Characterization of TiO2/Ce2S3 heterostructure. The obtained TiO2/Ce2S3 heterojunctions were denoted as TCx, where T and C represent TiO2 nanofiber and Ce2S3 nanoparticle, respectively; x is the mole percentage of Ce2S3 with respect to TiO2. As can be seen from the field emission scanning electron microscopy (FESEM) and transmission electron microscopy (TEM) images, Ce2S3 nanoparticles uniformly deposited on the surface of TiO2 nanofibers (Fig. 1a, b). The phase structure of the hybrid nanofibers was determined by X-ray diffraction (XRD) (Supplementary Fig. 2) and high-resolution transmission electron microscopy (HRTEM). Two sets of lattice spacings of 0.352 and 0.325 nm, corresponding to anatase (101) and rutile (110) planes, respectively, were observed in HRTEM image (Supplementary Fig. 3b). The lattice spacings of TiO2, as well as Ce2S3, appeared in the HRTEM image simultaneously (Fig. 1c) for the hybrid nanofibers. The energy-dispersive X-ray spectroscopy (EDX) spectrum (Fig. 1d) and the high-angle annular dark field (HAADF) image (Fig. 1e) both revealed the existence of Ti, O, Ce and S elements in the composite sample TC5, further confirming the formation of TiO2/Ce2S3 nanohybrids.
The surface chemical states of the samples were studied by X-ray photoelectron spectroscopy (XPS). Two symmetrical peaks (Ti 2p3/2 and Ti 2p1/2) assigning to Ti4+ ions can be seen in the high resolution XPS spectra of Ti 2p of TiO2 and TC5 (Fig. 2a). The Ce 3d spectra can be resolved into two pairs of spin-orbit peaks as shown in Fig. 2b. The peaks located at 885.8 and 903.9 eV were attributed to the state of Ce3+ in Ce2S3. The other pair of peaks at 882.4 and 900.5 eV were assigned to the satellite peak of Ce. Noticeably, in TC5 sample, the binding energies (BEs) of Ti 2p and O 1s (Supplementary Fig. 4a) shifted 0.2 eV towards a lower value comparing with those of pristine TiO2; conversely, the BEs of Ce 3d and S 2p (Supplementary Fig. 4b) of TC5 became more positive in comparison with those of Ce2S3, implying the electrons transfer from Ce2S3 to TiO2 upon hybridization in dark. Such electron transfer between TiO2 and Ce2S3 resulted in an internal electric field (IEF) pointing from Ce2S3 to TiO2, which will drive the efficient separation of photoinduced charge carriers as revealed by in-situ XPS (discussed below) and form S-scheme TiO2/Ce2S3 heterojunctions to promote the photocatalytic performance.
The difference of work functions (Φ), which can be explored by the contact potential difference (CPD) between the samples and the standard gold tip of Kelvin probe instrument, is an essential parameter leading to the electron transfer within duplicate semiconductor heterostructures in dark. As shown in Fig. 2c, the CPD of TiO2, TC5 and Ce2S3 reached a stable value in dark. The work function (Φ) and Fermi level (EF) can be calculated by the following equations:
Φ = Φtip + e × CPD (1)
E F = −Φ (2)
where e represents the electron charge, Φtip is the standard work function of the gold tip (~ 4.25 eV).32 As shown in Fig. 2d, the work functions of TiO2, TC5 and Ce2S3 were estimated to be 4.31, 4.27 and 3.99 eV, respectively, indicating that the Fermi level of TiO2 is lower than that of Ce2S3 (− 4.31 vs. −3.99 eV), consistent with our above DFT analysis (Supplementary Fig. 1). Once they begin to contact with each other, the electrons tend to transport form Ce2S3 to TiO2, forcing the phases to reach a same Fermi level (− 4.27 eV) and thus creates an IEF at TiO2/Ce2S3 interfaces directing from Ce2S3 to TiO2 (Fig. 2d), which is in accordance with above XPS results.
Insights into charge separation mechanism of S-scheme heterojunction. To explore the charge separation mechanism, the band structures of TiO2 and Ce2S3 were first to investigate. The VB XPS spectra showed that the energy levels of valence band maximum (VBM) of TiO2 and Ce2S3 were 2.67 and 0.95 eV (vs. NHE), respectively (Supplementary Fig. 5a). The flat-band potentials of TiO2 and Ce2S3 were 0.07 and 0.05 eV (vs. NHE), respectively, derived from the Mott-Schottky plots (Supplementary Fig. 5b). Combined with the band gap of TiO2 (3.27 eV) and Ce2S3 (2.01 eV) (Supplementary Fig. 5c), the positions of VBM and the conduction band minimum (CBM) of TiO2 and Ce2S3 were derived as shown in Supplementary Fig. 5d.
Based on the above XPS, DFT calculation and CPD analyses, the Fermi level of TiO2 was lower than that of Ce2S3, inducing the electrons to flow from Ce2S3 to TiO2 upon contact with each other, forming an IEF directing from Ce2S3 to TiO2, meanwhile bending the energy bands of TiO2 and Ce2S3 at interfaces. Upon light excitation, the VB electrons of TiO2 and Ce2S3 both jumped to their CBs. Driven by the interfacial IEF and bent bands, the photoinduced electrons in TiO2 CB transferred to Ce2S3 VB spontaneously and recombined with the holes in Ce2S3 VB. While the powerful photogenerated electrons in Ce2S3 CB and the holes in TiO2 VB were preserved to engage in photocatalytic reactions. Such charge-transfer route resembles ‘‘step’’ in macroscopic level (from TiO2 CB to Ce2S3 CB) and letter of “N” in microscopic level, following the S-scheme charge transfer pathway (Fig. 3a), which can efficiently separate the photogenerated electron-hole pairs and in the meantime maintain the high redox ability of electrons in Ce2S3 CB and holes in TiO2 VB, respectively. Therefore, the obtained S-scheme TiO2/Ce2S3 heterojunctions will endow excellent photocatalytic nitrobenzene hydrogenation activity.
To evidence the charge-transfer route in S-scheme heterojunction more intuitively, in-situ XPS spectra were measured under light irradiation (Fig. 2a, b and Supplementary Fig. 4). Obviously, the BEs of Ti 2p and O 1s for TC5 shifted positively, while the BEs of Ce 3d and S 2p of TC5 showed a negative shift under light irradiation, with reference to the corresponding BEs in the XPS spectra in dark. Such shifts soundly proved that the photoexcited electrons in TiO2 CB transferred to Ce2S3 VB under light irradiation. From macroscopic perspective, electrons in the TiO2/Ce2S3 nanohybrids steps from low energy level (TiO2 CB) to high energy lever (Ce2S3 CB), following the S-scheme charge-transfer pathway.33–35
The S-scheme charge transfer dynamics was further investigated by the femtosecond transient absorption spectroscopy (fs-TAS), which is widely recognized as the most powerful method to disclose the photoinduced charge transport mechanism in semiconductors.36–38 As depicted in Fig. 3b-e, a pronounced negative peak at 645 nm was observed for both TiO2 and TiO2/Ce2S3 nanohybrids (TC5), which can be interpreted as the stimulated radiation of TiO2 and corresponded to the recombination of photoinduced electrons and holes.39–41 This assignment was further confirmed by the experiment performed in the presence of an electron scavenger (AgNO3), in which the signal virtually disappeared (Supplementary Fig. 6), indicating that the photogenerated electrons were trapped by the electron scavenger with no electrons to recombine with the holes. Through monitoring the signal strength at 645 nm with different delay times, we noticed that the composite sample TC5 exhibited weaker intensity than pure TiO2 nanofibers, suggesting less recombination of photocarriers in TC5 than in pristine TiO2. Moreover, the spectroscopic nature of the charge carriers of TC5 altered at a much slower rate after 50 ps, indicating a relatively slow recombination of electrons and holes over the nanohybrids. The decay kinetics at 645 nm within 100 ps of TiO2 and TC5 were fitted and analyzed using a tri-exponential function and the results were presented in Fig. 3f and Table 1. In general, the whole series of process can be partitioned into three-time regions. The shortest lifetime (τ1) corresponds to the process of electron trapping, which means the photoexcited electrons in CB quickly relax into the impurity state of Ti3+ first, and this process is the fastest. The nanohybrids TC5 possessed more Ti3+ than pure TiO2 as evidenced by electron spin resonance (ESR) spectra shown in Supplementary Fig. 7. Herein, for TC5, more photogenerated electrons in TiO2 CB were trapped by the impurity state, leading to longer lifetime of τ1 (3.92 ns) in comparison with pure TiO2 (1.88 ns). The lifetime τ2 is assigned to the process of interfacial electron transfer from the impurity state of TiO2 to Ce2S3 VB, and τ3 can be ascribed to the recombination of holes in TiO2 VB with electrons at the impurity state. When Ce2S3 nanoparticles deposited on TiO2 nanofibers, the photoexcited electrons in TiO2 can find a new relaxation channel via Ce2S3, which leads to longer lifetime of τ2 (TC5: 43 ns, TiO2: 27 ns) (inset of Fig. 3f) and thus shorter lifetime of τ3 (TC5: 39 ns, TiO2: 103 ns) (Fig. 3f) comparing with pristine TiO2. It is worth noting that as the delay time prolonged, the kinetic trace curves rose tardily again, especially for TC5 (Supplementary Fig. 8), which was owing to the fact that free electrons trapped at the impurity state slowly released later and thus prolonged the lifetime. Therefore, it is apparent that the rapid surface trapping of the electrons as well as a timely and spatial charge separation occurred in the TiO2/Ce2S3 nanohybrids led to the suppressed electron-hole recombination and supported the formation of S-scheme charge transfer between TiO2 and Ce2S3. Accordingly, the schematic illustration of the S-scheme charge transfer mechanism in TiO2/Ce2S3 nanohybrids was depicted in Fig. 3g.42
Table 1
Parameters derived from fitted kinetics at 645 nm within 100 ps for TiO2 and TC5.
Sample
|
τ1
(ps)
|
τ2
(ps)
|
τ3
(ps)
|
TiO2
|
1.88
|
27
|
103
|
TC5
|
3.92
|
43
|
69
|
The transient photocurrent responses for TiO2 and TC5 were recorded to reaffirm the charge separation between TiO2 and Ce2S3 (Supplementary Fig. 9a), where the composite TC5 exhibited a higher photocurrent density compared with pure TiO2, testifying the superior charge separation efficiency of TiO2/Ce2S3 nanohybrids. Interestingly, under initial light irradiation, TC5 exhibited an obvious photocurrent decrease; when light was switched off, the photocurrent dropped steeply then increased smoothly. Such unique phenomenon was supposed to be induced by the trapping of electrons by the impurity state of Ti3+. Upon light irradiation, the electrons first jumped from VB to CB, then quickly trapped by the impurity state, resulting in the instantaneous decrease of photocurrent. Accordingly, when the light was off, the photocurrent declined then rose steadily, indicating that the trapped electrons gradually migrated to the cathode. Herein, the electron trapping at the impurity state is beneficial to the separation of photogenerated electron-hole pairs and can improve the photocatalytic performance, as also evidenced by above fs-TAS results. Additionally, the electrochemical impedance spectra (EIS) of TiO2, TC5 and Ce2S3 (Supplementary Fig. 9b) showed that the composite TC5 exhibited the smallest semicircle in comparison with pristine TiO2 and Ce2S3, indicating that the TiO2/Ce2S3 nanohybrids revealed lowest charge transfer resistance. The polarization curves of TiO2, TC5 and Ce2S3 under UV light irradiation (Supplementary Fig. 9c) showed that the overpotential for TC5 was lower than that of TiO2 and Ce2S3, indicating better reduction capability of TiO2/Ce2S3 nanohybrids. These electrochemical results proved that Ce2S3 nanoparticles could form S-scheme heterojunction with TiO2 nanofibers promoting the electron transfer and separation of electron-hole pairs. Based on these analyses, it is expected that the TiO2/Ce2S3 S-scheme heterojunction shall exhibit excellent performance toward photocatalytic hydrogenation of nitrobenzene to aniline.
Photocatalytic nitrobenzene hydrogenation over TiO2/Ce2S3 hybrids. The photocatalytic nitrobenzene hydrogenation performance of the resultant samples was evaluated by using H2O as hydrogen source. Controlled experiments demonstrated that no aniline was detected in dark or in the absence of either H2O or nitrobenzene, indicating that the light irradiation, input H2O and nitrobenzene were indispensable for the photocatalytic hydrogenation reaction. As shown in Fig. 4a and b, pristine TiO2 nanofibers and Ce2S3 nanoparticles showed relatively lower production rate of aniline owing to the rapid recombination and lower utilization of photoinduced charge carriers. After depositing Ce2S3 nanoparticles on TiO2 nanofibers, the concentration of aniline was greatly enhanced and achieved the maximum yield of 99% when TC5 was irradiated for 90 min. Such excellent photocatalytic hydrogenation performance of TiO2/Ce2S3 nanohybrids was in our expectation and outperformed most photocatalysts with deposited noble metals as cocatalysts.43–45 Nevertheless, further loading of Ce2S3 (e.g., TC7) would deteriorate the photocatalytic activity of the nanohybrids because the overloading could shield the light absorption of TiO2. Note that the preserved holes in TiO2 VB oxidized triethanolamine (TEOA) to produce diethanolamine (DEA), prolonging the lifetime of electrons in Ce2S3 CB. To determine the hydrogen source of the hydrogenation reaction, an isotope-labeling experiment was conducted by replacing H2O with D2O over TC5. As shown in Fig. 4c, the total ion chromatographic peak at 3.79 min corresponded to aniline, which produced two main ion signals, as well as some fragment peaks in the mass spectrum. The m/z values of 95 and 67 were assigned to aniline (C6H5D2N) and the fragmentation benzene of aniline (C6H5D), respectively, verifying that the hydrogen source was the input water rather than any other hydrogen-containing agents such as triethanolamine or organic solvents. To better demonstrate the liability of the resultant samples, we evaluated the recyclability and stability of the catalyst TC5 for photocatalytic hydrogenation. As shown in Supplementary Fig. 10, after five-time cycles, neglecting the influence of the catalyst loss during washing and drying processes, the decay of the aniline yield was hardly perceptible. The applied photocatalyst showed no detectable change in phase and morphology, evidenced by the XRD pattern (Supplementary Fig. 11), FESEM and TEM images (Supplementary Fig. 12). The chemical states were consistent with those of the fresh one as reveled by XPS spectra in Supplementary Fig. 13. All the results suggest the acceptable photostability of TiO2/Ce2S3 nanohybrids during the photocatalytic nitrobenzene hydrogenation reaction.
The adsorption of the reactant and the desorption of the product on the heterostructured photocatalyst were calculated using DFT theory. Supplementary Fig. 14 compared the optimized H2O and nitrobenzene molecules adsorbed on TiO2 and Ce2S3. The adsorption energy (Ead) of H2O on anatase TiO2 (− 0.9 eV), rutile TiO2 (− 1.16 eV) and Ce2S3 (− 1.01 eV) was comparable, indicating that H2O was adsorbed equally on TiO2 and Ce2S3 from a thermodynamic perspective. However, Ce2S3 was located at the outer shell of TiO2 nanofibers. H2O had shorter diffusion distance to reach Ce2S3, enabling preferential adsorption on the Ce2S3 rather than the inner TiO2 in a kinetic view. Further investigation shows that nitrobenzene preferred adsorption on Ce2S3 with more negative adsorption energy (− 1.76 eV) than on anatase TiO2 (− 1.34 eV) or rutile TiO2 (− 1.46 eV), suggesting that Ce2S3 was the prior adsorption site for both reactants. We also investigated the desorption energy of aniline on each photocatalyst (Supplementary Fig. 15). The aniline demanded less energy to desorb from Ce2S3 (0.46 eV) than from rutile (1.13 eV) or anatase TiO2 (0.49 eV). The results manifest that Ce2S3 nanoparticles are in favor of both reactant adsorption and product desorption, which definitely behave as the active sites (not TiO2 nanofibers) for the photocatalytic hydrogenation of nitrobenzene.
The intermediate species was explored with in-situ DRIFTS to uncover the mechanism of the photocatalytic hydrogenation reaction (Fig. 4d). As H2O and nitrobenzene vapors were introduced into the system in dark, there appeared characteristic absorption bands belonging to the adsorbed H2O (2350 cm− 1) and nitrobenzene (2860 and 2965 cm− 1). Their intensities gradually increased with the adsorption time, confirming the simultaneous adsorption of H2O and nitrobenzene on the photocatalyst. Upon light irradiation, new absorption band appeared at 3230 cm− 1 corresponding to (N) − OH species, while the bands at 1653 and 3435 cm− 1 could be assigned to N = O and –NH, which belonged to the intermediate species during the photocatalytic hydrogenation reaction. In addition, the − NH2 absorption band of aniline was also detected at 1157 cm− 1 under light irradiation, suggesting the production of aniline. The most likely pathway for photocatalytic hydrogenation of nitrobenzene to aniline was proposed as follows (Supplementary Fig. 16):
where the asterisk denotes the photocatalyst and Ph represents the benzene ring.
To confirm the reaction mechanism, we performed DFT calculations on Gibbs free energy change of each elementary reaction (Supplementary Fig. 16). As shown in Fig. 4e and Supplementary Fig. 17, the photocatalytic nitrobenzene hydrogenation reaction was an exothermic and spontaneous process, indicating that the proposed hydrogenation pathway is reasonable in a thermodynamic view. However, the desorption of aniline was endothermic and required an extra energy, which was the rate-limiting step for all the samples. As mentioned above, the desorption energy of aniline on Ce2S3 surface (0.46 eV) was lower than that on anatase and rutile TiO2 surfaces (0.49 and 1.13 eV) (Fig. 4f), which implied that Ce2S3 is more beneficial for aniline desorption and promotes the photocatalytic hydrogenation reaction.
In summary, we have constructed a unique TiO2/Ce2S3 S-scheme heterojunction photocatalyst for enhanced aniline production by nitrobenzene hydrogenation with water as proton source. DFT calculation and CPD results revealed the Fermi level of TiO2 was lower than that of Ce2S3, implying electrons transfer from Ce2S3 to TiO2 as evidenced by XPS, which bent the bands and created an IEF at the interfaces. Under light irradiation, the photoinduced electrons in TiO2 CB would immigrate to Ce2S3 VB driven by the bent bands and IEF, confirming the formation of S-scheme heterojunction between TiO2 and Ce2S3 and achieving efficient separation of photocarriers, thus improve the photocatalytic performance greatly. In-situ XPS and fs-TAS spectra further confirmed the charge-transfer route of S-scheme heterojunction. In-situ DRIFTS together with DFT calculations shed light on the step-by-step photocatalytic reaction mechanism toward hydrogenation of nitrobenzene to aniline intuitively and detailedly. Isotope (D2O) tracer results verified that the hydrogen was from water rather than from other hydrogen-containing agents. This work provides a pathway for the design of S-scheme photocatalysts and mechanism insights into boosted photocatalytic nitrobenzene hydrogenation performance.