Introduction

Photocatalytic hydrogen evolution via water splitting is a promising way to mitigate current energy and environmental issues1,2,3,4,5,6,7. Since Fujishima and Honda first reported solar energy conversion for hydrogen evolution using semiconductor-based photocatalysts, we have witnessed the development of artificial photocatalytic systems for enhanced hydrogen evolution reactions (HER)8,9,10. The primary drawback in using a pure semiconductor was fast recombination of photoinduced carriers, which seriously limited photocatalytic efficiency11,12. Meanwhile, co-catalysts, such as Pt nanoparticles (NPs) and RuO2, have been alternatives that provide an active center for redox reductions, reduce overpotentials for HER or oxidation reactions, and promote fast separations of photoinduced electrons and holes13,14,15,16. Thus, complex heterostructures with spatial charge separations via fine control of photoinduced carrier dynamics have been fabricated to improve photocatalytic activity.

Metal-organic frameworks (MOFs) are porous crystalline materials with high specific-surface areas, and tunable structures and functionalities. As alternatives to semiconductor photocatalysts that are receiving much interest in a wide range of applications. Unfortunately, MOFs, such as UiO-66 (Zr)17,18, exhibit poor visible-light absorption. The introduction of amino groups into the terephthalic acid ligands of UiO-66 (Zr) broaden light absorption to the visible region, and the presence of amino groups does not affect the structural stability of the UiO-66 host19. Photoinduced electron dynamics in MOFs have been well investigated, but the kinetics of photogenerated holes and their effect on photocatalytic activity remain poorly understood. Cadmium sulfide (CdS) has been the preferred visible-light photocatalyst for HER. However, photoinduced corrosion and fast recombination of photoinduced electron−hole pairs have severely restricted improvements in its catalytic activity. Thus, fine-tuning photoinduced carrier dynamics is necessary to suppress corrosion by accumulated holes3,20,21,22,23,24,25,26,27.

Here, we synthesized a spatial charge structure for the MOF-based photocatalyst (Pt@NH2-UiO-66/CdS). Relative to other MOF-based systems (see Supplementary Table 1), it exhibited the highest visible-light photocatalytic activity for HER, with an apparent quantum yield of 40.3% at 400 nm. Direct observation of the carrier dynamics in Pt@NH2-UiO-66/CdS, via transient absorption spectroscopy (TAS), revealed that a unique hole-transfer pathway was responsible for the enhanced and stable HER. The present results could help design spatial, multi-phase, heterostructured photocatalysts with high photocatalytic activity and facile control of photoinduced carrier dynamics.

Results and discussion

Characterization of materials

We synthesized Pt@NH2-UiO-66/CdS heterostructured nanocrystals (HNCs) by in-situ encapsulation of Pt nanoparticles (NPs) into MOFs having regular 100-nm octahedral shapes, and by growing CdS on the outer MOF surface, as shown in Supplementary Fig. 1. Figure 1a shows representative transmission electron microscopy (TEM) images of NH2-UiO-66 NCs with regular octahedral structures, which were consistent with scanning electron microscopy images in Supplementary Fig. 2. The Pt NPs (Supplementary Fig. 3) were encapsulated in situ in the MOF shown in Fig. 1b. By growing multiple CdS satellites of 10.5 ± 2.1 nm in size on the outer layer of Pt@NH2-UiO-66, Pt@NH2-UiO-66/CdS HNCs were formed with a spatial configuration of inner Pt NPs and outer CdS, as shown in Fig. 1c. Figure 1d is a scanning electron microscopy image of Pt@NH2-UiO-66/CdS HNCs that shows the CdS NPs attached to the surface of the MOFs. X-ray diffraction patterns in Supplementary Fig. 4 show that the Pt@NH2-UiO-66/CdS HNCs were composed of NH2-UiO-66 and zinc blend CdS, with Joint Committee on Power Diffraction Standards (JCPDS) no. 10-0454 phases. The Zr/Cd weight ratio was 0.28:14.3, as determined by inductively coupled plasma optical emission spectroscopy (ICP-OES). However, the Pt NPs did not display a diffraction peak due to the low amount of Pt (about 0.023 wt %) determined by ICP-OES (Supplementary Table 2). The high-resolution TEM image in Fig. 1e and Supplementary Fig. 5 revealed the composition of Pt NPs and CdS NPs. The 0.34-nm and 0.20-nm lattice fringes were assigned to zinc blend CdS (111) and Pt (200), respectively, which were consistent with fast Fourier transform patterns [Fig. 1f, g]. High-angle annular dark-field scanning TEM and energy-dispersive spectrometry elemental mapping (Fig. 1h) also verified the formation of ternary heterostructures composed of Pt, NH2-UiO-66, and CdS phases.

Furthermore, the Brunauer–Emmett–Teller surface area and pore structures analysis were performed, as shown in Supplementary Fig. 6 and Supplementary Table 3. By comparing the specific-surface area of the NH2-UiO-66 (820 m2 g−1), Pt@NH2-UiO-66 (722 m2 g−1), Pt@NH2-UiO-66/CdS (612 m2 g−1) HNCs, the smaller specific-surface area of Pt@NH2-UiO-66/CdS indicated that the CdS occupies the surface sites of the MOF. In addition, compared with that of NH2-UiO-66 and Pt@NH2-UiO-66, the porous volume of Pt@NH2-UiO-66/CdS was increased due to the effects of CdS NPs. We also further verified the composition and interface characteristics of Pt@NH2-UiO-66/CdS composites by the X-ray photoelectron spectroscopy (XPS) measurements (Supplementary Fig. 7). First, the XPS spectrum of Supplementary Fig. 7a shows that the Pt@NH2-UiO-66/CdS composites contain the elements such as C, N, O, Zr, Pt, Cd, and S, and there are no other impurity elements. Secondly, the binding energies of Cd 3d3/2 and 3d5/2 of the sample Pt@NH2-UiO-66/CdS are around 411.9 and 405.2 eV, respectively, indicating that the Cd in the composite material exists in the +2 valence28. In addition, the binding energies of S 2p1/2 and 2p3/2 are around 162.7 and 161.5 eV, respectively, indicating that S is −2 valence29 (Supplementary Fig. 7d, e). Therefore, these XPS results can fully demonstrate the existence of CdS in the Pt@NH2-UiO-66/CdS composites. Compared with NH2-UiO-66 and Pt@NH2-UiO-66, the Zr 3d binding energy of Pt@NH2-UiO-66/CdS composite was shifted by 0.1 eV (Supplementary Fig. 7b). It suggested that there was a strong interaction between CdS and NH2-UiO-66, rather than a simple physical contact30. As shown in Supplementary Fig. 7c, the binding energies of Pt 4f5/2 and 4f7/2 in the Pt@NH2-UiO-66 are around 74.4 and 71.1 eV, respectively, which are assigned to the metallic Pt31. However, the binding energies of Pt f5/2 and 4f7/2 in Pt@NH2-UiO-66/CdS were shifted to around 74.7 and 71.3 eV, respectively, indicating that the loading of CdS in the composite caused the binding energy of Pt to be shifted by 0.3 eV, but the valence state of Pt has not changed.

Optical properties and photocatalytic activity

The visible-light-driven photocatalytic activity and stability of Pt@NH2-UiO-66/CdS HNCs for HER were investigated. Figure 2a shows the ultraviolet-visible absorption spectra of NH2-UiO-66, Pt@NH2-UiO-66, Pt@NH2-UiO-66/CdS, and CdS NPs (also, see Supplementary Fig. 8). Pt@NH2-UiO-66/CdS featured the absorption characteristics of MOF at 370 nm and the band-edge absorption of the CdS phase at 500 nm. Figure 2b shows the energy levels of NH2-UiO-66 and CdS that were determined in Supplementary Fig. 9; they were consistent with the previous reports3,32. As shown in Fig. 2c, the HER visible-light-driven photocatalytic activity (using lactic acid as a sacrificial agent) for Pt@NH2-UiO-66/CdS was 37.76 mmol h−1 g−1, which was higher than that of NH2-UiO-66 (0.011 mmol h−1 g−1), Pt@NH2-UiO-66 (0.12 mmol h−1 g−1), NH2-UiO-66/CdS (1.65 mmol h−1 g−1, Supplementary Fig. 10), and CdS NPs (0.41 mmol h−1 g−1). The photocatalytic H2 evolution has increased with the increase of the amount of catalyst, then decreased when the dosage of catalyst was 10 mg (Supplementary Fig. 11). Thus, the 5 mg of the optimized photocatalyst dosage displays the best catalytic performance. When the photocatalytic H2 evolution was normalized by the quality of the catalyst, however, with the increase of the catalyst dosage, the hydrogen evolution rate gradually decreases, possibly due to blocked or scattered light by the excess suspended photocatalysts in the reaction medium21,22,23. To validate the benefits of spatial separation of the co-catalysts in the MOFs, we examined the HER photocatalytic activity for physical mixtures of materials with normalized Pt contents. Supplementary Fig. 12 shows that Pt@NH2-UiO-66/CdS exhibited higher activity than the others, which was consistent with better photocatalytic activity of Pt implanted in the MOF33. Furthermore, the cascade photoinduced carrier transfer with the spatial charge separation between the ternary phases with strong interactions contributed much to the high activity. The Pt@NH2-UiO-66/CdS maintained good stability after seven cycles over twenty-one hours and its morphology had no significant changes, as shown in Supplementary Fig. 13. Various sacrificial agents from small to large molecules were used to reveal the roles in HER photocatalytic activity by consuming photogenerated holes. Supplementary Fig. 14 shows that the lactic acid was the best sacrificial agent, where the carboxyl groups played a more important role than the hydroxyl groups. The large molecules could be oxidized by photogenerated holes in Pt@NH2-UiO-66/CdS. In the photocatalytic degradation of macromolecules, the generation of •OH radicals could be detected by fluorescence emission and electron-paramagnetic-resonance spectra. Larger signals were trapped by 5,5-dimethyl-1-pyrroline N-oxide (DMPO)32 (see Supplementary Fig. 15), which restricted the HER photocatalytic activity. As shown in Supplementary Fig. 16, with the increase of lactic acid contents (decrease of the pH) in the system, the H2 evolution rate of the catalyst gradually increases, which further indicates that the timely removal of holes is beneficial to improve the H2 evolution rate. The wavelength-dependent apparent quantum yields (AQYs) of Pt@NH2-UiO-66/CdS shown in Fig. 2d reproduced its absorption spectrum, indicating that HER proceeded via photoexcitation. The AQY at 400 nm was 40.3%, which was the highest efficiency relative to previous reports (Supplementary Table 1). These results demonstrated that Pt@NH2-UiO-66/CdS had the highest photocatalytic activity and stability, due to the spatial separation of the photoinduced electrons and holes. To reveal the separation and recombination of these charge carriers, we characterized the photoluminescence (PL), transient photocurrent spectra and electrochemical impedance spectroscopy. Pt@NH2-UiO-66/CdS exhibited a very low fluorescence intensity relative to that for NH2-UiO-66, Pt@NH2-UiO-66 and NH2-UiO-66/CdS (Supplementary Fig. 17). This suggested greatly inhibited recombination of electron and hole pairs in the HNCs34. However, the enhanced photocurrent density of the Pt@NH2-UiO-66/CdS promoted transfer kinetics of the photoexcited charge carriers (Supplementary Fig. 17). Meanwhile, the interfacial properties between the electrode and the electrolyte were also investigated using electrochemical impedance spectroscopy measurements. As known, the semicircle at a high frequency in the Nyquist plot may illustrate the charge-transfer process. The diameter of the semicircle is dependent on the charge-transfer resistance, as shown in Supplementary Fig. 17, the Pt@NH2-UiO-66/CdS exhibits the smallest arc due to its lowest charge-transfer resistance. Meanwhile, the characteristic frequency in the Bode phase plot of the Pt@NH2-UiO-66/CdS film could be estimated a little shift to a lower frequency relative to this of the Pt@NH2-UiO-66 film (Supplementary Fig. 17), further indicating that the charge-recombination rate is reduced in the Pt@NH2-UiO-66/CdS35. These photo-electrochemical properties indicated that Pt@NH2-UiO-66/CdS enhanced separation and migration of the photoinduced charge carriers, which was attributed to its unique spatial-phase separation and hierarchical structure.

Photoinduced carrier dynamics

Time-resolved transient absorption (TA) measurements were performed to track photoinduced carrier dynamics to investigate electron and hole transfers in the mechanism of photocatalytic HER. Figure 3a shows time-resolved TA spectra of NH2-UiO-66 after selective excitation with a 400-nm laser. Band-edge bleaching at less than 450 nm and broad absorption at 600 nm were observed. Fast and slow recovery monitoring at 650 nm is shown in Fig. 3d. Supplementary Table 4 lists the two recovery time constants of 9.8 ps (τ1) and 225 ps (τ2), respectively, which may be attributed to electron-trap states in the intermediate band16,33,36,37. For Pt@NH2-UiO-66, similar features were observed in Fig. 3b, with two decay components of 6.2 ps and 69 ps, respectively. The fast component could be assigned to fast electron transfer from the conduction-band minimum of the MOF to the trap state and then to the Pt sites. The TA characteristics for Pt@NH2-UiO-66/CdS shown in Fig. 3c were bleaching at the 480-nm peak and a broad absorption at 600 nm that was weaker than that of CdS NPs and NH2-UiO-66/CdS (Supplementary Fig. 18). The first component (29 ps) was the fast-trap process, from possible coupled interactions with the hole transfer from NH2-UiO-66 to CdS, and an electron-trap state in NH2-UiO-66, where CdS NPs have no hole-trap states. The latter of CdS NPs is indicated by featureless absorption from the TAS and kinetic profiles at 650 nm in the previous report1,3. The bleaching at 480 nm for Pt@NH2-UiO-66/CdS with the fast-quenching component is shown in Fig. 3e, indicating electron transfers from CdS to NH2-UiO-6638. This feature was also observed in NH2-UiO-66/CdS, with about a 64% decrease in ultrafast quenching (<100 fs), suggesting electron transfers from CdS to NH2-UiO-66. The fast recovery with the 9.3-ps time constant indicated that the electron from NH2-UiO-66 transferred to the Pt phase. The fs TAS suggested that the electron transferred from CdS to NH2-UiO-66, and was then trapped at NH2-UiO-66 before finally transferring to the Pt phase. The hole from NH2-UiO-66 was transferred to CdS. Time-resolved PL spectroscopy was used to investigate the lifetimes of photoinduced carriers in Pt@NH2-UiO-66/CdS33. Figure 3f shows the 450-nm PL kinetics for each sample following excitation at the band-edge absorption. The mean PL lifetimes were 1.97 ns, 0.99 ns, 1.05 ns, and 0.95 ns for NH2-UiO-66, Pt@NH2-UiO-66, NH2-UiO-66/CdS and Pt@NH2-UiO-66/CdS, respectively (Supplementary Table 5). The shorter PL lifetime of Pt@NH2-UiO-66/CdS indicated that the spatial charge-separation of the ternary composites was suppressed by accelerating the radiative recombination of the photoinduced electrons and holes. This strongly indicated that photogenerated carriers transferred in Pt@NH2-UiO-66/CdS could have other pathways of nonradiative recombination for high photocatalytic activity.

The HER mechanism

The photoinduced carrier dynamics for photocatalytic HER by Pt@NH2-UiO-66/CdS HNC are illustrated in Fig. 4. When visible-light irradiated the Pt@NH2-UiO-66/CdS, both phases were excited, the electron from the CdS conduction band was transferred to the conduction band of NH2-UiO-66, and the electron was trapped. Then it was transferred to the Pt surface for HER under visible-light irradiation. The photoinduced hole in NH2-UiO-66 was extracted from the trapped state and transferred to the CdS valence band. Then it migrated to the surface, where it was oxidized by the sacrificial reagent. Thus, the TAS and quenching experiments using hole tracers strongly supported the mechanism of a photogenerated hole-transfer band-trap and spatial charge separation for photocatalytic HER.

Conclusions

We reported a hole-trap transfer pathway in MOF-based ternary HER photocatalysts with spatial-separation structures. The Pt@NH2-UiO-66/CdS photocatalyst separated photogenerated electrons and holes by combining Pt NPs and CdS NPs, which greatly prolonged the lifetime of the hole-trap-mediated pathway and improved the HER photocatalytic efficiency. This work provides a deeper understanding of electron and hole transfer in co-catalyst-NH2-UiO-66-semiconductor ternary composites with spatial-separation structures. Considering their synergistic enhancement of the photocatalytic activity, the results highlight the benefits of fabricating these structures and also advance the development of highly efficient photocatalytic composites.

Methods

Synthesis of NH2-UiO-66

The procedures for synthesizing NH2-UiO-66 were reported previously21. Briefly, a mixture of 5 mL of a N,N-dimethyl formamide (DMF) solution of ZrCl4 (18.64 mg) and 5 mL of a DMF solution of 2-amino-1,4-benzenedicarboxylic acid (NH2-BDC) (14.5 mg) were mixed in a beaker. Then, 1.2 mL of acetic acid was added to the solution and transferred to a 50-mL Teflon-lined stainless-steel autoclave at 120 °C for 24 h. The product was purified via centrifugation and washed with ethanol and hexane. The NH2-UiO-66 was dried overnight at 60 °C under vacuum.

Synthesis of Pt@NH2-UiO-66

The mixture containing ZrCl4 (20 mM, 10 mL) and NH2-BDC (20 mM, 10 mL) was shaken to form a homogeneous solution. Then, a Pt NP solution (70 μL, 30 mg mL−1, ~2.1 mg Pt) and acetic acid (2.74 mL) were added. After sonication for 10 min, the mixed solution was transferred to a Teflon-lined stainless-steel autoclave at 120 °Cfor 24 h. Finally, the products were collected by centrifugation and dried overnight at 60 °C under vacuum.

Synthesis of Pt@NH2-UiO-66/CdS

Typically, 24.3 mg of Cd(CH3COO)2·2H2O was dissolved in 10 mL of ethanol forming a homogeneous solution. Then, 40 mg of Pt@NH2-UiO-66 was added to the solution that was then sonicated for 10 min. The suspension was heated to 80 °C at 6 °C min−1. At this point, 10 mL of an aqueous solution of thioacetamide (6.9 mg) was slowly injected into the flask with the rate of 0.3 mL min−1. It was kept at 80 °C for another 30 min. The precipitates were filtered and washed with water and ethanol several times. Finally, the product was dried at 60 °C under vacuum for 8 h.

Characterization

TEM characterization was performed with a HT7820 (Hitachi, Japan) electron microscope at an acceleration voltage of 120 kV. High-resolution TEM, high-angle annular dark-field scanning TEM, and energy-dispersive spectrometry mapping were performed with a TalosF200x (FEI, USA, equipped with Super-X EDS) electron microscope at an acceleration voltage of 200 kV. An X-ray diffractometer (RigakuD/MAX-2000) with Cu Kα radiation (1.5406 Å) at 40 kV and 30 mA was used to record powder X-ray diffraction patterns. The patterns were collected over a 2θ range of 3°–80° at a scanning speed of 5° min−1. Optical absorbances of all samples were acquired with a Shimadzu UV-1900i spectrophotometer. PL spectra and time-resolved PL spectra were performed on a Hitachi F-77800 fluorescence spectrophotometer. The room temperature PL spectra were recorded with an excitation wavelength of 380 nm. The Pt content was determined via ICP-OES (Vista-MPX, Varian). Electron-paramagnetic-resonance spectra were acquired with a JEOL FA300 spectrometer with a 9.05-GHz magnetic field modulation at a microwave power of 0.998 mW. 5,5-dimethyl-1-pyrroline N-oxide (DMPO) was used as the spin-trapping reagent. Catalyst (5 mg) was suspended in the mixture of water (500 μL) and DMPO (50 μL) for the detection of •OH radicals. After ultrasonication, the detection of •OH in a N2 atmosphere was performed under irradiation with a 300-W Xe lamp with a 420-nm filter for 5 min. Time-resolved fluorescence decay spectra were acquired with an EI FLS-1000 fluorescence spectrometer.

Photocatalytic activity for HER

For photocatalytic HER, the photocatalyst (5 mg) was dispersed in 25 mL of acetonitrile and 3 mL of deionized water with 3 mL of lactic acid. Then the suspension was stirred in a photocatalytic reactor and purged with N2 for 30 min to remove dissolved oxygen, followed by 300-W Xe light irradiation with a UV cutoff filter (>420 nm). Gas chromatography (Shimadzu GC-2014, N2 as a carrier gas) using a thermal conductivity detector was used to measure the HER.

The apparent quantum yield (AQY) in the above photocatalytic reaction conditions was determined. The excitation light was regulated by a bandpass filter. The intensity of the light was measured with a power meter. The AQY was calculated according to the following equation:

$${{{{{\rm{AQY}}}}}} =\frac{{{{{{\rm{number}}}}}}\,{{{{{\rm{of}}}}}}\,{{{{{\rm{reacted}}}}}}\,{{{{{\rm{electrons}}}}}}}{{{{{{\rm{number}}}}}}\,{{{{{\rm{of}}}}}}\,{{{{{\rm{incident}}}}}}\,{{{{{\rm{photons}}}}}}}\times 100 \% \\ \, =\frac{{{{{{\rm{number}}}}}}\,{{{{{\rm{of}}}}}}\,{{{{{\rm{evolved}}}}}}\,{{{{{{\rm{H}}}}}}}_{2}\,{{{{{\rm{molecules}}}}}}\,\times \,2}{{{{{{\rm{number}}}}}}\,{{{{{\rm{of}}}}}}\,{{{{{\rm{incident}}}}}}\,{{{{{\rm{photons}}}}}}}\times 100 \%$$