journal article·2015-08Open access

The Complex Role of Carbon Nitride as a Sensitizer in Photoelectrochemical Cells

Jingsan Xu, Isaac Herraiz‐Cardona, Xiaofei Yang, Sixto Gimenez, Markus Antonietti, Menny Shalom

Advanced Optical Materials· Volume 3· Issue 8

Abstract

The role of carbon nitride (C3N4) as an absorber in a photoelectrochemical cell is reported. C3N4-sensitized TiO2 mesoporous film created via in-situ, vapor-transport growth results in a direct Ti–O–C bonding. The material hybridization shows a unique electronic transition at the interface, leading to strongly enhanced visible-light absorption and photoactivity. In the last years, enormous attention has been focused on the metal-free semiconductor graphitic carbon nitride (simplified as C3N4) with a bandgap of 2.7 eV, which holds a great promise in the fields of electrocatalysis,1 bioimaging,2 solar cells,3 and especially photocatalysis.4, 5 Many approaches were introduced to enhance the photoactivity of C3N4, such as doping6 with different heteroatoms and by surface area increase, usually by hard-templating method.5 Recently, an easy, safe, and highly effective approach has been developed to promote the photocatalytic activity of C3N4, which employs supramolecular precursors for thermal polymerization into C3N4. The supramolecular complex is generated by combining different organic compounds (triazine derivatives) in solvents and then assemblies are formed because of noncovalent interactions such as hydrogen bonds. Typically, cyanuric acid-melamine system has been shown to yield preorganized micro/nanostructures and morphologies, and as a result enables the modification of optical and electrical properties of final C3N4 product and significant improvement of photoactivity.7 Moreover, the final material composition can be tuned by the insertion of other molecules (e.g., barbituric acid) into the starting complex. As stated before, C3N4 features good stability over high temperature and corrosive chemical environments.8 C3N4 has been frequently used in combination with other semiconductor materials, such as MoS2,9 In2O3,10 and Ag3PO4,11 for better light harvesting. When the chemistry is successfully executed, heterojunctions are formed between C3N4 and the other component, which improve the charge separation process and the overall performance. However, while the high activity of C3N4 as photocatalyst is well understood and studied, its performance in photoelectrochemical cells (PEC) remained low due to a more complicate charge separation/transport process. In order to promote PEC performance, uniform, continuous, and well-attached thin film electrodes should be produced. Furthermore, it is crucial to better understand the photophysical properties and the chemical interactions of carbon nitride materials in such systems. Herein, we carefully studied the chemical interactions alongside the photophysical properties of TiO2/C3N4 in a model photoelectrochemical cell. A TiO2/C3N4 hybrid thin film was prepared by an in situ vapor-transport growth mode, using cyanuric acid-melamine (CM) or cyanuric acid-melamine-barbituric acid (CMB) supramolecular complexes as the precursor (labeled here as TiO2/CM and TiO2/CMB, respectively). The hybrid materials were characterized using a range of techniques (summarized in Table S1, Supporting Information). After thermal polymerization, the 20-nm TiO2 nanoparticles were uniformly coated with 3–5 nm of C3N4 layers. The TiO2/C3N4 system showed unique electronic coupling, leading to a greatly modified electronic structure and enhanced optical absorption. Consequently, the visible-light driving photocurrent was raised by ≈30 times for TiO2/C3N4 as compared with a comparable pristine TiO2 cell. The material growth approach is developed on the basis of previous reports on improving photoactivities of C3N412 with certain modifications, as schematically shown in Figure 1. First, mesoporous TiO2 thin film on fluorine-doped tin oxide (FTO) glass was prepared by a doctor-blade method using commercial paste followed by air-annealing. Second, CM or CMB precursor was placed on top of the substrate, fully covering the TiO2 film. Then the system was put in crucible, capped, and heated in nitrogen (detailed information is given in the Supporting Information). The X-ray diffraction (XRD) patterns of the substrates are shown in Figure S1 (Supporting Information). Compared with bare FTO, the additional diffraction peaks confirmed the formation of TiO2 film, which was mainly composed of the anatase phase. After the growth of C3N4 layers, the TiO2 phase remained unchanged and no impurity peaks from, e.g., TiN or TiC were detected. It is noticeable that the diffraction patterns of C3N4 layer cannot be resolved because of its thinness (as shown in Figure 2) and the high crystallinity of TiO2 and FTO. However, the Fourier transform infrared spectroscopy spectra (Figure S1b, Supporting Information) of TiO2/CM and TiO2/CMB clearly demonstrated the vibration peaks at 1200–1650 cm−1, which correspond to the typical stretching modes of C–N heterocycles that implied the C3N4 formation. The invisibility of triazine peak at ≈810 cm−1 was due to the overlapping with the broad strong absorption band of TiO2 located in the range of 550–800 cm−1. Further support was given by the elemental analysis results, showing C/N molar ratio of 0.71 for TiO2/CM, which is a typical value for normal C3N4 samples.13 The ratio increased to 0.90 for TiO2/CMB, due to the incorporation of barbituric acid molecules into the precursor. The Raman spectra (Figure S1c, Supporting Information) shows the bands at 193, 392, 512, and 635 cm−1 corresponding to the characteristic Eg(1), B1g(1), A1g + B1g(2) doublet, and Eg(2) mode of anatase TiO2, respectively. The Eg(1) and B1g(1) modes are O–Ti–O bending vibrations while A1g + B1g(2) and Eg(2) are Ti–O stretching vibrations.14 The spectra demonstrated that after C3N4 growth the intensity of Eg(1) mode was significantly increased and the peak showed slight blue shift, meaning the enhancement of the bending vibration. On the contrary, A1g and Eg(2) modes were broadened and shifted to lower wavenumbers, suggesting the increasing of lattice disorder and weakening of the stretching vibrations. The structure evolution of TiO2 indicated the strong interaction between TiO2 and C3N4, as discussed in more detail later in this manuscript. Atomic force microscopy was employed to characterize the surface of the TiO2 and TiO2/C3N4 substrates (Figure S2, Supporting Information). The morphology and roughness of the substrates were not changed after thermal polymerization (with CM or CMB), e.g., no generation of disordered layers on top was found as reported before.12, 15 The cross-section scanning electron microscopy images also confirmed the absence of extra top layer (Figure S3a, Supporting Information) and only the mesoporous structure of the primary TiO2 film was observed (Figure S3b,c, Supporting Information). On the contrary, it is very interesting to find out that the TiO2 nanoparticles were uniformly coated by a thin C3N4 layer (3–5 nm), as illustrated by high-resolution transmission electron microscopy (HRTEM) images in Figure 2a. Further evidence of the successful C3N4 deposition is given by the electron energy loss spectroscopy (EELS) spectra (Figure 2b). Compared with pristine TiO2, TiO2/CM and TiO2/CMB show the transitions of 1s → π* and 1s → σ* in both carbon K edge and nitrogen K edge. More importantly, the relative intensities of the two transitions (the 1s → π* transition peaks are very sharp and intense for TiO2/CM and TiO2/CMB) demonstrated that the outer layers of the TiO2/C3N4 nanoparticles exclusively consisted of sp2 hybridized carbon and nitrogen atoms,16 which further proves the thin layers are composed of carbon nitride. Corresponding selected area electron diffraction patterns illustrates that the C3N4 does not have obvious impact on the crystallinity of the TiO2 nanoparticles (Figure S4, Supporting Information). It is worth noting that we randomly studied ≈10 different spots and we found that all the TiO2 nanoparticles were coated with C3N4 (Figure S5, Supporting Information). In addition, line-scanning energy-dispersive X-ray spectroscopy shows that the nitrogen and carbon contents remain constant across the TiO­2/C3N4 film (Figure S3d, Supporting Information). Therefore, it is reasonable to assume that the growth of C3N4 took place all across the TiO2 film. Besides, we found that the CM complex (with and without the TiO2) did not show a liquid-phase intermediate up to 500 °C (Figure S6, Supporting Information), such that no liquid would penetrate into the TiO2 thin film during the growth. Given these facts as well as the thickness (≈7 μm) and the mesoporous structure of the TiO2 film, we can suppose that during the heating process an in situ vapor-transport growth of the C3N4 layers occurred. Specifically, sublimation of the CM precursor would be triggered at elevated temperature in the sealed crucible (around 430 °C based on TG-DSC curves, Figure S6b, Supporting Information) and the resulted vapor precursor can transport downward to the TiO2 substrate driven by concentration gradient. This process is followed by redeposition at the TiO2 nanoparticles driven by surface energy minimization, as schemed in Figure 1b. Afterwards, thermal polymerization of the precursor towards C3N4 occurred on the surface of the TiO2 nanoparticles with further temperature increase (Figure 1c) and the phase transformations of CM precursor were monitored by XRD measurement (Figure S6d–g, Supporting Information). The growth could have been terminated by several reasons: (1) the oxide surfaces are preferentially coated by C3N4 due to surface energy effects; this effect levels off at about 5-nm thickness; (2) the transport of the vapor as well as the thickening of C3N4 layer is slowed down or even prohibited because of space limitations. The chemical states of the material systems were investigated by X-ray photoelectron spectroscopy (XPS), (Figure 3a,b). The pristine TiO2 substrate showed typical binding energies of 458.4 and 529.6 eV for Ti 2p3/2 and O 1s. After carbon nitride growth the Ti 2p3/2 and O 1s peaks shifted to higher binding energies (458.8 and 530.0 eV, respectively). Significantly, the binding energy of Ti–O–C bonding appears at 532.8 eV in the O 1s spectrum (Figure 3b). In addition, the C 1s spectrum also clearly shows the C–O bonding at 286.7 eV (Figure S7a, Supporting Information), implying the formation of direct electronic coupling between TiO2 and C3N4, which is responsible for the positive shifts of the Ti 2p3/2 and the O 1s peaks because C3N4 acts apparently as a strong electron withdrawing group on the TiO2 surface. Interestingly, this kind of chemical interaction between TiO2 and C3N4 has not been reported previously although a few TiO2/C3N4 composites were prepared,17, 18 indicating the peculiarity of the in situ vapor transport based growth mode of C3N4 on mesoporous TiO2. In addition, the shift of Ti 2p3/2 peak to higher binding energies implies that nitrogen doping of TiO2 during the synthesis could be excluded. Moreover, no Ti–C or Ti–N coordination schemes were detected. We note that free C3N4 powders made from CM complex did not show any C–O or C=O bonds (Figure S8, Supporting Information), indicating that the generation of C–O bonding is due to the chemical coupling between TiO2 and C3N4 (Figure 3c). The N 1s XPS spectra are shown in Figure S7a (Supporting Information) and the spectrum of TiO2/CM was fitted into several binding energies (Figure S7c, Supporting Information). The main peak at 398.7 and 399.8 eV can be attributed to the C–N–C coordination and tertiary nitrogen N–(C)3 groups, respectively.13 The peak at 400.8 eV resulted from the bonding of C–N–H. The C 1s spectrum is shown in Figure S7b (Supporting Information) (fitted curve in Figure S7d, Supporting Information), demonstrating the occurrence of C–N–C group with binding energy at 288.3 eV. The N and C spectrum corresponds well with that of typical C3N4 samples. The UV–vis absorption spectra are shown in Figure 3d. The pristine TiO2 substrate showed a clear absorption edge at 400 nm, corresponding to the bandgap of 3.1 eV of anatase. After thermal polymerization with the supramolecular precursors, the absorption of TiO2/CM was strongly enhanced in the visible-light range, and the edge shifted to ≈600 nm and beyond for TiO2/CMB (see also Figure S9, Supporting Information). It is important to note that corresponding to the XPS measurements, the optical absorption is fundamentally different from that of C3N4 made from pure CM and CMB, N-doped or C-decorated TiO219 and conventional TiO2/C3N4 composites. Given the promotion of light absorption, we investigated the photocurrents (Figure 4a) of the substrates under visible light illumination (λ > 410 nm) using a typical three-electrode setup in Na2S (0.1 m) aqueous solution without applying bias. As expected, the pristine TiO2 showed low photocurrent (0.045 mA cm–2) because of its poor visible-light absorption. The photoactivity was significantly enhanced for TiO2/CM and the photocurrent reached 1.4 mA cm–2, mostly related to the improved light absorption and to the recombination inhibition as will show later in the manuscript. The photocurrent decreased to 0.7 mA cm–2 for TiO2/CMB although it showed further enhanced light absorbance. The decrease of the photocurrent is probably due to higher defects levels (because of a too high carbon doping) which can result in higher recombination of charge carriers. In order to further study the photoactivity of carbon nitride, we measured the photocurrent versus the light intensity (Figure S10, Supporting Information). The photocurrent profile showed a typical solar cell behavior and can be well fitted into a linear model for both TiO2/CM and TiO2/CMB. Furthermore, we note that the photocurrent profiles can be divided into two parts, namely the current increased more quickly at low light intensity (<2000 μmol m−2 s−1) than at higher light intensity (up to 20000 μmol m−2 s−1). This can be understood due to the enhanced recombination of the photo­generated charge carriers at high illumination intensities.20 The TiO2/C3N4 substrates also illustrated considerable photocurrent under blue light (465 nm), i.e., 0.65 and 0.24 mA cm–2 for TiO2/CM and TiO2/CMB (Figure S11, Supporting Information), while pristine TiO2 did not show any noticeable photoresponse. Besides, the hydrophilicity of TiO2 substrate was checked before and after C3N4 sensitization and there was no obvious change based on contact angle measurement (Figure S12, Supporting Information), indicating the improvement of photocurrent should be ascribed to modified electronic/optical properties. Combining with the chemical coupling between TiO2 and C3N4 as revealed by XPS alongside the unique absorption spectra, we suppose that a novel, interface bound energy transition was created in this system. Normally speaking, upon absorption of light with energy higher than 2.7 eV (bandgap of C3N4), electrons in the ground state (HOMO) of C3N4 are excited to the conduction band and are relaxing to the lowest unoccupied molecular orbital (LUMO). From there, they usually transfer to the TiO2 conduction band (CB) due to the difference in energy (path 1 and 2, Figure 4b).18, 21 The thermodynamically favorable electron injection is thereby driving charge separation. Subsequently, the negative charges transport through the TiO2 nanoparticles and reach the conducting transparent electrode, while holes transfer to the electrolyte. In the present case, however, we suggest that the strong electronic coupling enables hybridization between the two semiconductors, and new molecular orbitals are generated at the interface (Figure S13, Supporting Information), according to the molecular orbital theory.22 We note that it differs from the conventional heterojunction due to the strong chemical coupling. Consequently, photoinduced charge transition can take place between the new levels (blue arrow), which have significantly narrower energy gap and thus results in optical absorption far beyond the bandgap of either C3N4 or TiO2. This TiO2/C3N4 can be considered as a joint electronic system and behaves in the way of charge-transfer complex. Similar phenomenon was previously observed in carbon@TiO2 nanocomposite, wherein a “dyade” structure was suggested to describe the properties.23 Furthermore, the electronic coupling can enable an alternative charge transfer (under illumination) from C3N4 HOMO directly into the unoccupied states in TiO2 CB24 (path 3, Figure 4b). This kind of transfer has also been demonstrated between TiO2 and other sensitizers such as catechol.25 However, one of the key issues arising from this transition is the fast recombination at the interface of TiO2/C3N4, induced by the electron transition back from the metal oxide to the C3N4 (path 4, Figure 4b) which would severely diminish the photocurrent. This can explain the relatively low photocurrent obtained from TiO2/C3N4 system compared with standard sensitized solar cells (≈10 mA cm–2),25, 26 although the absorption was strongly extended to the visible light range. Moreover, the obtained photovoltage (Figure S14, Supporting Information) in the C3N4/TiO2 cell is relatively low (≈170 mV). The photovoltage in PEC is the energy gap between the TiO2 Fermi level (electrons) and the chemical potential of the electrolyte (holes) and for the normal TiO2 PEC cell in Na2S electrolyte the photovoltage is ≈300 mV.26 The decreased photovoltage can be explained by (1) the downshift of the TiO2 energy level with respect to the electrolyte (2) the high recombination rate limits the electron concentration in the TiO2 (3) the negative potential of the Na2S electrolyte. A deeper analysis of the photoelectrochemical behavior of the TiO2/C3N4 heterostructures was carried out by means of cyclic voltammetry and impedance spectroscopy. Figure 4c shows the cyclic voltammetry curves obtained for both CM and CMB-modified TiO2 substrates and the comparison with pristine TiO2 electrodes. These curves show the classical capacitive behavior of TiO2 under cathodic bias related to the increased density of states as we approach the conduction band of the material. In addition, the reduction peak (at –0.6 V vs Ag/AgCl for pristine TiO2), which reflects electron transfer from the TiO2 CB to localized surface/trap state was clearly shifted for the C3N4-modified TiO2 substrates to more positive potentials, underlining the existence of new electron transfer mechanisms in the dyadic material.27 In order to characterize this shift further, electrochemical impedance spectroscopy (EIS) in dark was carried out. Figure S15 (Supporting Information) shows an example of the complex plane plots recorded on the tested samples. The obtained impedance spectra were fitted by using the well-known models to liquid The model employed in the present study is shown as Figure (Supporting Information). The chemical for TiO2, is in Figure as a of the potential and the density of states of TiO2 the conduction The of the versus the potential in the is for all the which that the sensitization with C3N4 does not significantly the density of states of TiO2. However, a clear shift of about is observed for of the C3N4-sensitized TiO2 electrodes. The shift of at the can be attributed to (1) downshift of the energy bands (Figure Supporting Information) with respect to the electrolyte due to the of the electronic states in TiO2 states are to the or (2) of the TiO2 density of states (Figure Supporting Information). However, based on the of versus we can the band further the strong electronic interactions between the Figure shows the recombination of electrons in the conduction band of TiO2 with in the electrolyte note that in the TiO2/C3N4 recombination cannot be obtained by this In order to the different we have the to a conduction in order to this on the basis of a density of electrons the between the electron Fermi level and TiO2 on the density of electrons in the TiO2 The lowest are obtained for pristine TiO2 and the for the TiO2/CM the lowest recombination indicating that C3N4 obtained with the CM precursor the charge recombination from the TiO2 to the electrolyte. However, the for TiO2/CMB is lower compared with TiO2/CM, for a of and approaches that of pristine TiO2 at the tested in the present which well the lower photocurrent found and showed in Figure In a study of carbon nitride as a potential for TiO2 electrodes with a structure for improved visible-light activity is reported. The synthesis carried out by in vapor transport growth of C3N4 the nanoparticles of TiO2 mesoporous thin film a strong chemical coupling, indicated by the generation of a direct Ti–O–C This bonding of additional molecular orbitals at the interface, in a strong of the optical absorption. The chemical and photophysical properties of the new hybrid materials were further investigated by that C3N4 acts as a surface structure the energy levels over the TiO2 surface. We strongly that this new to more and electrochemical cells by the electron between the organic semiconductor and the metal oxide substrate alongside the Moreover, the on the photophysical properties and the chemical interactions the to of hybrid such as nitrogen carbon at metal for As a to and this information by the materials are and be for are not or support issues arising from information than should be to the The is not responsible for the or of any information by the than should be to the corresponding for the
Source: openalex_inverted_index · Quality: baseline

Research context

OpenAlex topics: Advanced Photocatalysis Techniques · 2D Materials and Applications · Ga2O3 and related materials

SciVX discovery tags are terminology-based. OpenAlex topics are source-provided classifications.

Affiliations

Department of Colloid Chemistry Max Planck Institute of Colloids and Interfaces Potsdam 14424 Germany

Departament de Física Photovoltaics and Optoelectronic Devices Group Universitat Jaume I Castelló 12071 Spain

Funding

Minerva Fellowship

References (43)

Show source-provided references
1. 10.1021/ja209206c10.1021/ja209206c ↗
2. 10.1038/ncomms478310.1038/ncomms4783 ↗
3. 10.1021/ja308249k10.1021/ja308249k ↗
4. 10.1021/ja508329c10.1021/ja508329c ↗
5. 10.1021/la900923z10.1021/la900923z ↗
6. 10.1021/jp911803z10.1021/jp911803z ↗
7. 10.1021/jp304169210.1021/jp3041692 ↗
8. 10.1021/ja809307s10.1021/ja809307s ↗
9. 10.1021/ja103798k10.1021/ja103798k ↗
10. 10.1002/adfm.20120373210.1002/adfm.201203732 ↗
11. 10.1021/ja402521s10.1021/ja402521s ↗
12. 10.1039/c2ee03479d10.1039/c2ee03479d ↗
13. 10.1002/adma.20130311610.1002/adma.201303116 ↗
14. 10.1016/j.apcatb.2013.10.02910.1016/j.apcatb.2013.10.029 ↗
15. 10.1021/ie501203610.1021/ie5012036 ↗
16. 10.1002/anie.20130941510.1002/anie.201309415 ↗
17. 10.1002/anie.20090388610.1002/anie.200903886 ↗
18. 10.1021/cm503258z10.1021/cm503258z ↗
19. 10.1039/C4DT01347F10.1039/C4DT01347F ↗
20. 10.1039/b800274f10.1039/b800274f ↗
21. 10.1002/jrs.125007060610.1002/jrs.1250070606 ↗
22. 10.1002/cssc.20110069110.1002/cssc.201100691 ↗
23. 10.1103/PhysRevB.57.R318510.1103/PhysRevB.57.R3185 ↗
24. 10.1039/c2cp42484c10.1039/c2cp42484c ↗
25. 10.1016/j.electacta.2012.03.10210.1016/j.electacta.2012.03.102 ↗
26. 10.1016/j.jcis.2013.11.07210.1016/j.jcis.2013.11.072 ↗
27. 10.1039/C3TA14047D10.1039/C3TA14047D ↗
28. 10.1021/ie402820v10.1021/ie402820v ↗
29. 10.1021/jp053547e10.1021/jp053547e ↗
30. 10.1002/adfm.20070147810.1002/adfm.200701478 ↗
31. 10.1038/srep0560310.1038/srep05603 ↗
32. 10.1021/am503674e10.1021/am503674e ↗
33. 10.1038/nphoton.2012.1110.1038/nphoton.2012.11 ↗
34. 10.1002/adma.20100066010.1002/adma.201000660 ↗
35. 10.1021/jp111949a10.1021/jp111949a ↗
36. 10.1021/jp972197w10.1021/jp972197w ↗
37. 10.1021/jp101097j10.1021/jp101097j ↗
38. 10.1021/cm802254u10.1021/cm802254u ↗
39. 10.1016/j.jelechem.2011.12.01910.1016/j.jelechem.2011.12.019 ↗
40. 10.1016/j.solmat.2004.07.01710.1016/j.solmat.2004.07.017 ↗
41. 10.1021/jp011942910.1021/jp0119429 ↗
42. 10.1021/nn101534y10.1021/nn101534y ↗
43. 10.1021/jz200011210.1021/jz2000112 ↗