There is a growing need to develop sustainable "green" energy sources for replacing the classical fossil fuels for protecting our natural environment and alleviating the problem of global energy crisis [1]. Hydrogen (H₂), with its high energy capacity and environmental friendliness, is a source of clean energy that is generated from renewable sources such as solar [2, 3]. Notably, photocatalytic (PC) H₂ generation through water splitting has received tremendous attention in recent decades [4]. However, many conventional PC materials (namely, metal sulphides and (oxy) nitrides such as TiO₂, Zn_0.8Cd_0.2S, Cu₂O, WO₃, Ga₂O₃, g-C₃N₄, ZnIn₂S₄, CdS, and ZnS) cannot satisfy the requirements of low toxicity, low cost, high stability, and remarkable PC activity for practical applications [5-13]. In this context, the development of new photocatalysts, micro/nano structure engineering, and surface/interface engineering of heterogeneous semiconductors by optimizing suitable material combinations has been identified as a viable approach to improving the H₂ production rates of functional compounds [14-21]. Hybridization of semiconductor photocatalysts with carbon nanotubes (CNTs) affords a powerful strategy of designing advanced PC materials [22, 23]. Particularly, incorporation of CNTs into heterogeneous photocatalysts can offer potential advantages in terms of improved PC performances. On one hand, the one-dimensional (1-D) conductive channels of CNTs provide pathways for rapid electron transfer, which decrease the recombination of the photoinduced electron-hole pairs [24]. On the other hand, CNTs with an elaborated 1-D morphology could serve as good templates to induce the growth of semiconductors with 1-D nanostructures [25, 26] and core-shell architectures [27, 28] that display enhanced charge transport and improved PC efficiency. Indeed, previous studies focusing on the coupling of CNTs and TiO₂ have documented prominent improvements in the PC performance of the hybrid over those of the individual constituents [29-31]. For example, the activity of core-shell CNTs/TiO₂ nanocomposites for the PC degradation of methylene blue increased 2-3 times upon hybridization with ~16 wt% of CNTs [32]. This considerable enhancement is attributed to the uniform coating of nanocrystal TiO₂ onto CNTs to form a core-shell structure with high-quality interfacial contacts between CNTs and TiO₂, which largely facilitate the separation of the electron-hole pairs and lead to a better PC performance. Furthermore, micro/nanocomposites that combine CNTs with other commonly used PC materials such as ZnO [33], ZnS [34], CdS [35], WO₃ [36], and C₃N₄ [37], have also been demonstrated as hybrids that reveal enhanced PC performances through similar mechanisms.

It is well known that β-SiC (cubic phase silicon carbide) is a potential photocatalyst that exhibits an appropriate band gap (E_g = 2.4 eV, E_CB = −0.9 V vs. NHE at room temperature) for PC H₂ generation [38]. Until now, there has been limited research on SiC photocatalysts. The main reason for this is the rapid recombination of the photoinduced electron-hole pairs in SiC photocatalysts [39, 40]. Therefore, various engineering strategies, including the formation of unique SiC nanostructures (e.g., quantum dots [41], nanoparticles [42], nanowires [43], and hollow spheres[44]), construction of heterostructures (e.g., SiC-TiO₂ [45], SiC-ZnS [46], SiC-MoS₂ [47], SnO₂-SiC [48, 49], and SiC-CdS [50]), hybridization of SiC with metal co-catalysts (e.g., SiC-Pt [43], SiC-IrO₂ [51]), and nanocarbon materials (e.g., SiC-graphene [52, 53]), have been employed to promote the performance and durability of SiC photocatalysts since the initial research on water splitting in 1990 [54]. Further investigations show that the SiC-graphene nanoheterojunction with intimate interfacial contacts between SiC and graphene exhibits enhanced photoactivities for water splitting owing to the improved charge separation resulting from the formation of Schottky-junction interfaces [55]. In our previous studies, we have demonstrated that the well-designed CNTs/SiC nanowire nanoheterostructures exhibit enhanced PC activity, compared with that of individual SiC nanowires [56]. Owing to vapor-liquid-solid catalytic growth, the CNTs/SiC nanoheterostructures are composed of straight SiC nanowires and crooked CNTs. In the present work, we demonstrate that CNTs can be uniformly covered with SiC nanoshells to form coaxial CNT@SiC core-shell nanotubes (C@SiCNTs) via an in situ vapor-solid reaction between silicon vapor and CNTs. Owing to the rational coaxial core-shell structure with tight Schottky junctions between CNTs and SiC, both the interfacial coupling and charge separation of C@SiCNT nanoheterostructures can be greatly improved. As a result, the metal-free C@SiCNTs exhibits significantly improved performance in PC generation of H₂ from pure water. The enhanced mechanisms of PC H₂ generation over C@SiCNTs are discussed in detail.

Sodium sulphide, hydrofluoric acid, sulfuric acid, and nitric acid were supplied by Sinopharm Chemical Reagent Co., Ltd. Micro-grade silicon powder (40–200 mesh) was provided by Aladdin. Multi-walled carbon nanotubes (CNTs; length, 5–15 µm; specific surface area, 40–70 m² g^–1; diameter, 60–100 nm; purity, > 97%) were purchased from Shenzhen Nanotech Port Co., Ltd.

In a typical process, the CNTs were first purified with a mixed concentrated H₂SO₄/HNO₃ solution (volume ratio, 3:1) at room temperature. Then, 0.6 g of the acid-purified CNTs and excess silicon powder were mixed and placed inside a horizontal furnace. Subsequently, the furnace was heated to 1320 ℃. Simultaneously, argon gas was made to flow at 250 mL/min at one end of the tubular furnace. The temperature was kept at 1320 ℃ for 2 h. Subsequently, the sample was cooled and washed with a mixed HF/HNO₃ diluted (aqueous) solution. Finally, the obtained C@SiCNT nanoheterostructures sample was filtered and dried for further use.

The SiCNTs were obtained by increasing the reaction time to 4 h and keeping the other parameters unchanged, so that the carbon nanotubes were almost completely consumed.

Reference SiC nanocrystals were synthesized by annealing the C@SiCNTs at 750 ℃ in air for 2 h, which consumed the CNTs by oxidation.

The crystalline structures of the specimens were characterized by X-ray diffraction (XRD) (Rigaku, Cu K_α radiation, λ = 0.15418 nm) in the scanning range 10°–80°. The N₂ adsorption-desorption isotherm was determined at 77 K by using a surface area analyzer (Micromeritics Co., Gemini-2360). Then, Brunauer-Emmett-Teller (BET) method was used to determine the specific surface areas. X-ray photoelectron spectroscopy (XPS) was carried out with a VGESCALAB250 surface analysis system. The morphologies, structures, and elemental distributions of the products were examined by a scanning electron microscopy (SEM) (FEI Quanta 200 FEG) and a high-resolution transmission electron microscopy (TEM) (FEI Glacios Cryo-TEM, acceleration voltage: 200 kV). The carbon content was analyzed by a thermogravimetric analyzer under air flow. The UV-vis diffuse reflectance spectra were recorded by using a spectrophotometer (Shimadzu, model 2501 PC). The photoluminescence (PL) spectra were obtained with the help of a PerkinElmer fluorescence spectrophotometer at the excitation wavelength of 290 nm. The transient PL spectra were recorded by a fluorescence lifetime spectrophotometer (Edinburgh Instruments, UK FLS920) at the excitation wavelength of 290 nm.

80 µL of 0.5 wt% Nafion solution and 10 mg of the photocatalyst powders were added to 2.0 ml ethanol, and then, ultrasonic treatment was performed for 30 min to obtain a suspension. Then, 400 µL of the suspension was dropped onto a 3 × 2 cm² FTO glass substrate. Subsequently, the FTO glass substrate was dried and annealed at 150 ℃ for 1 h in argon flow. The transient photocurrent properties of the sample at 0.2 V bias were measured by an electrochemical analyzer by using a three-electrode system. A platinum electrode and the Ag/AgCl electrode (with saturated KCl) were used as the counter electrode and reference electrode, respectively. The light source was a 300 W Xe lamp (with AM 1.5 simulated sunlight cut-off filter). The electrolyte was 0.5 mol/L Na₂SO₄ solution. The Mott-Schottky plots were recorded at an ac amplitude of 5 mV with a frequency of 1000 Hz under dark condition. The electrochemical impedance spectra of the samples were obtained under the condition that the ac amplitude was 5 mV and the frequency range 0.01–10⁵ Hz.

Electrocatalytic H₂ evolution was also carried out by using the three-electrode cell. The polarization curves were recorded at 5 mV s^–1 scan rate. To convert the measured potentials (versus Ag/AgCl) to the RHE scale, Eq. (1) was employed.

(1)

A 300 W Xe lamp equipped with a simulated sunlight cut-off filter (AM 1.5) was used as the light source. The PC H₂ evolution was observed in a 200 mL Pyrex flask (three-necked, flat-bottom) at room temperature and pressure. The flask was sealed with a silicon rubber ring. The distance from the reactor to the light source was 20 cm. In a typical PC H₂ evolution experiment, 15 mg of the photocatalyst was added to 100 mL of aqueous Na₂S (0.1 mol/L) solution, then ultrasonic treatment was carried out for 30 min to form a suspension in the flask reactor. Nitrogen was bubbled into the suspension for 30 min to remove oxygen. 400 µL of the gas was extracted and analyzed by using a gas chromatograph (GC-7900, TCD, argon atmosphere) after 1 h of irradiation.

A cyclic experiment was carried out to determine the PC stability of C@SiCNTs. After 5 h of irradiation, the H₂ produced was substituted by N₂, which was followed by another 5 h of irradiation. According to the practical H₂ generation under 420 nm illumination, the apparent quantum efficiencies (AQEs) were calculated to determine the solar energy conversion abilities of C@SiCNTs, SiCNTs, and the reference SiC by using the following equation [57].

(2)

where N_H2 is the number of molecules of H₂ produced, λ is the wavelength of the irradiated monochromatic light, E_λ and P_λ are the average photon energy and the light intensity per unit area corresponding to λ, respectively.

The fabrication of C@SiCNTs is illustrated in Scheme 1. Silicon powder and CNTs were grinded together and placed inside a horizontal furnace. Then, the furnace was heated to 1320 ℃ in argon atmosphere because the sublimation temperature and melting point of the silicon powder are 1127 and 1420 ℃, respectively. The formation of SiC shells on the CNTs involved a vapor-solid reaction between the solid CNTs and the silicon vapor (Eq. (3)) [55]:

(3)

After the in situ vapor-solid reaction between CNTs and silicon vapor for 2 h, C@SiCNTs were obtained. When the vapor-solid reaction time increased to 4 h, almost all the CNTs were consumed completely by the silicon vapor and converted to SiC, then, SiCNTs were obtained. If the furnace was heated to 1500 ℃, micron SiC crystal of large particles was obtained. Because the synthesis temperature is higher than the melting point of silicon, the reaction between silicon and carbon changes from a vapor-solid reaction to a liquid-solid reaction. This phenomenon was observed and discussed in detail in our previous report [48]. Therefore, C@SiCNTs can only be formed with the right combination of reaction conditions. The reference SiC was obtained by annealing the C@SiCNTs at 750 ℃ for 4 h in air. The obtained specimens were characterized by XRD, and the patterns are presented in Fig. 1(a). It is observed that the original CNTs exhibit hexagonal structure, and the space group is P6₃/mmc (JCPDS PDF no. 41-1487). The diffraction peak angles of 26.2°, 42.8°, and 44.6° correspond to the (002), (100), and (101) lattice planes of CNTs, respectively. After the reaction with silicon vapor for 2 h, the diffraction peaks at 42.8° and 44.6° disappeared, due to the formation of the SiC coating on the CNTs. On the other hand, diffraction peaks at 35.4°, 41.2°, 59.8°, 71.6°, and 75.4°, corresponding to the (111), (200), (220), (311), and (222) lattice planes of β-SiC (JCPDS PDF no. 29-1129), respectively, can be found in the C@SiCNTs. Further, a diffraction peak at 26.2°, indexed to the (002) reflections of CNTs, can also be observed in the XRD pattern of C@SiCNTs. The low intensity diffraction peak at 33.6° corresponds to the stacking faults of SiC. It is confirmed that the as-prepared C@SiCNTs are composed of CNTs and β-SiC [58]. Upon increasing the reaction time to 4 h, CNTs were almost completely transformed to β-SiC in situ, so that only the diffraction peaks of β-SiC are observed in the XRD pattern of SiCNTs. Similarly, the reference SiC nanocrystals are composed of pure β-SiC.

The SiC contents of the as-prepared samples were determined by thermogravimetric analysis (Fig. 1(b)). The mass fraction of SiC in C@SiCNTs is about 70.5 wt%. It is found that SiCNTs and SiC are stable (up to 900 ℃ in air), since no remarkable weight losses (2.0% for SiCNTs and 1.5% for the reference SiC) are observed in this temperature range.

The textural properties of the C@SiCNTs, SiCNTs, and reference SiC samples are revealed in Fig. 1(c); obviously, all the specimens exhibit type III adsorption-desorption isotherms with type H3 hysteretic loops, which indicate the presence of mesopores connected by macropores in the samples [59]. Moreover, the isotherms reveal mass absorption in the high relative pressure zone (inset in Fig. 1(c)), which indicates the presence of large mesopores, which are observed in Fig. 1(d). As presented in Table 1, the BET surface areas of C@SiCNTs (62.57 m² g^–1) and SiCNTs (55.66 m² g^–1) are about 3.4 and 3.0 times that of the reference SiC (18.56 m² g^–1), respectively, which further suggest that C@SiCNTs and SiCNTs are topologically transformed from acid-purified CNTs (specific surface area, 40–70 m² g^–1; diameter, 60–100 nm). More interestingly, the mean pore diameters of C@SiCNTs (39.52 nm) and SiCNTs (32.94 nm) are smaller than that of CNTs (60–100 nm), which maybe correspond to the inner diameter of the CNTs. On the contrary, the mean pore diameter of the reference SiC (60.61 nm) is greater than that of the CNTs, which is because of the presence of slit-like pores that originated from the aggregation of particles. Such a nanostructure can significantly enhance the PC activity, because it provides open pores for guest molecules and also improves the light absorption through multiple scattering effects [60].

To gain further insight into the surface compositions of the catalysts, XPS was employed. The high-resolution Si 2p spectrum (Fig. 2(a)) of C@SiCNTs can be deconvoluted into two peaks corresponding to the Si-C bonds (101.18 eV) of CNTs and Si–O bonds (102.03 eV) of SiO_x. The deconvoluted C 1s spectrum (Fig. 2(b)) of the C@SiCNTs shows two peaks at the binding energies of 284.72 eV and 283.31 eV, which are indexed to the C–C bonds of CNTs and the C–Si bonds of SiC, respectively [61]. In addition, the deconvoluted Si 2p spectrum (Fig. 2(c)) of the SiCNTs reveals two peaks at 101.34 and 102.22 eV, which correspond to the Si–C bonds of SiC and the Si–O bonds of SiO_x, respectively. The deconvoluted C 1s spectrum (Fig. 2(d)) of the SiCNTs displays two peaks at the binding energies of 284.70 and 283.44 eV, which are indexed to the C–C bonds of CNTs and the C–Si bonds of SiC, respectively. Notably, the peak intensity of the C–C bonds of SiCNTs is much lower than that of the C–C bonds of C@SiCNTs, which indicates a low content of carbon in the SiCNTs. The atomic fractions of silicon, carbon, and oxygen in the C@SiCNTs are 29.35%, 65.62%, and 5.03%, respectively. Similarly, the atomic fractions of silicon, carbon, and oxygen in the SiCNTs are 45.54%, 49.01%, and 5.45%, respectively. Calculated from the atomic fractions, the mass fraction of SiC in the C@SiCNTs is about 72.5 wt%, which is close to the results of thermogravimetric analysis (Fig. 1(b)). In addition, the mass fraction of SiC in the SiCNTs is about 96.2 wt%, which suggests that 3.8 wt% of the CNTs in the SiCNTs that are coated by SiC cannot be consumed by the silicon vapor.

As presented in Fig. 3(a), the acid-purified CNTs display diameters of 60–100 nm, a smooth surface, and a crooked morphology. Comparing the morphologies of the acid-purified CNTs and the C@SiCNTs (Fig. 3(b)), it is interesting to note that the 1-D nanotube structures are still maintained. It indicates that CNTs with suitable acid-purification maintain their nanotube structures during ultrahigh-temperature sintering. In addition, the surfaces of C@SiCNTs are rougher than those of CNTs, because of the coating of SiC on the surfaces of CNTs. Furthermore, the lengths of the C@SiCNTs and SiCNTs nanotubes decrease slightly, compared with that of the CNTs. No SiC nanowires are observed in the C@SiCNTs and SiCNTs. On the other hand, the reference SiC sample consists of nanorods and nanoparticles (Fig. 3(d)).

A perfect heterojunction interface with robust intrinsic contact in the C@SiCNTs was examined by using TEM and HRTEM. As presented in Fig. 4(a), the C@SiCNTs show a nanotube constructure with a rough surface, which is consistent with the SEM image (Fig. 3(b)), as discussed previously. For further investigation, the distributions of carbon and silicon elements are uniform, and the shapes are the same as the morphology of the C@SiCNTs. Compared to that of elemental carbon, the distribution shape of Si element is shrinked, which indicates that the SiC coating on the CNTs have formed a coaxial core-shell nanotube structure. This is further confirmed from the HRTEM image (Fig. 4(b)), in which the C@SiCNTs show two crystal lattice spacings of ca. 0.25 and 0.34 nm, which are indexed to the (111) planes of β-SiC and the (002) planes of CNTs, respectively. Obviously, the SiC (111) planes and the CNT (002) planes connect together to form an excellent heterojunction with intrinsic atomic contact (inset in Fig. 4(b)). In addition, the HRTEM image shows that the (002) planes of CNTs are covered by the (111) planes of SiC, which suggests the formation of a coaxial core-shell nanotube structure. As a result, rapid electron transfer from SiC to CNTs and electron-hole separation on the SiC shell will be realized, which is significant for improving the PC H₂ evolution.

The UV-vis diffuse reflectance spectra of the samples were recorded by using a spectrophotometer. As shown in Fig. 5(a), it is clear that the three specimens exhibit an absorption band in the range 200–800 nm. Furthermore, it is also seen that the absorption band of the C@SiCNTs nanocomposites is stronger than those of the SiCNTs and the reference SiC, which indicates better light response in the case of the C@SiCNTs, which is beneficial for the improvement of the PC activity. In addition, the band gap energies (E_g) of the SiC hybrids were calculated by the Kubelka-Munk method based on the tangent lines of (αhν)^1/2 versus hν plots, where hν is the photon energy and α is the absorption coefficient [62]. As shown in Fig. 5(b), the E_g values of the C@SiCNTs, SiCNTs, and reference SiC are estimated to be 2.42, 2.64, and 2.72 eV, respectively. Obviously, the CNTs are incorporated into the SiC nanotubes, thus playing a powerful role in reducing the E_g of bulk SiC through a chemical interaction between SiC and CNTs. This phenomenon is similar to the result obtained in the case of ZnO-CNT composite materials [63]. In conclusion, narrowing the E_g is beneficial for improving the PC activity.

The photogenerated carrier transition behaviors of the samples were further investigated based on the PL spectra. In general, a lower PL signal strength indicates better charge capture and more efficient transport [64, 65]. As shown in Fig. 5(c), the reference SiC reveals a strong emission peak at 468 nm. Simultaneously, a clear fluorescence decay is seen for the C@SiCNTs sample, which indicates efficient charge separation between SiC and CNTs in the C@SiCNTs. Particularly, the emission peak intensity of the SiCNTs is lower than that of the reference SiC, because the 1-D nanotube structures permit efficient electron transfer [66]. These results suggest that coating SiC onto the surfaces of CNTs and 1-D nanotubes can distinctly promote the separation and transfer of the photogenerated charge carriers.

The transient photocurrent responses of the three samples were measured for several simulated sunlight (AM 1.5) on-off cycles to demonstrate the enhanced separation of the photogenerated charge carriers. The C@SiCNTs exhibit a higher photocurrent density than the reference SiC (Fig. 5(d)), which indicates efficient charge separation after the hybridization of SiC with CNTs. Therefore, more electrons can be photoexcited and used to yield H₂. In detail, the photocurrent density of the C@SiCNTs (about 1.3 µA cm^–2) is 8.7 and 3.7 times higher than those of the reference SiC (about 0.15 µA cm^–2) and the SiCNTs (about 0.35 µA cm^–2), respectively. Understandably, the higher the photocurrent, the greater is the number of photogenerated electrons that can be transferred from the collected product to the counter electrode when a bias is applied [67]. Fluorescence lifetime measurements were carried out to reveal the separation of the photogenerated charge carriers, As seen in the transient PL spectrum (Fig. 6), the C@SiCNTs exhibit a slower exponential decay, with an average lifetime of 6.66 ns, which is 2.46 and 1.42 times longer than those of the reference SiC (2.70 ns) and the SiCNTs (4.69 ns), respectively. The increased exciton lifetime of the C@SiCNTs suggests that the photocarrier is more likely to be involved in surface redox reactions, which is conducive to improving the PC performance.

The PC H₂ evolutions of the C@SiCNTs, SiCNTs, and reference SiC were measured under simulated sunlight irradiation (AM 1.5) in an aqueous Na₂S solution (0.10 mol L^–1). The total amount of H₂ evolved increases linearly with the irradiation time during the PC reactions over all the samples (Fig. 7(a)), which suggests that all the samples display good photostability under simulated sunlight irradiation. The total H₂ productions after 5 h of irradiation for the C@SiCNTs, SiCNTs, and reference SiC are 592.6, 302.8, and 105.3 µmol g^–1, respectively.

The average H₂ evolution rates over the three samples after 5 h of irradiation were estimated, and the values are displayed in Fig. 7(b). The H₂ production rate of the C@SiCNTs (118.5 µmol g^–1 h^–1) is about 5.62 and 1.95 times those of the reference SiC (21.1 µmol g^–1 h^–1) and the SiCNTs (60.6 µmol g^–1 h^–1) under simulated sunlight irradiation, respectively. A summary of the previously reported SiC/carbon-based photocatalysts is presented in Table 2. Even though the activity obtained here (118.5 µmol g^–1 h^–1) is lower than those of graphene-covered SiC/Pt [68], SiC-covered graphene [55], and SiC/carbon nanofibers[61], it is better than those of other SiC/carbon-based photocatalysts such as GO/SiC [53], SiC-graphene [52], and MWCNTs/SiC nanowires, which suggests an excellent synergetic effect between the CNT cores and the SiC shells.

In addition, the stability of the C@SiCNTs in PC H₂ evolution was investigated by cyclic experiments. As displayed in Fig. 7(c), after four repeated runs, no remarkable decay in the H₂ evolution is observed after the PC reaction for 20 h under simulated sunlight irradiation. It is further demonstrated that the C@SiCNTs photocatalyst also exhibits excellent stability during PC H₂ production. To further confirm the good stability of the C@SiCNTs photocatalyst, the structural morphology of the C@SiCNTs was characterized after the cyclic experiments. The XRD pattern and SEM image of the C@SiCNTs obtained after the cyclic experiments are shown in Fig. 7(e) and Fig. 7(f), respectively. As can be seen, there is no obvious change in the structure and morphology of the C@SiCNTs after the cyclic experiments.

In order to clarify that the photogenerated holes were consumed by Na₂S, the PC H₂ evolution over the C@SiCNTs was investigated in pure under simulated sunlight irradiation. The total amount (Fig. 7(d)) of H₂ produced increases linearly. The average H₂ evolution rate over the C@SiCNTs after 5 h irradiation is 5.3 µmol g^–1 h^–1. The results confirmed that the H₂ production rate of C@SiCNTs in a Na₂S solution is higher than that in pure water under simulated sunlight irradiation, since the recombination of the photogenerated charge carriers is remarkably suppressed.

In view of the above analysis of the hybrids, the enhanced PC performance of the C@SiCNTs can be attributed to the 1-D coaxial core-shell nanoheterostructures and to the Schottky junction formed between CNTs and SiC, which results in efficient intrinsic interfacial charge transfer. The Schottky-type junction of C@SiCNTs improves the PC activity in three ways: (1) the 1-D coaxial core-shell nanotube structures exhibit excellent performances such as rapid electron transport and enhanced light absorption and scattering; (2) CNTs act as photosensitizers by enhancing the absorption of light. This is confirmed by UV-vis spectroscopy (Fig. 5(a)), which reveals that the absorption band of C@SiCNTs is stronger than those of SiCNTs and the reference SiC [69]; (3) a robust Schottky junction is formed between CNTs and the semiconductor by in situ growth of SiC nanoshells on the CNTs [70]. Such a junction decreases the barrier for charge transfer and thus suppresses the electron-hole recombination.

The flat band (fb) potentials of the SiC nanostructures were estimated based on the Mott-Schottky plots by using Eq. (4). The surface carrier density was determined according to Eq. (5).

(4)

(5)

where C is the space charge layer capacitance, N_d stands for electron donor density, e₀ stands for electron charge, E denotes the applied potential, and ɛ and ɛ₀ represent the permittivity of the material and the permittivity of vacuum, respectively. According to a previous report, the ɛ value of β-SiC is 6.9 and the ɛ₀ value is 8.85 × 10^–12 F m^–1 [71]. V_fb is the flat band potential, which is calculated by extrapolating the fitted line to C^–2 = 0. The positive tangent slopes imply that the three photocatalysts are n-type semiconductors (as displayed in Fig. 8(a)). The estimated V_fb for the C@SiCNTs, SiCNTs, and reference SiC are –0.102, 0.002, and 0.017 V (vs RHE), respectively. Obviously, the V_fb of the C@SiCNTs exhibits a positive shift, compared to that of the reference SiC, since the Fermi level of SiC, which is below the conduction band, is more negative than the Fermi level of CNTs. The electrons of SiC are transferred to CNTs across the interface between SiC and CNTs. These electron movements result in band bending at the interface between SiC and CNTs, and consequently, generate an interfacial electric field. In addition, the C@SiCNTs exhibit a much smaller positive slope than the reference SiC, which suggests a larger electron density for the C@SiCNT photocatalyst. The N_d are 1.18 × 10²², 9.67 × 10²¹, and 6.69 × 10²¹ for the C@SiCNTs, SiCNTs, and reference SiC, respectively. The aforementioned photoelectrochemical measurement results are in accordance with the charge densities of the three samples.

The charge transfer kinetics of the C@SiCNTs, SiCNTs, and reference SiC were further investigated based on the electrochemical impedance spectra that were obtained under illumination at an ac amplitude of 5 mV in the frequency range 0.01–10⁵ Hz. A smaller radius of the semicircle in the EIS curve indicates a better ability to transport charges [72]. Obviously, the radius of the semicircle of the C@SiCNTs is smaller than those of the SiCNTs and reference SiC (as seen in Fig. 8(b)), which suggests that the coaxial core-shell nanotube heterostructures can effectively reduce the interfacial charge transfer resistance and enhance the separation efficiency of the photogenerated charge carriers.

By using a Xe lamp equipped with pieces of band-pass filters, the H₂ generation properties of the C@SiCNTs, SiCNTs, and reference SiC were further measured under 420 nm monochromatic light irradiation. The average rates of PC H₂ generation over the C@SiCNTs, SiCNTs, and reference SiC were 53.5, 24.3, and 6.3 µmol g^–1 h^–1, respectively (Fig. 8(c)). The corresponding AQEs of the C@SiCNTs, SiCNTs, and reference SiC under 420 nm visible light irradiation reached 0.081%, 0.037%, and 0.01%, respectively (Fig. 8(d)). The AQE of the C@SiCNTs is much higher than that of the reference SiC, which further indicates that C@SiCNTs can deliver more photoelectrons for the H₂ reduction reaction.

The polarization curves for the electrocatalytic H₂ evolutions over the C@SiCNTs, SiCNTs, and reference SiC were obtained at 5 mV s^–1 scan rate from 0 V to −1.0 V vs RHE. As shown in Fig. 9, the onset potentials for electrocatalytic H₂ evolutions over the C@SiCNTs, SiCNTs, and reference SiC are about −0.45, −0.69, and −0.78 V (vs RHE), respectively. The C@SiCNTs exhibit reduced onset overpotentials, which suggest that the formation of the heterojunction between SiC and CNTs improves the catalytic activity. All the SiC samples exhibit much lower overpotentials than the bare FTO, which indicates that SiC displays a higher catalytic activity [47].

Illustrations of the electron-hole transfer in C@SiCNTs and the PC H₂ production mechanism over C@SiCNTs photocatalysts are shown in Scheme 2. The electrons in the VB of SiC can be excited to the CB of SiC under simulated sunlight irradiation. The photogenerated electrons from the excited SiC can be transferred to the CNTs, because CNTs display a higher electrical conductivity and a more positive Fermi level than SiC (based on the work functions of SiC (4.4 eV) [73] and CNTs (4.8 eV)) [74]. Furthermore, the 1-D coaxial core-shell nanotube structures consisting of CNT cores and SiC shells offer abundant channels for charge transfer, which suppresses the recombination of the photogenerated electron-hole pairs. CNTs function as an electron donor to accept the photogenerated electrons from the host SiC to produce H₂. Meanwhile, the holes can be consumed by the sacrificial reagent (Na₂S) to regenerate the ground-state SiC. As a result, a significantly enhanced reaction activity and stable PC H₂ evolution can be achieved over the 1-D coaxial core-shell CNTs@SiC nanotube photocatalysts.

We demonstrated that the PC performance of the SiC/CNTs hybrid can be effectively improved through rational engineering of the morphology and interfaces of the SiC active layer. C@SiCNTs with a coaxial core-shell nanotube structure were synthesized by an in situ vapor-solid chemical reaction between CNTs and silicon vapor. The obtained C@SiCNTs exhibited a high photoactivity and stability in water splitting. The enhanced PC activity of the CNTs/SiC hybrids is attributed to the synergistic effects between SiC and CNTs, including the 1-D core-shell nanotube heterostructure, enhanced light absorption, as well as the formation of tight Schottky-type heterojunctions between CNTs and SiC, which promote charge separation at the interfaces of nanoheterostructures.