D. H.
Wang
ab,
J. N.
Pan
ab,
H. H.
Li
ab,
J. J.
Liu
ab,
Y. B.
Wang
ab,
L. T.
Kang
*ab and
J. N.
Yao
*c
aKey Laboratory of Design and Assembly of Functional Nanostructures, Chinese Academy of Sciences, Fuzhou, 350002, P.R. China. E-mail: longtiank@fjirsm.ac.cn
bFujian Provincial Key Laboratory of Nanomaterials, Fujian Institute of Research on the Structure of Matter, Chinese Academy of Sciences, Fuzhou, Fujian 350002, P.R. China
cBeijing National Laboratory for Molecular Sciences (BNLMS), Institute of Chemistry, Chinese Academy of Sciences, Beijing 100190, P. R. China. E-mail: jnyao@iccas.ac.cn
First published on 18th November 2015
Due to the two-dimensional flexible structure and abundant pendant amine, graphitic-C3N4 (g-C3N4) may be easily modified by organic molecules as a promising photocatalyst for solar H2 production from water. Here, through a simple liquid chemical reaction between g-C3N4 and the precursor of μ-oxo dimeric iron(III) porphyrin [(FeTPP)2O], we provide a novel route to construct a pure organic heterostructure of g-C3N4/(FeTPP)2O on the basis of the π–π and the Fe–amine interactions. The experimental results demonstrated that (FeTPP)2O acted not just as a photosensitizer, but also played the role of a charge promotor to prohibit the recombination of the excited electrons and holes of g-C3N4. As compared to pure or mixed g-C3N4 and/or (FeTPP)2O, the obtained pure organic g-C3N4/(FeTPP)2O heterostructure exhibited dramatic photocatalytic H2 production under solar light without any cocatalysts.
g-C3N4, a new metal-free polymeric semiconductor,11,12 stands out from a mass of photocatalysts as a shining star due to its appealing electronic structure with a medium band gap for both water reduction and oxidation, and good chemical and physical stability.13 Especially, the polymeric semiconductor is only composed of C and N elements, and can be easily prepared from low-cost and environmentally friendly precursors,14–17 such as urea,15 thiourea16 and melamine.17,18 The adjustable photocatalytic activity in visible light combining with its low cost make it potentially useful in a variety of applications.19 The utilization of g-C3N4 as a photocatalyst for hydrogen evolution has been intensively studied in recent years, nevertheless, the photocatalytic efficiency of pristine g-C3N4 is still not satisfactory due to the rapid recombination of photo-excited electron–hole pairs, low visible utilization efficiency and small surface area.20 To break these limitations, many strategies such as textural managing,22–24 doping,25,26 heterostructure construction,20,27 dye sensitization,28,29etc., have been developed.21–29 Among them, the formation of heterostructures holds great potential to separate the electron–hole pairs because it is helpful for the charge transfer across the interface of the heterostructure and restrains the recombination of electrons and holes.20,30 Now, semiconductor heterostructures are commonly applied in inorganic/inorganic and inorganic/organic semiconductors, but seldom observed in an organic/organic system.20,30,31 For organic semiconductors, Coulomb binding energies are observed in the range of several hundred meV for the Frenkel exciton, which is higher as compared to that for the Wannier exciton in inorganic semiconductors.32,33 Thus, the fabrication and study of pure organic semiconductor heterostructures as photocatalysts, for example the g-C3N4–P3HT polymer composite34,35 and CNS–CN heterojunction,36 is very important in order to more efficiently avoid the rapid recombination of photo-generated holes and electrons.
In fact, g-C3N4 may be easily modified with organic small molecules as a promising photocatalyst due to its two-dimensional flexible structure and abundant pendant amine. Here, (FeTPP)2O was chosen to construct a pure organic heterostructure with g-C3N4 due to its strong absorption bands in the visible region, large π-conjugated aromatic system and outstanding chemical and thermal stability.37 Furthermore, the (FeTPP)2O is found to be an active centre in many enzymes, which has proven to be the versatile, potential catalyst for the oxidation of hydrocarbons in air in the absence of co-reductants or solvents.38 Here, through a simple liquid chemical reaction between g-C3N4 and a precursor of (FeTPP)2O, we provide a novel route to construct the heterostructure between g-C3N4 and organic small molecules on the basis of the π–π and the Fe–amine interactions. Unlike the similar reported several organic dye-sensitized C3N4 systems,39–41 in which they do not efficiently work without cocatalysts, (FeTPP)2O in our composite is not only a photosensitizer for wider photon absorption, but also obviously promoted the separation of the photo-induced electrons and holes and the production of photocatalytic H2 in the absence of a cocatalyst.
The Fourier transform infrared spectroscopy (FTIR) spectra of pristine g-C3N4, pure (FeTPP)2O and a series of g-C3N4/(FeTPP)2O composites are obtained and shown in Fig. S3.† For the pristine g-C3N4, the intense band at 812.1 cm−1 represents the out-of-plane bending vibration characteristic of heptazine rings. The bands from 1000–1750 cm−1 are from the stretching and bending modes of nitrogen containing heterocycles. The broad feature at 3250 cm−1 is usually assigned to the stretching modes of –NH– groups.42,43 Unfortunately, FTIR spectra in Fig. S3† exhibits the typical bands of g-C3N4, however no noticeable trace of (FeTPP)2O can be found on account of the low loading amount of (FeTPP)2O (lower than 10 wt%). Based on the above experimental results and analysis, the existence of (FeTPP)2O and g-C3N4 in as-obtained samples were proved, which indicates the successful synthesis of the composite of g-C3N4/(FeTPP)2O in this work.
To clarify whether the heterostructure is successfully formed in the composite of g-C3N4/(FeTPP)2O, the optical properties of the above-mentioned materials were studied. Fig. 1a shows the UV-vis diffuse reflectance spectroscopy (DRS) of g-C3N4/(FeTPP)2O composites, pure g-C3N4 and (FeTPP)2O. The absorption peak of sole g-C3N4 is up to ∼450 nm, and has a small absorption tail which is probably due to n–π* transitions.44 The (FeTPP)2O has strong absorption in the visible region from 400 to 700 nm, mainly including two Q bands around 567 nm and 613 nm, which are similar to that of the (FeTPP)2O solution (Fig. S1†). The g-C3N4/(FeTPP)2O composites show the absorption features combining g-C3N4 and (FeTPP)2O, and exhibit wider absorption of solar light. As the content of g-C3N4 increased, the obvious red-shift of the Q band of (FeTPP)2O in composites can be found, which suggests π–π stacking interactions between g-C3N4 and (FeTPP)2O.45,46
Photoluminescence (PL) is an effective and commonly used method to investigate the electron transfer property of semiconductor materials.45,46Fig. 1b displays the PL spectra of g-C3N4, pure (FeTPP)2O and g-C3N4/(FeTPP)2O nanocomposites. Obviously, g-C3N4 has a strong and wide emission, in contrast, (FeTPP)2O has negligible PL signal response when the excited wavelength is 400 nm. The emission intensity of g-C3N4 remarkably decreased after it was combined with (FeTPP)2O. It means that the recombination of photo-excited holes and electrons of g-C3N4 is largely prohibited by (FeTPP)2O and that the charge transfer occurs in the composites. The big red shift of PL spectra of g-C3N4 in these composites also demonstrates the existence of interaction between g-C3N4 and (FeTPP)2O. On the basis of the above optical properties, it was guessed that their heterojunctions have been successfully prepared and the charge transfer may occur between them.
Wide-angle X-ray diffraction (XRD) patterns also confirm the presence of g-C3N4 in as-prepared samples. As shown in Fig. 1c, XRD patterns of g-C3N4 with different (FeTPP)2O contents exhibit a typical diffraction peak at 27.41°, which corresponds to the interlayer stacking of aromatic segments and indexed as (002). As compared to pure g-C3N4, no obvious (FeTPP)2O characteristic peaks can be found, indicating that g-C3N4/(FeTPP)2O retains the original crystal structure of g-C3N4. The same conclusion can also be made from BET experiments. The nitrogen adsorption–desorption isotherms of pure g-C3N4 and (FeTPP)2O-mediated g-C3N4 show type IV with hysteresis loops, indicating the presence of mesopores (Fig. 1d). The specific surface areas of pure g-C3N4 and g-C3N4/(FeTPP)2O (5.0 wt%) composites are 74 and 64 cm2 g−1, respectively. The inserted image in Fig. 1d indicates the similar mesopore structure between pure g-C3N4 and g-C3N4/(FeTPP)2O. These data suggest that the introduction of (FeTPP)2O into g-C3N4 lowers the specific surface areas of g-C3N4, however, it does not have any obvious effect on the mesopore structure of g-C3N4, which is closely related to its photocatalytic activity.
In order to further investigate the components and structure of g-C3N4/(FeTPP)2O composites, scanning electron microscopy (SEM, JSM-6700F, JEOL, Inc.) and transmission electron microscopy (TEM, F20, FEI) were used. The corresponding images are shown in Fig. 2. As compared to g-C3N4 shown in Fig. S4,† which has smooth folded nanosheets besides some fragments, (FeTPP)2O in Fig. S5† is composed of a lot of aggregations of amorphous nanoparticles (NPs) or naonoplates. After (FeTPP)2O was introduced into the as-obtained g-C3N4, many small NPs can be found on g-C3N4 sheets and quite a lot of aggregations are attached on g-C3N4, as shown in Fig. 2a and b. The TEM image in Fig. 2b exhibits that the g-C3N4/(FeTPP)2O heterostructure still retains the morphology of g-C3N4. Fig. 2c clearly displays a typical layer-by-layer structure of g-C3N4 according to the difference of image contrast, and the inserted selected area electron diffraction (SAED) image indicates a low crystallinity of g-C3N4 in the composite. A clue to the existence of (FeTPP)2O can be found in the TEM image shown in Fig. 2d. In order to further prove the existence of (FeTPP)2O, the corresponding energy dispersive X-ray spectroscopy (EDS) was also performed on the sample. The result in Fig. 2d shows that Fe–K peaks exist in the spectrum and their percent content is calculated to be ∼0.11 at%, which is close to the content in theory (5 wt%). At the same region of the sample, TEM-assisted elemental mappings were carried out and the results are shown in Fig. 2e. The Fe–K mapping in Fig. 2e further indicates that the NPs of (FeTPP)2O exist and they should be equably distributed on the surface of g-C3N4 nanosheets. The above results further demonstrate the successful formation of the g-C3N4/(FeTPP)2O heterostructure and the intimate contact between g-C3N4 and (FeTPP)2O. Interestingly, although at least two obvious NPs can be found in the orange square in Fig. 2e, as compared to Fig. 2d, only one (FeTPP)2O nanoparticle can be found in the Fe–K mapping. It means that most of the NPs or nanoplates should be fragments of g-C3N4 as a result of the reaction between g-C3N4 and the precursor of (FeTPP)2O.
To further understand the interaction of the two components, X-ray photoelectron spectroscopy (XPS) was performed on these samples. The binding energies of Fe 2p3/2 of (FeTPP)2O and g-C3N4/(FeTPP)2O are 711.5 eV and 711.1 eV (Fig. 3a), respectively, the decrease of 0.4 eV in the binding energy of Fe 2p3/2 might be caused by the interaction between (FeTPP)2O and g-C3N4, which indicates the formation of an Fe–N bond because the N element has lower electron affinity than O.47 The binding energy of Fe 2p1/2 also has the same tendency. To prove the interaction, N 1s XPS spectra of pure g-C3N4 and g-C3N4/(FeTPP)2O were also measured and compared, and the results are exhibited in Fig. 3b. Three peaks at ca. 398.3, 399.4 and 400.9 eV are observed for single g-C3N4. The main N 1s peak at 398.3 eV corresponds to the sp2 hybridized aromatic N bonded to carbon atoms (C–N–C). The peak at 399.4 eV is assigned to the tertiary N bonded to carbon atoms in the form of N–(C)3 or H–N–(C)2. And the weak peak with a high binding energy at 400.9 eV is attributed to the quaternary N bonded to three carbon atoms in the aromatic cycles.15,42,43 After forming the heterostructure with (FeTPP)2O, all the N 1s peaks of g-C3N4 shift to a higher binding energy, corresponding to the decrease of the Fe 2p binding energy. Apparently, a strong interaction between (FeTPP)2O and g-C3N4 is built through the formation of the Fe–N bond.48 As compared to the relative atomic concentration of Fe (1.29 at%) in the pure (FeTPP)2O composite shown in Fig. 3c, that of the g-C3N4/(FeTPP)2O composite is 0.21 atom% estimated from the XPS data. The result is higher than that from the EDS and our theoretical data because the XPS data are usually collected from the sample surface. The above XPS data indicate that in the as-obtained samples, (FeTPP)2O is on the surface of g-C3N4 and they form the heterostructure through π–π and the Fe–amine interactions.
Between g-C3N4 and (FeTPP)2O, the charge transfer has been revealed in the PL spectra, the SEM, TEM and XPS data have also demonstrated the formation of the heterostructure. However, only the relative band positions of the two components can tell us the direction of charge transfer and the type of the heterojunction. Based on the UV-vis diffuse reflectance data, the band-gap energy was calculated according to Fig. 4a and c. The band gap of g-C3N4 and (FeTPP)2O are 2.73 eV and 2.09 eV, respectively. The valence band XPS (VB XPS) spectra of g-C3N4 and (FeTPP)2O were measured, as shown in Fig. 4b and d, the valence band (VB) of (FeTPP)2O is more negative about 1.56 eV than that of g-C3N4. Together with their band gaps we can deduce that the conduction band (CB) of (FeTPP)2O is more negative about 0.96 eV than that of g-C3N4. As illustrated in Scheme 1, the band alignment of g-C3N4 and (FeTPP)2O forms the typical type II heterojunction.49,50 The large CB and VB offsets between g-C3N4 and (FeTPP)2O make the photo-generated electrons transfer from (FeTPP)2O to g-C3N4, while the photo-generated holes migrate from g-C3N4 to (FeTPP)2O in a thermodynamically favorable manner, which leads to an efficient charge separation and significant enhancement in photocatalytic activity.36,50
Solar-light-induced H2 production was investigated to examine the photocatalytic activity of the as-prepared samples. Fig. 5a shows the varied amount of H2 evolution under solar light for various photocatalysts. Almost no H2 can be detected when only g-C3N4 was used as the photocatalyst. It may be due to the detrimental electron-hole recombination and the Frenkel exciton effect. In contrast, the g-C3N4/(FeTPP)2O composite exhibited dramatic photocatalytic H2 production. Here, the loading amount of (FeTPP)2O plays an important role in tailoring the photocatalytic activity of the heterostructures. With the content of (FeTPP)2O increased, the amount of H2 first increases and then decreases. The experiments in Fig. 5 show that the optimal amount of (FeTPP)2O is about 5.0 wt%, and the best sample produces 59.2 μmol H2 under solar light irradiation for 4 h. As we know, for the photocatalytic activity of the heterostructure nanocomposites, two factors have a serious effect on the hydrogen evolution. One is the capability of light absorption, the other is the amount of activity sites on the surface of the composite. In our system, the (FeTPP)2O plays a dual role of dye-sensitization and charge transfer for g-C3N4. In some ranges, more doping of the guest means the formation of more heterojunctions (activity sites) and stronger absorption in visible-light. With the further increase of doping beyond some amount, however, surface activity sites and/or absorption actually decrease. Therefore, there is always an optimal doping ratio. As far as the g-C3N4/(FeTPP)2O composite is concerned, the decrease of its specific surface areas with the increase of the amount of (FeTPP)2O (see Fig. 1d) means that the optimal ratio of (FeTPP)2O in the heterostructure is determined by its visible-light absorption. UV-vis DRS in Fig. 1a clearly shows that the ratio should be ∼5.0 wt%, which is consistent with our experimental results in Fig. 5.
Furthermore, it was found that sacrificial electron donors are also essential for the consumption of photo-generated holes and the regeneration of (FeTPP)2O. As shown in Fig. 5b, without any electron donors, no H2 is produced. The presence of ascorbic acid (AA) also has little effect on promoting the activity of g-C3N4/(FeTPP)2O heterostructures, while the addition of triethanolamine (TEOA) significantly enhances the photocatalytic H2 production. The reason must be ascribed to the highest redox potential and the created basic circumstance of TEOA.51 Apparently, sole (FeTPP)2O shows no activity toward water splitting, as shown in Fig. 5c. It must be noted that the physical mixture of g-C3N4 and (FeTPP)2O failed to produce the heterostructure, thus the mixture also shows negligible photocatalytic activity.52 In addition, commercial (FeTPP)2O was also thought to combine with g-C3N4 in a physical manner. Through impregnation of g-C3N4 with a dichloromethane solution of commercial (FeTPP)2O overnight, the composite of g-C3N4/(FeTPP)2O was prepared. However, Fig. 5d shows that the corresponding composite also exhibits negligible H2 production. Therefore, the formation the Fe–N bond between g-C3N4 and (FeTPP)2O should play a key role in the solar H2 production from water splitting.
Hydrogen evolution measurement under UV-vis light irradiation with a 300 W Xenon lamp was also investigated. The durability of g-C3N4/(FeTPP)2O (5.0 wt%) acting as the photocatalyst for H2 evolution was evaluated by three consecutive operations. As shown in Fig. S6,† UV-vis-light can stimulate the photocatalyst to produce H2 at a rate of ∼40 μmol h−1 and the amount of H2 increases with irradiation time. Followed by the first run, some deactivation was observed in the second run. Interestingly, after the second cycle and lighting off for several hours, some photocatalytic activity was recovered in the third cycle. The deactivation should have occurred due to the non-renewable consumption of the (FeTPP)2O in the photocatalytic process,47 while the recovery is ascribed to the partial regeneration of (FeTPP)2O with TEOA in the dark. In addition, under visible light (λ > 420 nm) generated by using an Xenon lamp with a UV-cutoff filter, H2 evolution is 11 μmol h−1 (see Fig. S7†). At 420 nm, a quantum efficiency (QE) of 0.0415% was obtained.53
FeTPPCl + AgClO4 = AgCl + FeTPP+·ClO4+ | (1) |
In a typical synthesis, a solution of dry silver perchlorate in a minimum amount of anhydrous acetonitrile (0.6 mL, 0.30 mM) was added into a 16 mL solution of 0.27 mM FeTPPCl in anhydrous dichloromethane with stirring. After stirring for about 3 h, the mixture was separated by centrifuging at 10000 rpm for 10 min. Relying on the difference in solubility between silver chloride and FeTPPClO4 in acetonitrile, a brown FeTPPClO4 solution is separated from the admixture. The solution was poured into 50 mL n-hexane and placed in a fridge at −22 °C for 24 h. Then the precipitation was obtained through centrifuging at 10000 rpm for 2 min and further washed with n-hexane two times. The as-obtained product was dried for further use.
Footnote |
† Electronic supplementary information (ESI) available. See DOI: 10.1039/c5ta07278f |
This journal is © The Royal Society of Chemistry 2016 |