradiation.10–13 Pristine g-C3N4, however, suffers from many drawbacks, including narrow optical absorption range, high recombination efficiency of charge carriers, low specific surface area, and so on, which make its photocatalytic performance unsatisfactory.14–18 Up to now, many strategies, including building heterojunctions,19,20 doping,21 and morphology controlling,22,23 have been developed to improve the photocatalytic performance of g-C3N4 materials.
In the process of thermal polycondensation, there are a large amount of non-condensed surface amino groups in g-C3N4 due to the solid-state kinetic hindrance. As a result, g-C3N4 used in most studies is essentially amorphous or semi-crystalline.24 These amino groups act as structural defects that can trap photogenerated charges, leading to low photocatalytic activity.25 To improve the crystallinity of g-C3N4, the amino groups need to be fully reacted in the thermal polycondensation. One of the common solutions to overcome the solid-state kinetic hindrance is reacting in liquid molten salt. For example, Xiang's group synthesized hollow spherical precursors via the self-assembly of melamine and cyanuric acid, and then annealed the precursors in KCl–LiCl molten salt to prepare highly crystalline g-C3N4 hollow spheres;26 Zhong's group reported a kind of Zn-doped crystalline g-C3N4 through calcining melamine in KCl–ZnCl2 molten salt.27 However, the molten salt method is faced with some disadvantages: (i) for different materials, it is important and difficult to find suitable molten salt (or a molten salt system); (ii) cleaning and collecting the product is time-consuming. Chen's group improved the crystallinity of g-C3N4 through the Ni-foam template method.28 In a typical process, Ni-foam was immersed in the dicyandiamide solution and then dried, leading to the recrystallization of dicyandiamide anchored on the Ni-foam. High-crystalline g-C3N4 nanosheets could be obtained by heating the Ni-foam at 550 °C and then removing it with hydrochloric acid. In the process of heating, Ni-foam could not only play as the template for g-C3N4 deposition but also the catalyst for the dehydrogenation of amino groups. But analogous to the molten salt method, the template method also required a long time to collect the products. Moreover, some attempts to improve the crystallinity of g-C3N4 through the high temperature–high pressure (HT–HP) route have been made according to the graphite-to-diamond phase transition.29,30 However, harsh reaction conditions limit the practical application of the HT–HP route.
The above investigations suggest us look for a simple and fast way to reduce the content of amino groups. Considering that the amino groups are the residue of incomplete thermal polycondensation, one of the most convenient methods to remove them is increasing the reaction temperature. It is well-known that the increase of the reaction temperature can improve the reaction kinetics, as well as make the chemical equilibrium move to the direction of product formation. In addition, high-temperature thermal treatment can exfoliate g-C3N4, which can help increase specific surface area and create active sites.31 However, prolonged calcination at high temperatures (≥550 °C) may lead to the decomposition of g-C3N4. Therefore, inspired by residual cooking, a modified calcination method—residual heating—has been developed. In the process of residual heating, an empty furnace is heated to a high temperature (≥650 °C) and then powered off, the waste heat in the furnace is used to calcine the precursor to obtain high-crystalline g-C3N4.
In this work, we have successfully prepared g-C3N4 with excellent photocatalytic performance through residual heating, and deeply studied the influence of the parameters in this method on the products. The effects of the chemical structure, morphology, optical property, and band structure of our samples on photocatalytic performance have also been systematically discussed according to a series of characterization results.
To investigate the effect of reaction time on the products, the preparation procedure of DCN-750 was slightly adjusted: the covered crucible with dicyandiamide was taken out of the muffle furnace after a T-minute calcination (T = 2, 4, 6, 8, 10, 20, 30, and 60 min) and quickly cooled to room temperature. After naturally cooling to room temperature, the sample was ground and collected. This sample was denoted as DCN-750-T.
Pristine g-C3N4 was synthesized via traditional calcination.5 Briefly, 5 g precursor (urea, dicyandiamide, or melamine) was put into a covered crucible and calcined at 550 °C in the air for 4 h, with a heating rate of 5 °C min−1. After naturally cooling to room temperature, the sample was ground and collected. This sample was denoted as pristine UCN, DCN, and MCN, respectively.
Fig. 1 (a) Change of the furnace temperature during the preparation process of DCN-X; (b) curves of temperature change of DCN-800 and MCN-800, and their differential curves; (c) yields of DCN-X. |
As shown in Fig. 1(c), the yields of DCN-800, DCN-750, DCN-700, and DCN-650 were 3%, 28%, 39%, and 41%, respectively, while the yield of pristine DCN was 40%. The yields of UCN-X and MCN-X were shown in Fig. S1(c):† the yield of UCN-650 was only 1%, while the yield of pristine UCN was 5%; the yields of MCN-800, MCN-750, MCN-700, and MCN-650 were 3%, 8%, 15%, and 23%, respectively, while the yield of pristine DCN was 42%. The yields of the samples prepared through residual heating were generally lower than that through traditional calcination, which should be caused by the strong thermal decomposition of precursors under high temperatures in the early stage of residual heating.
Fig. 2 TEM images of (a) pristine DCN, (b) DCN-650, (c) DCN-700, (d) DCN-750, and (e) DCN-800, the insets were their SAED patterns; (f) HRTEM image of DCN-750. |
Further morphological analysis was performed via HRTEM image. As shown in Fig. 2(f), lattice fringes were found in the HRTEM image of DCN-750, which illustrated the high crystallinity of DCN-750. Two types of exposed facets could be determined in DCN-750: the facet with a lattice distance of 0.336 nm could be attributed to the (002) facet of g-C3N4,32 and the facet with a lattice distance of 0.688 nm indicated the exposure of the (100) facet of g-C3N4.33 It was noteworthy that (100) as an exposed facet was quite different from the results in most literature.26,34–36 A relevant study by Wang's group has proved that electrons were energetically favorable to transfer toward the (100) facet in photocatalytic reactions, so the exposure of the (100) facet was beneficial to the enhancement of the photocatalytic performance.37
SAED was used to further characterize the crystallinity of DCN-X samples, and the results were shown in the insets of Fig. 2(a–e). In line with our expectation, DCN-X samples exhibited high crystallinity: there was only an incomplete ring that represented the (002) facet in the SAED pattern of pristine DCN, while many diffraction rings accompanied by irregularly distributed spots could be observed in the SAED patterns of DCN-X samples. Taking the SAED pattern of DCN-750 in the inset of Fig. 2(f) as an example, some diffraction rings could be attributed to the (100), (101), and (002) facets of g-C3N4, respectively.
Fig. 3 (a) XRD patterns and (b) FTIR spectra of DCN-X samples; (c) NMR spectra of pristine DCN and DCN-750; (d) structure of g-C3N4. |
The chemical structures of DCN-X samples were preliminarily determined by FTIR measurements. As shown in Fig. 3(b), DCN-X samples had the same FTIR spectra as pristine DCN. The peak at 808 cm−1 was ascribed to the vibrations of heptazine cores, the multiple peaks located between 1240 and 1650 cm−1 were attributed to the various vibration modes of CN heterocyclic ring units, the broad band at 2900–3300 cm−1 was related to the vibrational stretching modes of O–H and N–H.40,41 The FTIR spectra of UCN-X and MCN-X samples in Fig. S2(c and d)† also showed similar results, indicating that the chemical structure of g-C3N4 prepared through residual heating was the same as that of pristine g-C3N4.
To learn the structural difference between pristine and our samples, NMR spectra of pristine DCN and DCN-750 were measured in Fig. 3(c). In the 13C NMR spectrum of pristine DCN, three peaks located at 164.5 ppm, 162.5 ppm, and 155.3 ppm, which were belong to C–NH2, C–NH, and C–N3 bonds, respectively.42 In the 13C NMR spectrum of DCN-750, the chemical shift of the C–N3 was ∼0.4 ppm higher than that for pristine DCN. As seen in Fig. 3(d), the C atom C–N3 located in the heptazine rings, so the chemical shift meant that the heptazine rings in DCN-750 were intact. This should thank to the short calcination time during residual heating, which could not break too many heptazine rings. Owing to these heptazine rings, the exposed (100) facet of DCN-750 could be observed in the HRTEM image.
More detailed chemical structures of the samples were analyzed through XPS. The XPS full survey spectrum of DCN-750 was presented in Fig. 4(a): only C, N, and O elements existed in the as-prepared photocatalyst, which was consistent with our expectations. The high-resolution C 1s spectrum in Fig. 4(b) showed the peaks at 288.2 eV and 284.8 eV, which could be assigned to CN and C–C in the CN heterocyclic structures of g-C3N4, respectively.40,43 The N 1s spectrum, as shown in Fig. 4(c), could be divided into three characteristic peaks at 400.5 eV, 399.1 eV, and 398.5 eV, which were attributed to amino functions carrying hydrogen (C–N–H), tertiary nitrogen groups (C–N3), and sp2-hybridized nitrogen in triazine rings (C–NC), respectively.44 In addition, a small peak at 404.7 eV was generated by the π-electron excitation.45 These results were consistent with the XPS spectra of pristine DCN (Fig. S3†), which revealed the existence of g-C3N4 microstructure in DCN-750. As for the O 1s spectrum in Fig. 4(d), it could be deconvoluted into two peaks at 533.0 eV and 531.9 eV, which could be assigned to C–N–O and the surface hydroxyl groups, respectively.46
Fig. 4 (a) XPS survey spectrum of DCN-750; high-resolution XPS spectra of (b) C 1s, (c) N 1s, and (d) O 1s of DCN-750. |
To investigate the influence of the initial temperature of residual heating on the amino groups in the products, high-resolution XPS spectra of N 1s of DCN-650, DCN-700, and DCN-800 were also deconvoluted in Fig. S4.† The specific intensity values of the peak for C–N–H to that for C–NC, which could be used to evince the completion of C–N conjugate framework,25 were listed in Table 1. It could be seen that the specific peak intensity declined when the initial temperature rose from 600 °C to 750 °C, indicating the increase of initial temperature could reduce the content of amino groups, and thus lead to more complete heptazine rings.25 There was a slight increase in the specific intensity when the initial temperature reached 800 °C, which might be due to the high-temperature-induced structural destruction in g-C3N4. Moreover, the contents of the amino groups in DCN-700, DCN-750, and DCN-800 were less than that in pristine DCN, which elucidated that the unreacted amino groups could be efficiently removed through residual heating. Furthermore, the mass ratios of C/N were also estimated in Table 1. The mass ratio of C/N increased from 0.55 to 0.59 with the rising of the initial temperature, which was close to the value of highly crystalline carbon nitride hollow spheres (C/Nmass = 0.6) prepared by Xiang's group.26 Such a result suggested less content of amino groups and a better-aligned crystalline structure of g-C3N4 in DCN-750. Overall, the XPS results strongly demonstrated the feasibility of residual heating in reducing the amino groups and increasing the crystallinity of g-C3N4.
Samples | Peak intensity (C–N–H)/peak intensity (C–NC) | C/N |
---|---|---|
Pristine DCN | 0.700 | 0.549 |
DCN-650 | 0.989 | 0.555 |
DCN-700 | 0.530 | 0.577 |
DCN-750 | 0.417 | 0.589 |
DCN-800 | 0.496 | 0.585 |
Fig. 5 (a) DRS spectra and (b) Tauc plots of DCN-X samples; (c) XPS valence-band spectrum of DCN-750; (d) PL spectra of DCN-X samples. |
To determine the band gap (Eg), Kubelka–Munk model was used to obtain the Tauc plots of DCN-X samples:48
(αhν)1/n = A(hν − Eg) |
XPS valence-band spectrum was carried out to evaluate the valence band maximum (VB) and conduction band minimum (CB) of DCN-750. As shown in Fig. 5(c), the binding energy of DCN-750 was 0.96 eV, which was obtained by linear extrapolation. As a result, the band structure of DCN-750 could be obtained, which was displayed in the inset of Fig. 5(c). The valence band maximum was EV = 0.96 eV (vs. NHE), and the conduction band minimum was EC = −1.16 eV (vs. NHE).
PL measurement was performed to clarify the recombination efficiency of photogenerated carriers in our samples. It was well-known that the intensity of PL peak could reflect the recombination efficiency of photogenerated carriers, which influenced the photocatalytic activity of semiconductors. The PL spectra of DCN-X samples were measured with the excitation wavelength at 382 nm at room temperature, as depicted in Fig. 5(d). The intensive fluorescence emission peaks of all samples appeared at 440–450 nm, suggesting the transition of photogenerated electrons from n to π*.52 In terms of the intensity of the PL peaks, the peak intensity of pristine DCN was the largest, and that of DCN-650 and DCN-700 were slightly smaller; the peak intensities of DCN-750 and DCN-800 decreased strongly with the increase of initial temperature. This demonstrated the increase of initial temperature could promote the migration of photogenerated electrons and thereby reduced the recombination efficiency, which suggested the improvement of photocatalytic performance.
Fig. 6 (a) Photocatalytic rates and (b) kinetics of DCN-X; (c) photocatalytic rates and (d) kinetics of UCN-X; (e) photocatalytic rates and (f) kinetics of MCN-X. |
Photocatalytic degradation abilities of UCN-X and MCN-X samples for RhB were also tested. As shown in Fig. 6(c), 98.0% of RhB could be degraded by pristine UCN in 10 min, while the time required was 30 min by DCN-650. The comparison of kinetic rate constants in Fig. 6(d) showed that the photocatalytic performance of UCN-650 was only 38% of that of pristine UCN. Therefore, the initial temperature of 650 °C was exorbitant for the residual heating of urea, which led to the poor photocatalytic activity of UCN-650. In addition, considering the intense thermal decomposition of urea, we thought that the amount of urea should be increased while rising the initial temperature to obtain UCN samples, such as UCN-700 and UCN-800. The photocatalytic performance of MCN-X samples was shown in Fig. 6(e), which was similar to that of DCN-X samples. Within 30 min, the degradation rates of pristine MCN, MCN-650, MCN-700, MCN-750, and MCN-800 for RhB were 36.4%, 16.5%, 22.2%, 88.6%, and 82.7%, respectively. So when the initial temperature was between 650 and 750 °C, the photocatalytic performance of the MCN samples improved with the increase of initial temperature. The photocatalytic performance, however, slightly decreased when the initial temperature reached 800 °C. Such a situation was similar to that of the DCN samples, suggesting the overhigh temperature of 800 °C might break the structure of MCN-800. These results demonstrated that the optimal initial temperature for the residual heating of melamine was 750 °C. Furthermore, the photocatalytic performance of MCN-650 and MCN-700 was worse than that of pristine MCN, while MCN-750 and MCN-800 exhibited better photocatalytic performance than pristine MCN. By calculating the kinetic rate constants (Fig. 6(f)), it could be known that the photocatalytic abilities of MCN-750 and MCN-800 were 3.5 and 2.8 times higher than that of pristine MCN, respectively, which suggested the universality of residual heating to enhance the photocatalytic performance of g-C3N4 materials.
Firstly, DCN-750-T samples were preliminarily characterized. The morphology of DCN-750-T samples could be observed in TEM images, which were shown in Fig. 7. It could be found that the thin 2D structure appeared since DCN-750-4 and went thinner with the increase of reaction time. Furthermore, the crystallinity became high after 6 min according to the SAED patterns in the insets of Fig. 7. As shown in Fig. 8(a), XRD was used to study the crystal structure of DCN-750-T samples. A typical XRD pattern of g-C3N4 appeared at 6 min, suggesting the formation of the graphitic structure was rapid. With the increase of reaction time, the characteristic peak of the (002) facet slightly right-shifted, which illustrated that the crystallinity of the samples went higher gradually. The results of FTIR (Fig. 8(b)) showed that the chemical structure of g-C3N4 has formed since 2 min. These characterization results indicated that the g-C3N4 with high crystallinity formed quickly within 6 min.
Fig. 7 TEM images of DCN-750-T samples with different reaction times, the insets are their respective SAED patterns. |
Fig. 8 (a) XRD patterns, (b) FTIR spectra, (c) optimization of photocatalytic performance and yield, and (d) kinetics of DCN-750-T. |
As shown in Fig. 8(c), the photocatalytic performance of DCN-750-T samples for RhB was tested, and the yields were also calculated. The degradation rate of DCN-750-2 was only 22.2% within 30 min. When the reaction time became 10 min (DCN-750-10), the degradation rate within 30 min soared to 95.5%. With a further increase of reaction time, the degradation rate would remain stable. As for yield, it would decrease from 44% to 27% with the increase of reaction time, which could be due to the broken structure of g-C3N4 caused by the long-time calcination during the process of residual heating. The kinetic rate constants of DCN-750-T samples were estimated, which were depicted in Fig. 8(d). It could be found that DCN-750-20 had the maximal k value, which was 0.10476 min−1, and 7.8 times higher than that of pristine DCN. The yield of DCN-750-20 reached 30%, which was higher than that of DCN-750. Therefore, g-C3N4 with higher yield and better photocatalytic performance could be conveniently obtained by controlling the reaction time of residual heating. The XPS and NMR spectra of DCN-750-20 were displayed in Fig. S5,† which could demonstrate the chemical structure of DCN-750-20 was consistent with that of DCN-750.
In addition, a g-C3N4 sample was also prepared by calcining dicyandiamide under 750 °C for 20 min. The degradation rate of this sample for RhB within 30 min was 94.7% under 200 W visible light, which was similar to the photocatalytic ability of DCN-750-20. The yield of this sample, however, was only 20%, which was lower than that of DCN-750-20. This result reflected the advantage of residual heating in yield.
To comprehensively illustrate the feasibility of residual heating, we found the photocatalytic performance data of RhB degradation over other g-C3N4 materials from recent reports and made a comparison with our photocatalysts. Other g-C3N4 materials usually needed 300 W visible-light irradiation, and the degradation time was long (more than 40 min).50,53,54 As for DCN-750-20 and MCN-750 in our study, they could effectively degrade RhB within a shorter time under 200 W visible-light irradiation.
Scavenger experiments were used to identify the active oxidation species during the process of RhB degradation. Oxidation species, such as holes (h+), hydroxyl radicals (·OH), and superoxide radicals (·O2−), can play a crucial role in the reaction of dye degradation. To investigate the photocatalytic mechanism of our samples, EDTA, BQ, and IPA were introduced into the photocatalytic experiment of DCN-750-20 for RhB degradation as scavengers for h+, ·O2− and ·OH, respectively.46,56 As shown in Fig. 9(c), the degradation efficiency decreased slightly with the addition of IPA, suggesting that ·OH might play a role in the photocatalytic degradation reaction. The addition of EDTA promoted the degradation efficiency, implying that e− could influence the degradation efficiency. Considering that e− was usually ineffective in oxidation reactions, the reason for this phenomenon might be that EDTA captures h+, and then improved the separation efficiency of photogenerated carriers.56 The photocatalyst became completely useless with the addition of BQ, indicating that photogenerated ·O2− was the reactive species during the photocatalytic degradation for RhB. The upward degradation curve of DCN-750-20 + BQ after 0 min could be attributed to the change of the adsorption–desorption equilibrium after irradiation.
Given the discussion above, the plausible mechanism of photocatalytic degradation for RhB over DCN-750-20 through multiple pathways could be speculated:57–59 on one hand, photo-generated carriers were activated in the photocatalyst under the light irradiation. The photo-generated electrons moved to the surface of the photocatalyst and reacted with the dissolved oxygen in the solution to create ·O2−; simultaneously, the photo-generated holes could react with H2O molecules to produce ·OH. Then, these oxidative radicals could attack a more liable oxidation azo group to break the dye molecules, leading to the degradation of RhB. The degradation mechanism induced by photocatalyst could be described as follows:
DCN-750-20 + hν → DCN-750-20(h+) + DCN-750-20(e−) | (1) |
DCN-750-20(e−) + O2 → DCN-750-20 + ·O2− | (2) |
DCN-750-20(h+) + H2O → DCN-750-20 + ·OH | (3) |
·O2−/·OH + RhB → degradation products | (4) |
On the other hand, with the transfer of the photo-generated electrons from the excited dye molecules to the photocatalyst surface, the oxidized dye molecules might also undergo self-degradation. The excited dye molecules absorbed on the surface of the photocatalyst (RhB@DCN-750-20) could also react with hydroxyl ions to generate hydroxyl radical, which was helpful for the degradation of RhB. The degradation mechanism induced by dye could be described as follows:
RhB@DCN-750-20 + hν → ·RhB+@DCN-750-20(e−) | (5) |
·RhB+@DCN-750-20(e−) + O2 → · + RhB@DCN-750-20(·O2−) | (6) |
RhB@DCN-750-20(·O2−) + H2O → RhB@DCN-750-20(OH−) | (7) |
· + RhB@DCN-750-20(OH−) → RhB@DCN-750-20(·OH) → DCN-750-20 + degradation products | (8) |
Preparation method | Traditional calcination | Residual heating |
---|---|---|
Sample | Pristine DCN | DCN-750-20 |
Duration of heating | More than 350 min | 95 min |
Electricity consumption | 5 kW h | 3.5 kW h |
Yield | ∼40% | ∼30% |
Degradation efficiency for RhB | 36.1% in 30 min | 95.8% in 30 min |
Footnote |
† Electronic supplementary information (ESI) available. See DOI: https://doi.org/10.1039/d2ra08324h |
This journal is © The Royal Society of Chemistry 2023 |