With the rapid development of society, the energy shortage and environmental pollution have become two major problems in the 21st century [1-4]. Solar energy, which is inexhaustible, low cost and environment friendly, is considered to have the greatest potential to solve these two major problems [5-7]. Hence, a large number of researchers are devoted to converting solar energy into chemical energy and degradation of pollutants by utilizing the solar energy in various ways, such as solar cells [8-10] photocatalysis [11-20] photoelectrochemical catalysis [21-23] and so on. In recent years, photoelectrochemical catalytic materials have been widely concerned, including TiO2 [24-26], CdS [27, 28], WO3 [29], g-C3N4 [30-34]. Among them, Graphitic carbon nitride (g-C3N4) is one of the most extensively researched materials due to its appropriate energy band position and band gap, visible-light response, low cost and good chemical stability. Nevertheless, pure g-C3N4 possesses shortcomings of low specific surface area, a low quantum efficiency and rapid recombination of photo-exciting electron-hole pairs [35]. Therefore, lots of strategies have been utilized to enhance photoelectrochemical catalytic activity of g-C3N4, such as morphology adjustment [36], producing defects [37], noble metal loading [38, 39] and construction of heterojunctions [28, 40-43]. Among them, heterojunction construction is a very effective method because it can not only increase light absorption, but also can greatly reduce recombination of electron-hole pairs.
The one-dimensional nanomaterial arrays including nanotubes, nanowires and nanorods have been widely studied because of their large specific surface area and one dimensional electron transport path with low electron-hole recombination rate [27, 44]. Especially, metal oxide materials and metal nitrides are the most common. Among metal nitrides, TiN nanotubes arrays is a promising electrode material as a result of its superior electrical conductivity (about 4000‒55500 S/cm) [45], large specific surface area and chemical stability, which also make TiN nanotube array suitable for a conductive substrate to support other photocatalytic material. In the past several years, g-C3N4-based heterostructures have been widely researched because pure g-C3N4 has many advantages mentioned before, and coupling with other semiconductor materials can be easy to effectively separate electron-hole pairs to increase photocatalytic performance [46, 47]. Nevertheless, most studies of the g-C3N4-based heterojunctions are based on g-C3N4 powders. Few studies have been carried out to form g-C3N4-based heterojunctions on conductive nanotube substrates to take full advantage of the superiorities of one-dimensional nanomaterials. However, for pure g-C3N4, its photogenerated electron-hole pairs have high recombination rate and light absorption efficiency is relatively low. Constructing heterojunction with low bandgap semiconductors is one of the most promising methods to overcome these shortcomings. Recently, semiconductor quantum dots (QDs), such as CdS, PbS and CdSe, have attracted much attention due to their intense visible-light absorption and tunable band gap. Particularly, CdS QDs, an II-IV semiconductor material with narrow band gap, are usually applied to decorate other semiconductors with wide band gap to improve optical absorption, such as ZnO/CdS [27], TiO2/CdS [28] and WO3/CdS [48, 49]. As far as we know, however, few works have been done to modify g-C3N4 nanotube arrays with CdS QDs.
In this paper, we designed a novel TiN/C3N4/CdS nanotube arrays core/shell structure photoanode for advanced photoelectrochemical performance and degradation efficiency. Among them, one-dimensional extremely conductive substrate TiN nanotube arrays stemed from 800 ℃-annealed TiO2 nanotubes in NH3. g-C3N4, which mainly acts as the electron acceptor to transport and separate electrons, was prepared by liquid atomic layer deposition (LALD). CdS exists in the form of quantum dots in order to more effectively absorb visible light, which was obtained by successive ionic layer adsorption reaction (SILAR). Their morphologies were observed by scanning electron microscopy (SEM) and transmission electron microscopy (TEM) and phase composition were well characterized by powder X-ray diffraction (XRD), Raman spectra and X-ray photoelectron spectroscopy (XPS). The photoelectrochemical (PEC) performance was detected by transient photocurrent responses, linear sweep voltammetry (LSV), electrochemical impedance spectroscopy (EIS) and photoelectrochemical degradation.
TiN nanotube arrays were synthesized by a two-step process (Fig. 1(a)). Firstly, TiO2 nanotube arrays were prepared by electrochemical anodization of Ti foils. Briefly, Ti foils (99.99% purity, 15 mm × 30 mm) were chemically polished and then ultrasonically cleaned with acetone, ethanol and deionized water, respectively. The prepared Ti foils were subsequently anodized in ethylene glycol electrolyte containing 2 vol% deionized water and 0.3 wt% NH4F at 60 V for 30 min. At the end of the electrochemical anodization, the samples were transferred to ethanol for ultrasonic cleaning for 5 min to remove the surface coating and then were annealed in muffle furnace at 450 ℃ for 3 h with a heating rate of 2 ℃/min. Secondly, the prepared TiO2 samples were further calcined in ammonia (NH3) atmosphere at 800 ℃ for 2 h to obtain TiN nanotube arrays without collapse.
TiN/C3N4 nanotube arrays were synthesized by a LALD method (Fig. 1(a)) [50]. The TiN samples were immersed in the cyanamide aqueous solution with 25% mass concentration and keep it for 30 min. Then, the treated samples were placed on glass slides and dried naturally for 12 h. The samples were subsequently annealed at 550 ℃ for 3 h with a heating rate of 2 ℃/min in a tube furnace with N2 atmosphere. In order to measure the UV-vis absorption spectrum of g-C3N4, the FTO/g-C3N4 sample was prepared by depositing g-C3N4 on FTO in the same method as the TiN/C3N4 sample.
CdS QDs were deposited on TiN/C3N4 nanotube arrays by a simple SILAR method (Fig. 1(a)) [27]. The sulfide source was 0.2 M Na2S solution and its solvent consisted of methanol and deionized water with the volume ratio of 1:1. Cadmium source was Cd(NO3)2 ethanol solution (0.2 M). The TiN/C3N4 samples were first dipped in the Cd(NO3)2 solution for 30 s and were immediately rinsed with ethanol for 30 s. Then, the samples were soaked in Na2S solution for 30 s and were subsequently rinsed with methanol and deionized water (volume ratio of 1:1) for 30 s. The above process was termed as one deposition cycle CdS QDs. The samples were deposited 10, 20, 30 and 40 cycles, respectively, and were called as TiN/C3N4/CdS 10C, TiN/C3N4/CdS 20C, TiN/C3N4/CdS 30C and TiN/C3N4/CdS 40C, respectively. For comparison, the TiN samples deposited 30 cycles of CdS QDs (TiN/CdS 30C) was also synthesized by the same method. Owing to serious photo corrosion and electron-hole recombination of CdS QDs, ZnS protecting layer was also deposited to reduce carrier recombination. Similar to CdS QDs deposition, all TiN/C3N4/CdS samples were immersed to Zn(NO3)2 (0.1 M) solution and Na2S solution (0.1 M) for 2 min each, respectively. Then, the prepared samples after two deposition cycles of ZnS were annealed by a two-step low temperature thermal treatment method developed by our group before [27]. Namely, the samples were first annealed for 10 min at 150 ℃ and for another 10 min at 250 ℃ on a heating plate.
The surface morphologies of the as-prepared samples were characterized by field emission SEM (FESEM, Hitachi S4800) and TEM (JEM-1200EX). The composition of the samples was detected by XRD (Bruker D8) with Cu Kα radiation in an angle range of 10°‒80°. A Renishaw InVia micro-Raman spectrometer (excitation wavelength of 514 nm) was used to obtain Raman spectra. XPS was employed on Thermo Scientific Escalab 250Xi (America) using monochromatic Al Kα source. The optical absorption spectra were measured by UV-vis spectrophotometer (Persee, TU-1901) using BaSO4 as a reference. The incident phonon-to-current efficiency (IPCE) was evaluated at 0 V using a 300 W xenon lamp equipped with different monochrome filters. The Brunauer-Emmett-Teller (BET) surface areas were measured using a Micromeritics ASAP 2460 at 200 ℃. Photoluminescence spectra (PL) were recorded on a JY HORIBA FluoroLog-3 spectrophotometer. Ultraviolet photoelectron spectroscopy (UPS) were measured with a VG Scienta R4000 analyzer and a sample bias of ‒5 V was applied. ESR signals were recorded on a Bruker A300-10/12 spectrometer and radicals were trapped by 5, 5-dimethyl-1-pyrroline N-oxide (DMPO).
All the photoelectrochemical tests were recorded by an electrochemical workstation (Zennium Zahner, German) and carried out in a conventional three-electrode system with the samples as the working electrode, an Ag/AgCl electrode as the reference electrode and a platinum plate as the counter electrode. The electrolyte was the mixed solution of 0.2 M Na2SO3 and 0.1 M Na2S (volume ratio of 1:1). A xenon lamp (500 W) equipped with an AM1.5 filter (Aulight, China) was used as the light source (100 mW/cm2). The voltage scan rate of 10 mV/s was utilized in the LSV test. Transient photocurrent responses were tested at 0 V bias. EIS was performed under the above light source and the amplitude was 10 mV. Its frequency ranged from 1 Hz to 100 kHz. The photoelectrocatalytic activities of the as-prepared samples were evaluated by the degradation of methyl orange (MO) under visible-light illumination without stirring at 0.2 V bias. The degradation solution was a mixture of 10 ml 0.02 g/L MO and 30 ml above electrolyte (pH = 11) and the electrodes were kept in the degradation solution in dark for 30 minutes to obtain adsorption desorption equilibrium. The light source was a 500 W xenon lamp with a UV cutoff filter (λ > 420 nm) and the illumination area of the electrodes was 1.4 cm2. The concentration of MO solution was analyzed by UV–vis spectrophotometer.
Fig. 1(b) displays the XRD patterns of the TiO2, TiN, TiN/C3N4 and TiN/C3N4/CdS 30C samples. For TiO2 nanotube arrays, the peaks appear at 25.44°, 38.17°, 48.22°, 54.17°, 55.24°, 63.35° in XRD pattern corresponding to the anatase TiO2 (101), (004), (200), (105), and (211) planes (JCPDS 21-1272), respectively [21]. And peaks at 38.79°, 40.46°, 53.31°, 71.01°, 76.41° could be identified to peaks of the titanium (JCPDS 65-9622). After annealing of titanium dioxide at 800 ℃ in NH3, the new peaks appear at 37.17°, 43.25°, 62.82° (JCPDS 38-1420), which demonstrates that the anatase TiO2 crystal phase is transformed into TiN phase [51]. For the TiN/C3N4 sample, a weak peak at 27.6° could be found, which corresponds to the (002) crystal plane of g-C3N4 (JCPDS 87-1526) [50]. After SILAR, a clear peak at 26.49° can be observed, which corresponds to the (002) plane of hexagonal CdS, indicating that CdS is formed on the surface of the TiN/C3N4 sample [27]. Hence, the XRD results demonstrate that TiN/C3N4/CdS of the core-shell structure has been successfully synthesized.
Raman spectroscopy was further characterized to illustrate crystal phase of the samples (Fig. 1(c)). For the TiO2 sample, five Raman characteristic peaks at 638, 516, 395, 197, 144 cm‒1 can be seen, which could be attributed to Eg, A1g(B1g), B1g, Eg, Eg modes of anatase TiO2, respectively [52]. For the TiN sample, Raman peaks are very weak, and four broad weak peaks at 150‒250, 250‒400, 400‒500, 500‒650 cm‒1 can be found, which arises from first-order transverse acoustic (TA), longitudinal acoustic (LA), the second-order acoustic (2A) and transverse optical (TO) modes of TiN, respectively [53]. After LALD, there are two broad peaks around at 1340 and 1589 cm‒1, and they can be ascribed to D mode and G mode of g-C3N4, respectively [54]. After SILAR, the prepared sample exhibits two characteristic peaks at 303 and 606 cm‒1 corresponding to the first-order longitudinal-optical phonons (1LO) and the second-order longitudinal-optical phonons (2LO) of CdS, respectively [27]. Therefore, according to the results of the above Raman spectroscopy, the core-shell structure of TiN/C3N4/CdS is further confirmed. This is in good agreement with the XRD results.
Fig. 2 shows SEM images of the TiO2, TiN, TiN/C3N4 and TiN/C3N4/CdS 30C nanotube arrays samples. According to Fig. 2(a), highly ordered TiO2 nanotube arrays were successfully synthesized by anodization. The average inner diameter of nanotube arrays is about 80 nm and the tube length is about 2μm. Fig. 2(b) is the SEM image of the TiN sample, indicating that the nanotube structure remains unchanged without collapse after 800 ℃ annealing at NH3. As shown in Fig. 2(c), we can observe that the surface of nanotube becomes rough after deposition of g-C3N4, suggesting that g-C3N4 has been successfully deposited on TiN nanotube arrays. Fig. 2(d) displays the SEM image of TiN/C3N4/CdS 30C. It can be found that the average inner diameter of nanotube arrays is significantly smaller than before, revealing that CdS QDs is well deposited on TiN/C3N4. In addition, the specific surface areas of the TiN, TiN/C3N4 and TiN/C3N4/CdS 30C samples were studied by nitrogen adsorption-desorption isotherms as shown in Fig. S1. We can observe that their specific surface areas gradually decrease and are 0.48, 0.46 and 0.41 m2/g respectively, further demonstrating that g-C3N4 and CdS QDs were deposited on TiN nanotube arrays in turn. All their specific areas are relatively low due to the large mass of Ti foils.
To further investigate the microstructure and chemical composition of the samples, TEM images and corresponding EDS elemental mappings were measured as displayed in Fig. 3. The TEM image of ordered TiN/C3N4/CdS 30C nanotube arrays is clearly shown in Fig. 3(a). It can be seen that there is something attached to the inner walls of nanotubes, which is well consistent with SEM images in Fig. 2(d). This may be due to the deposition of g-C3N4 and CdS on nanotubes. In order to further discover specific compositions, the high-resolution TEM image was explored in Fig. 3b. The lattice spacings of 0.212 nm (or 0.213 nm), 0.335 and 0.337 nm corresponding to TiN (200) [55], g-C3N4 (002) (JCPDS 87-1526) and CdS (002) (JCPDS 75-1545) crystal planes respectively are found. In addition, it should be noted that HAADF-STEM image of the single nanotube of TiN/C3N4/CdS 30C is also well displayed in Fig. 3(c) and corresponding EDS elemental mappings of N Kα, Ti Kα, C Kα, Cd Lα and S Kα could be observed in Fig. 3(d)‒(h). It can be found that N, Ti, C, Cd and S elements exist in the nanotube and are very evenly distributed, further indicating that g-C3N4 and CdS are uniformly deposited on TiN, respectively.
To further affirm the surface chemical composition and the crystal phase, high-resolution XPS spectra were performed in Fig. 4. For TiN nanotube arrays, the Ti 2p and N1s core levels are measured as shown in Fig. 4(a)‒(b). The Ti 2p spectrum could be assigned to six peaks and exhibits Ti-N 2p3/2 binding energy of 455.6 eV, Ti-N-O 2p3/2 binding energy of 457.1 eV, Ti-O 2p3/2 binding energy of 458.7 eV, Ti-N 2p1/2 binding energy of 461.4 eV, Ti-N-O 2p1/2 binding energy of 463.1 eV and Ti-O 2p1/2 binding energy of 464.4 eV, respectively [45, 56]. The N1s spectrum could be divided into three peaks, which are the Ti-N peak of 396.1 eV, the Ti-N-O peak of 397.2 eV and the absorbed-N peak of 399.5 eV, respectively. These results show that the surface chemical states of the TiN sample are composed of Ti-N, Ti-N-O, Ti-O, indicating part of TiO2 and/or TiOxNy still exist in the inner of TiN nanotube arrays. However, XRD and Raman results can only detect TiN phase, which reveals that TiN is only formed on the surface of nanotube arrays. Fig. 4(c)‒(d) present high-resolution XPS spectra of C 1s and N 1s of the TiN/C3N4 sample. The C 1s spectrum is deconvolved into three peaks with the binding energies of 285.8, 286.4 and 288.4 eV assigned to the characteristic peaks of the graphitic carbon, C-OH and C-N=C bonds of g-C3N4, respectively. The N 1s spectrum of TiN/C3N4 nanotubes is resolved into four peaks at 396.1, 399.2, 400.3 and 402.2 eV, which are associated with Ti-N bond from TiN substrate, the sp2 hybrid nitrogen (C-N=C), tertiary nitrogen N-(C)3, (C)2-N-H related to structural defects and imperfect condensation [57]. For TiN/C3N4/CdS 30C nanotube arrays, XPS spectra of Cd 3d and S 2p can be observed in Fig. 4(e)‒(f). There are two characteristic peaks at binding energies of 404.9 and 411.7 eV in Fig. 4(e), which corresponds to the Cd 3d5/2 and Cd 3d3/2 states, respectively [58, 59]. Besides, S 2p spectrum in Fig. 4(f) is composed of S 2p3/2 (binding energy of 161.3 eV) and S 2p1/2 (binding energy of 162.6 eV), which further affirms that CdS QDs are formed and successfully covered on TiN/C3N4 nanotube arrays. Consequently, the high-resolution XPS spectra confirm the core-shell structure of TiN/C3N4/CdS.
Photoelectrochemical catalytic performance of the samples was characterized under simulated sunlight (AM1.5, 100 mW/cm2). Fig. 5(a) shows transient photocurrent responses of the TiN, TiN/C3N4, TiN/C3N4/CdS 10C, TiN/C3N4/CdS 20C, TiN/C3N4/CdS 30C and TiN/C3N4/CdS 40C samples under a potential of 0 V. We can find that the TiN sample does not respond to simulated solar light. Thus, TiN with excellent conductivity acts only as a conductive substrate to support other semiconductor materials. The transient photocurrent response of the TiN/C3N4 sample is only about 25 μA/cm2 and relatively small in comparison to that of the TiN/C3N4/CdS samples. However, to best of our knowledge, this photocurrent density is higher than that of most of the reported pure g-C3N4, which can be attributed to the good support of TiN nanotube arrays substrate [33, 38, 50, 60-63]. For the TiN/C3N4/CdS samples, we can find that their photocurrent densities have been greatly increased compared with the TiN/C3N4 sample, which can be ascribed to the intense visible-light absorption of CdS QDs. For comparison, transient photocurrent response of the TiN/CdS 30C sample without g-C3N4 deposition was also measured as shown in Fig. S2 and its photocurrent density is obvious less than that of TiN/C3N4/CdS 30C, which indicates that g-C3N4 plays an indispensable role in the core-shell structure. Besides, with the switching of light, the photocurrents rise and fall rapidly, illustrating that the samples have a quick light response. In order to further investigate the optimal CdS deposition cycles, the transient photocurrent responses of TiN/C3N4/CdS 10C, TiN/C3N4/CdS 20C, TiN/C3N4/CdS 30C and TiN/C3N4/CdS 40C were also measured as shown in Fig. 5(a). It can be discovered that the photocurrent density gradually enhances from 1.5 to 3 mA/cm2, as the deposition cycles increase from 10 to 30 cycles. In comparison to the photocatalytic performance of similar catalytic systems reported before, the TiN/C3N4/CdS 30C sample shows a superior normalized photocurrent of 30 mA/W as shown in Table S1. Nevertheless, when the deposition cycles continue to increase to 40 cycles, the photocurrent density declines to around 1.8 mA/cm2. This may be caused by the CdS exceeding the tunneling thickness limit, which is harmful to charge transfer across the photoelectrode/electrolyte interface [59, 13]. It can be proved by the EIS results in Fig. 5(b). Furthermore, it is worth noting that the photocurrent density of TiN/C3N4/CdS at 30 CdS deposition cycles is maximal, approximately 3 mA/cm2, which is about 120 times larger than that of the TiN/C3N4 sample and approximately twice as much as TiN/C3N4/CdS 10C. Additionally, stability tests of the TiN/C3N4/CdS 10C, TiN/C3N4/CdS 20C, TiN/C3N4/CdS 30C and TiN/C3N4/CdS 40C samples was carried out at 0 V for 7200 s under simulated sunlight as displayed in Fig. S3. There is no obvious change in current density, revealing their good stability.
LSV measurements of different CdS deposition cycles were carried out to further confirm the optimal deposition cycles. The dark current densities of the samples in Fig. S4 can be ignored because they are very low relative to photocurrents. As is shown in Fig. 5(c), the photocurrent densities of the samples are of the same order as the transient photocurrent responses plot. Still, the photocurrent density of TiN/C3N4 is relatively low and the photocurrent density of TiN/C3N4/CdS 30C is the largest. The onset potentials of the samples are almost at the same position (about -1.13 V), implying that the conduction band minimum of the samples is almost unchanged. In addition, the corresponding photoconversion efficiency plot calculated from LSV is displayed in Fig. 5(d). The maximal photoconversion efficiencies of TiN/C3N4, TiN/C3N4/CdS 10C, TiN/C3N4/CdS 20C, TiN/C3N4/CdS 30C and TiN/C3N4/CdS 40C are calculated and around 0.02%, 0.66%, 0.75%, 1.16% and 1.07%, respectively. Obviously, the photoconversion efficiency of TiN/C3N4/CdS 30C at -0.5 V reaches a maximum, about 1.16%, which is almost 58 times the efficiency obtained with the TiN/C3N4 sample and 1.76 times that obtained with the TiN/C3N4/CdS 10C sample. In order to further prove that the photoelectrochemical performance of TiN/C3N4/CdS 30C was greatly enhanced, the IPCE measurements were carried out at 0 V bias as shown in Fig. S5. The IPCE values were obtained according to the following equation:
where P is the monochromatic illumination power intensity (mW/cm2), λ is the light wavelength, and Iph is the photocurrent density (mA/cm2). Clearly, the IPCE values of TiN/C3N4/CdS 30C are greatly larger than that of TiN/C3N4 in the full wavelength range. Furthermore, with the decrease of wavelength, the IPCE value of TiN/C3N4/CdS 30C gradually increases and reaches a maximum at 420 nm, about 64.8%. When the wavelength continues to decrease, the IPCE starts to decrease. Consequently, the IPCE examination further supports the results of the linear sweep voltammograms. Furthermore, PL spectra of the TiN/C3N4 and TiN/C3N4/CdS 30C samples were performed to investigate the separation efficiency of photogenerated carriers in the photocatalytic system. As shown in Fig. S6, the TiN/C3N4 sample exhibits a strong emission peak at 440 nm. However, the emission peak intensity greatly decreases after deposition of CdS QDs, implying the effective separation of photogenerated carriers.
To further study the charge transfer characteristics of the samples, EIS tests were conducted as shown in Fig. 5(b). EIS results are displayed in the form of Nyquist plots at 0 V bias under AM1.5 simulated sunlight. The semicircle diameter reflects the charge transfer resistance (Rct) at the interfaces between the electrolyte and the electrodes. The smaller the semicircle diameter of the arc, the easier the charge carriers transfer is at the interfaces. Markedly, Rct follows the order of TiN > TiN/C3N4 > TiN/C3N4/CdS 10C > TiN/C3N4/CdS 40C > TiN/C3N4/CdS 20C > TiN/C3N4/CdS 30C. In other words, after g-C3N4 is deposited on TiN nanotube substrate, the interfacial carriers transfer impedance decreases, implying that electrons transfer faster at the g-C3N4/electrolyte interface than at the TiN/electrolyte interface. With further deposition of CdS QDs, the semicircle diameters of the TiN/C3N4/CdS samples are obviously reduced relative to the samples without CdS QDs deposition, demonstrating that Rct values of the TiN/C3N4/CdS samples further decrease, which is good for improving their photoelectrochemical properties. Particularly, TiN/C3N4/CdS 30C possesses the smallest charge transfer resistance. These results are well consistent with transient photocurrent responses in Fig. 5(a). In addition, we can see that the solution resistance (Rs, the intersection of curves and x-axis) values of all samples are small, which can be ascribed to highly conductive TiN substrate. It is worth noting that highly conductive TiN substrate can reduce the solution resistance, but its photoelectrochemical catalytic activity is quite low leading to its high charge transfer resistance value. Furthermore, it is noteworthy that even though the Rs values of the TiN/C3N4/CdS samples increase very slightly after deposition of CdS QDs in Fig. 5(b), only about 3 Ω, it hardly affects the photoelectrochemical catalytic performance.
The photoelectrocatalytic activities of TiN/C3N4, TiN/C3N4/CdS 10C, TiN/C3N4/CdS 20C, TiN/C3N4/CdS 30C and TiN/C3N4/CdS 40C were further probed with the degradation of methyl orange (MO) under visible-light illumination (λ > 420 nm) as displayed in Fig. 5(e)‒(f). For the blank sample without photocatalyst, methyl orange is hardly degraded under light irradiation. As is shown in Fig. 5(e), we can see that C/C0 values of the TiN/C3N4/CdS samples are noticeably smaller than that of TiN/C3N4, and TiN/C3N4/CdS 30C has the lowest value, demonstrating that g-C3N4 coupling with CdS QDs can greatly improve the degradation rate, especially for TiN/C3N4/CdS 30C, which may be due to the increasement of light absorption and effective separation of electron-hole pairs. Besides, corresponding degradation rates of MO of the samples are shown in Fig. 5(f). It can be found that the degradation rates of TiN/C3N4, TiN/C3N4/CdS 10C, TiN/C3N4/CdS 20C, TiN/C3N4/CdS 30C and TiN/C3N4/CdS 40C are 8.1%, 30.1%, 40.3%, 48.9% and 31.9%, respectively, after irradiation for 150 min. Their degradation process also follows the pseudo-first-order kinetics model, and their pseudo-first-order kinetics curves are displayed in Fig. S7. Their first-order rate constants ke could be calculated by the following equation [64]:
where C and C0 are the concentration of MO at illumination time t and 0, respectively. ke for TiN/C3N4, TiN/C3N4/CdS 10C, TiN/C3N4/CdS 20C, TiN/C3N4/CdS 30C and TiN/C3N4/CdS 40C were calculated to be 0.00063, 0.0023, 0.0034, 0.0044 and 0.0025 min‒1, respectively. This order is consistent with the results of transient photocurrent responses, the linear sweep voltammograms and EIS tests. In addition, in order to investigate the radical species of degradation, DMPO spin-trapping ESR spectra of the TiN/C3N4/CdS 30C sample were collected as shown in Figs. S10 and 11. The characteristic peaks of •O2‒ and •OH radicals are clearly displayed, revealing that both •O2‒ and •OH radicals are the active radical species to degrade MO [60, 65-67].
Since photoelectrocatalytic activities are determined by the separation, transfer and migration of photogenerated electrons and holes as well as the energy band positions of g-C3N4 and CdS, the possible photoelectrocatalytic mechanism is discussed as follows. Based on the UPS spectra in Fig. 6(a)-(b), the valence band positions (EVB) of g-C3N4 and CdS could be calculated to be 6.86 and 6.13 eV versus the vacuum level by subtracting the width of the He Ⅰ UPS from the excitation energy (21.2 eV) [68, 69]. Furthermore, the conduction band edge (ECB) can be obtained by formula ECB = EVB ‒ Eg, where Eg represents the band gap of semiconductor that can be obtained from the Tauc plots, (αhν)2 versus hν. Fig. 6(c) and (d) show the Tauc plots of the FTO/g-C3N4 sample and the TiN/C3N4/CdS 30C sample calculated from the UV-vis absorption spectrum in Figs. S8 and S9. The FTO/g-C3N4 sample is used here to measure the UV-vis absorption spectrum of g-C3N4. Thus, the band gap of g-C3N4 is 2.82 eV as shown in Fig. 6(c). According to the Tauc plot of the TiN/C3N4/CdS 30C sample, 2.27 eV of the band gap of CdS can be seen in Fig. 6(d). Therefore, ECB values of g-C3N4 and CdS were calculated as 4.04 and 3.86 eV, respectively. The final scheme diagram of the photocatalytic mechanism can be shown in Fig. 7. Under the condition of visible-light illumination, the photogenerated electrons on the valence bands of g-C3N4 and CdS are excited to their conduction bands, respectively. Since the VCB potential of CdS is more negative than that of g-C3N4, electrons excited to the CdS conduction band are spontaneously transferred to the VCB of g-C3N4, and then transferred to the external circuit thought TiN conductive substrate, and then reach the Pt counter electrode to react with the H+ in water to generate H2. While generating photogenerated electrons, the photogenerated holes are left in the valence band of g-C3N4 and CdS, respectively. Because the valence band of g-C3N4 is more positive than that of CdS, the holes of the g-C3N4 valence band can easily move to the valence band of CdS [70-72]. Consequently, the composite of g-C3N4 and CdS deposited on TiN substrate can significantly enhance the photoelectrocatalytic performance due to the effective separation of electron-hole pairs and broadening of the light absorption range.
In summary, a novel TiN/C3N4/CdS nanotube arrays core/shell structure was successfully synthesized by LALD and SILAR method. In this structure, the highly conductive TiN nanotube arrays, which does not respond to simulated solar light, only act as a conductive nanotube substrate. The one-dimensional nanotube arrays substrate possesses the large specific surface area and 1D electron transport path, which is conducive to increasing the area of illumination and reducing recombination of electron-hole pairs. Deposited g-C3N4 mainly acts as the electron accepter to transport and separate electrons. CdS QDs can respond to the visible light and is thus considered as a light absorber. The results show that the TiN/C3N4/CdS nanotube arrays structure can greatly improve the photoelectrocatalytic performance. Particularly, the optimized photocurrent density of TiN/C3N4/CdS is almost 120 times enhancement corresponding to TiN/C3N4 at 0 V bias under simulated sunlight. In addition, the photoelectrocatalytic degradation rate of methyl orange by the TiN/C3N4/CdS samples is also effectively promoted relative to that of TiN/C3N4. The performance improvement can be attributed to the expansion of the light absorption range and effective separation of electron-hole pairs after coupling with CdS QDs according to the analysis of photoelectrocatalytic mechanism. Our work provides a promising and flexible method for design of other advanced nanotube array core/shell structure heterojunctions for photoelectrochemical catalytic hydrogen production and organic pollutant degradation.
2020,
Vol. 41 Issue (10): 1645-1653 DOI: S1872-2067(19)63512-6 
PDF
2020,
Vol. 41 Issue (10): 1645-1653 DOI: S1872-2067(19)63512-6 PDF