0.68% of solar-to-hydrogen efficiency and high photostability of organic-inorganic membrane catalyst

Microstructure, topology and photoactivity of CdS@SiO2-Pt photocatalystSiO2 nanolayer was coated on the surface of CdS NRs to prepare the CdS@SiO2 NRs via a sol-gel method (Fig. 1a). Fourier transform infrared (FTIR) spectrum in Supplementary Fig. 1 shows strong absorption peaks assigned to the anti-symmetric stretching vibration of Si-O-Si bonds and bending vibration of Si-O bonds at 1080 cm−1 and 790 cm−1,20. Its high-resolution Si 2p and O 1s XPS spectra in Supplementary Fig. 2a, b respectively show a single characteristic peak at binding energy of 101.35 eV and two characteristic peaks at binding energies of 530.38 eV and 531.74 eV, which can be assigned to Si 2p electron orbital of SiO221 and exposed silicon hydroxyl groups and Si-O bonds22,23. Moreover, its X-ray diffraction (XRD) pattern in Fig. 1b still appear the crystal characteristic of hexagonal CdS (PDF#77-2306)24 without the observation of other crystal phases, and the typical diffraction corresponding to amorphous SiO2 is not observed around 2θ = 24°. Above results indicate that a small amount of SiO2 was successfully coated on the surface of CdS NRs, and transmission electron microscope (TEM) and high-resolution TEM (HRTEM) images in Fig. 1c provide the evidences for this conclusion. Obviously, the CdS@SiO2 composite still appears nanorod-like morphology with smooth surface and increased transverse size (Supplementary Fig. 3), and the lattice fringes of d = 0.32 nm in Supplementary Fig. 4a, b are assigned to the (101) crystal plane of hexagonal CdS25. After SiO2 coverage, a nanolayer with a thickness of ~4.86 nm can be clearly observed from the inset of Fig. 1c, forming a typical core-shell nanorod structure. As shown in Fig. 1d, compared with the bare CdS NRs (0.84 mmol g−1 h−1), a slight decrease of SSL-driven HTH rate (0.84 → 0.63 mmol g−1 h−1) is presented on CdS@SiO2 NRs, which ascribes to the inactive character of SiO2 nanolayer in HTH conversion. However, the unconspicuous photoactivity decrease indicates that the photoexcitons of internal CdS NRs can easily pass through the SiO2 nano-barrier to participate in surface reaction owing to the quantum tunneling effect of photoexcitons26,27.Fig. 1: The preparation, microstructure and photoactivity of CdS@SiO2-Pt photocatalyst.a Preparation diagram of CdS@SiO2-Pt photocatalyst. b XRD patterns and (d), HTH conversion performances of CdS NRs, CdS@SiO2 NRs and CdS@SiO2-Pt photocatalysts with different Pt loading capacities (3.53%, 6.28%, 9.68%). c TEM images of CdS@SiO2 NRs. The inset of (c) is the partial magnification. Error bars in (d) indicate the standard deviation for three measurements.Subsequently, in order to ensure the successful immobilization of Pt NPs on the surface of CdS@SiO2 NRs, its surface was amino-functionalized using 3-aminopropyl triethoxysilane (APTES). FTIR spectrum in Supplementary Fig. 1 shows obviously enhanced absorption peak assigned to the stretching vibration of N-H bonds at 3444 cm−1 20, demonstrating the successful amino-functionalization of CdS@SiO2 NRs. Then, Pt NPs was immobilized on the surface of core-shell CdS@SiO2 NRs through the complexation of amino groups and Pt4+ ions following a chemical reduction (NaBH4), and it can be clearly observed from the TEM and HRTEM images in Fig. 2a, b that a large number of well-dispersed nanoparticles with an average size of ~3 nm appear on the surface of SiO2 nanolayer. The lattice fringes (d = 0.23 nm) and diffraction spots assigned to the (111) crystal plane of face-centered cubic (fcc) phase Pt can be observed from the enlarged HRTEM image of Fig. 2c and selected area electron diffraction pattern of Fig. 2d28,29, and the nanoparticles can be further identified as Pt NPs from the energy dispersive X-ray mapping of Fig. 2e. Therefore, it proves the successful immobilization of Pt NPs on the surface of CdS@SiO2 NRs. Moreover, the loading capacity of Pt NPs on CdS@SiO2 NRs was regulated by changing the dosage of K2PtCl6, so a series of CdS@SiO2-Pt photocatalysts were prepared. The actual Pt loading capacities were determined as 3.53%, 6.28%, and 9.68% by ICP-MS technique, and the corresponding samples were represented as CdS@SiO2-Pt3.53%, CdS@SiO2-Pt6.28% and CdS@SiO2-Pt9.68%, respectively. Accordingly, with the increase of loading capacity of Pt NPs, the partially magnified XRD patterns in Fig. 1b appear gradually increased diffraction peak assigned to the (111) crystal plane of fcc phase Pt (PDF#87-0647) at 2θ = 39.9°30, which is consistent with the result of inductively coupled plasma mass spectrometer (ICP-MS) analysis. Figure 1d shows that loading Pt NPs on the surface of CdS@SiO2 NRs can effectively improve its SSL-driven HTH photoactivity in alkaline condition, which determines that the photoelectrons of internal CdS NRs can easily transfer to the surface Pt NPs through the SiO2 nano-barrier under photoirradiation. With the dosage increase of Pt NPs, the SSL-driven HTH rates of CdS@SiO2-Pt composite photocatalysts show a trend of first increasing and then decreasing (2.98 → 6.24 → 4.47 mmol g−1 h−1). When the loading capacity of Pt NPs is too low (e.g., CdS@SiO2-Pt3.53%, Supplementary Fig. 5a), the number of Pt NPs as the active sites is very small, so its ability to accept photoelectrons from internal CdS NRs is very limited, resulting low photoactivity. On the contrary, when the loading capacity of Pt NPs is too high (e.g., CdS@SiO2-Pt9.68%, Supplementary Fig. 5b), excess Pt NPs is easy to aggregate to form agglomerates due to the high surface energy, which significantly reduces the exposed quantity of active sites, resulting destroyed synergistic effect of cocatalyst and reducing the HTH photoactivity. Especially, the highest SSL-driven HTH rate (6.24 mmol g−1 h−1) was achieved by CdS@SiO2-Pt6.28% composite photocatalyst, which is about 7.4-fold greater of bare CdS NRs. Its high-resolution Pt 4 f X-ray photoelectron spectroscopy (XPS) spectrum in Supplementary Fig. 2c shows two characteristic peaks assigned to Pt 4f7/2 (69.68 eV) and Pt 4f5/2 (73.12 eV) electron orbitals of Pt0, respectively31, indicating that Pt NPs were immobilized on the surface of CdS@SiO2 NRs in zero-valence state.Fig. 2: The topology of CdS@SiO2-Pt photocatalyst.a TEM images of CdS@SiO2-Pt6.28% composite photocatalyst. b HRTEM image, (d), SAED pattern, (e), HAADF image and EDX mappings of CdS@SiO2-Pt6.28% composite photocatalyst. Inset of (b) is the particle size distribution of Pt NPs on the surface of CdS@SiO2-Pt6.28% composite photocatalyst. c Partial magnification of (b) and corresponding line profile of selected area.Photoelectric properties of inorganic catalystThe photoelectric property play an important role in determining the catalytic performance of photocatalysts. UV-Vis absorption spectra in Supplementary Fig. 6 shows that the maximum absorption edge (λmax) of bare CdS NRs blue-shifts to 527 nm from 540 nm after coating SiO2 nanolayer, which is attributed to the non-optical response feature of SiO2 coating. Due to the quantum size of SiO2 nanolayer, the λmax of CdS@SiO2 NRs is very close to bare CdS NRs, and excellent visible-light harvesting capacity is still maintained. Further immobilizing Pt NPs on the surface, obviously red-shifted λmax is observed (527 nm→563 nm), and its absorption capacity is also enhanced significantly in broadband light range (λ > 510 nm), which is attributed to the optical enhancement of synergetic host (CdS)-guest (Pt) interaction32 and surface plasmon resonance effect of Pt NPs33,34. Therefore, the CdS@SiO2-Pt6.28% composite photocatalyst can be confirmed to possess excellent broadband light harvesting capacity. Furthermore, steady-state photoluminescence (PL) spectra (λex = 380 nm) in Supplementary Fig. 7 shows that coating of SiO2 nanolayer significantly reduces the PL emission intensity of bare CdS NRs, indicating effectively suppressed photoexciton recombination kinetics, which is attributed to the comprehensive influence of quantum tunneling effect of photoelectrons and nano-barrier of SiO2 coating. After anchoring Pt NPs, the PL emission intensity is further decreased significantly, which ascribes to that Pt NPs with electron-acceptor character35,36 can rapidly accept photoelectrons tunneling from the internal CdS NRs through SiO2 nanolayer due to lower potential energy, thus greatly suppressing the photoexciton recombination kinetic, and acquiring higher photon utilization efficiency.All electrochemical data were tested on a light-assisted three-electrode system (Fig. 3a). Light-induced electrochemical impedance spectroscopies (EIS) in Supplementary Fig. 8 show slightly increased interfacial impedance of CdS NRs due to the coating of SiO2 nanolayer, but it further decreases significantly with immobilization of Pt NPs. This result powerfully proves that photoexcitons in CdS@SiO2-Pt6.28% composite photocatalyst can easily cross SiO2 nanolayer from internal CdS NRs to the surface Pt active sites due to its quantum tunneling effect. Furthermore, transient photocurrent-time curves in Supplementary Fig. 9 display the intensity trend of CdS@SiO2 < CdS <CdS@SiO2-Pt6.28%, which is basically consistent with the results of UV-Vis absorption spectra and EIS analysis. On the one hand, the SiO2 nanolayer does not significantly affect the photoexciton separation kinetics of CdS NRs under photoirradiation. On the other hand, the electron donor (CdS)-electron acceptor (Pt) interaction greatly promotes the photoelectron migration from internal CdS NRs to surface Pt active sites through the SiO2 nano-barrier, making the catalyst surface a main area for enriching photoelectrons37. In fact, it can be found from the light-induced linear sweep voltammetry curves of Supplementary Fig. 10 and Tafel polarization curves of Supplementary Fig. 11 that the CdS@SiO2-Pt6.28% composite photocatalyst shows lower onset potential (0.91 V vs. −2 mA cm−2) and smaller polarized overpotential (917 mV) for photoelectrochemical hydrogen evolution compared with CdS NRs and CdS@SiO2 NRs, which provides the powerful evidences for the faster HTH kinetics24.Fig. 3: The photoelectric properties and exciton dynamics of inorganic catalyst.a Schematic diagram of the three-electrode electrochemical test system, light density (99.1 mW·cm−2) of the irradiated light source and the working electrodes of inorganic catalyst and membrane catalyst. b 2D mapped TA spectra, (c), TA spectra, (d), attenuated TA spectra of CdS@SiO2-Pt6.28% composite photocatalyst. e, f Decay kinetic curves of CdS NRs and CdS@SiO2-Pt6.28% composite photocatalyst. g Band structure of CdS@SiO2-Pt6.28% composite photocatalyst. The fs-TA tests were performed under excitation of a 380 nm-pump pulse.Exciton dynamics and band structure of CdS@SiO2-Pt6.28% composite photocatalystFemtosecond transient absorption (fs-TA) spectroscopy technique was adopted under excitation of a 380 nm-pump pulse to investigate the exciton dynamics of the optimal CdS@SiO2-Pt6.28% composite photocatalyst. The two-dimensional (2D) mapped TA spectra of CdS NRs and CdS@SiO2-Pt6.28% composite photocatalyst display obvious excited state absorption (ESA) signal at 457 nm (Supplementary Fig. 12a and Fig. 3b), which ascribes to the generation of hot excitons in CdS NRs after excitation38. Compared with CdS NRs, the CdS@SiO2-Pt6.28% composite photocatalyst shows a slightly blue-shifted ground state bleaching (GSB) signal (516 nm→513 nm), indicating that it is easier to reach the excited state under the same excitation39, which is attributed to the synergistic host (CdS NRs)-guest (Pt NPs) electronic interaction. Supplementary Fig. 12b, c and Fig. 3c, d are the TA spectra of CdS NRs and CdS@SiO2-Pt6.28% composite photocatalyst at different relaxation times. It is obvious that the TA intensity of GSB peak increases rapidly in fs range (500 fs) after photoexcitation, and it decreases gradually in subsequent picosecond (ps) and nanosecond (ns) ranges, which ascribes to the instantaneous generation and subsequent recombination of excitons40. Comparatively, the maximum TA intensity of CdS@SiO2-Pt6.28% composite photocatalyst is stronger than that of bare CdS NRs, indicating that more hot excitons are generated under same excitation40. Furthermore, a new GSB signal (~740 nm) can be observed from the TA spectra in relaxation time range of 1−200 ps (Fig. 3c), corresponding to its 2D mapped TA spectrum, which is attributed to the CdS NRs-to-Pt NPs hot electron transfer through SiO2 nanolayer41. Moreover, compared with bare CdS NRs (Fig. 3e), two fast decay processes (τ1 = 14.43 ps, τ2 = 110.02 ps) can be observed in the fitted decay kinetics curve of CdS@SiO2-Pt6.28% composite photocatalyst (Fig. 3f). The former is attributed to the CdS NRs-to-Pt NPs electron transfer through SiO2 nanolayer, while the latter indicates the faster conduction band (CB)-to-trapping state (TS) electron transfer of CdS@SiO2-Pt6.28% composite photocatalyst than CdS NRs. This can be attributed to that the electron-acceptor feature of Pt NPs enables it to quickly accept tunneling electrons from CdS NRs, providing a strong driving force for accelerating exciton transfer42. Moreover, CdS@SiO2-Pt6.28% composite photocatalyst (τ3 = 1348.71 ps) has a longer hot exciton decay lifetime than bare CdS NRs (τ2 = 288.33 ps), demonstrating that the rapid CdS NRs-to-Pt NPs charge transfer ability greatly restrains the photoexciton recombination kinetics43.The flat-band potential of CdS NRs can be determined as Efb(Ag/AgCl) = −1.31 V according to its Mott-Schottky curves at different frequencies (Supplementary Fig. 13), and it can be further converted to −0.70 V vs. reversible hydrogen electrode (RHE) potential by following our previous method24. Since all the linear parts of Mott-Schottky curves show a positive slope, the ECB of CdS NRs can be determined as ECB(RHE) = -0.90 V due to the n-type semiconductor characteristic. Furthermore, its bandgap can be determined as Eg = 2.37 eV from the (ahν)n/2 ~ hν pattern in Supplementary Fig. 14, thereby its valence band position can be confirmed to be EVB(RHE) = 1.47 V. The specific calculation formulas are given in Supplementary Information. Due to the lower chemical potential of surface Pt NPs, the band structure of CdS@SiO2-Pt6.28% composite photocatalyst can be proposed as Fig. 3g.Membrane catalyst and its comprehensive performanceIn order to overcome the drawback of particulate photocatalyst in actual application, the optimal inorganic photocatalyst was compounded with organic ferroelectric PDVF to process into organic-inorganic membrane via a electrospinning technique (Fig. 4a, b). Scanning electron microscope (SEM) images in Fig. 4c show that this membrane is composed by abundant fibers exposing a large number of inorganic materials (Fig. 4d), which benefits to the expose of active sites. As shown in Fig. 4e, pure PVDF membrane can not drive HTH conversion under SSL-irradiation, while all PVDF/CSP membrane catalysts show prominent SSL-driven HTH photoactivity. Moreover, the expose of inorganic catalyst can be well regulated by adjusting its dosage (Supplementary Fig. 15a–f), so as to further control the number of active sites on organic-inorganic matrix. However, it is obvious that excessive inorganic catalyst leads to the aggregation in organic matrix (Supplementary Fig. 15f) due to the uneven dispersion, which will break the synergistic interaction of organic-inorganic interface. Therefore, with the dosage increase of inorganic catalyst, the photoactivities of membrane catalysts showed a trend of increasing first and then decreasing (Fig. 4e). Especially, when the dosage of inorganic catalyst was 0.14 g, the membrane catalyst (PVDF/CSP0.14) achieved the highest HTH rate (160.98 mmol m−2 h−1), and it further reached to 213.48 mmol m−2 h−1 and acquired 0.68% of STH efficiency by regulating the thickness of membrane catalyst (Fig. 4f), which considerably outperforms the existing membrane photocatalysts (Table 1)44,45,46,47,48,49.Fig. 4: The membrane catalyst and its comprehensive performance.a Digital photo of electrospinning equipment. b Schematic diagram of electrospinning technique. c SEM image and (d), EDS mappings of PVDF/CSP0.14 membrane catalyst. e HTH performances of membrane catalysts with different dosage of inorganic catalyst (0 g → 0.16 g) and (f), different thickness (50 μm → 190 μm). g Cyclic HTH performances of PVDF/CSP0.14 membrane catalyst. h Digital photos and SEM images of fresh and recovered PVDF/CSP0.14 membrane catalyst. Inset of (a) is the digital photo of PVDF/CSP0.14 membrane catalyst with the size of 21 cm × 24 cm. Inset of (c) is the partial magnification. Inset of (e) is the light density (212.9 mW·cm-2) used in HTH performance evaluation. Error bars in (e, f) indicate the standard deviation for three measurements.Table 1 HTH performance contrast of PVDF/CSP0.14 membrane catalyst with other reported membrane catalystsThe micro-structural variation and mass loss of particulate catalyst in cyclic process inevitably affect its photoactivity50,51. Satisfyingly, the PVDF networked membrane catalyst (PVDF/CSP0.14) with synergistic organic-inorganic interface maintained a highly stable photoactivity in 50 cycles (Fig. 4g) due to its high structural stability and convenient separation/recovery property. Owing to the limitation of experimental period (hundreds of hours), the photoactivity may still maintain satisfactory stability even continuously increasing the cyclic number. In addition, the photocurrent response remained stable under long-term (5 h) photoirradiation (Supplementary Fig. 16), which also fully illustrates the excellent light stability and well supports the claim of maintaining highly stable photoactivity over 50 cycles. FTIR spectra in Supplementary Fig. 17 shows that the recovered PVDF/CSP0.14 membrane catalyst still display obvious α, β and γ phase characteristics assigned to ferroelectric PVDF52 even being reused dozens of times, and the digital photos and SEM images in Fig. 4h also did not appear obvious variation in the appearance and micromorphology after 50-times of cycles, suggesting the strong regenerability. Therefore, it indicates that the drawback of particulate catalyst in long-term photo-activation was dramatically overcome to better meet practical application. Stress-strain curves in Supplementary Fig. 18 indicate that the compound of inorganic particulate catalyst significantly increase the tensile strength (3.34 MPa→4.11 MPa) and tensile elongation yield (0.65% → 0.88%) of PVDF membrane, suggesting significantly enhanced mechanical strength. Meanwhile, the PVDF/CSP0.14 membrane catalyst possesses smaller interfacial impedance than pure PVDF membrane (Supplementary Fig. 19), indicating that the formation of synergistic organic-inorganic interface significantly promotes exciton transfer. Based on the superior broadband light harvesting capacity of inorganic catalyst and piezoelectric effect of PVDF matrix, the PVDF/CSP0.14 membrane catalyst appears stronger response current signals under transient induction of photoirradiation (Supplementary Fig. 20) or ultrasound (Supplementary Fig. 21) than pure PVDF membrane, suggesting the potential of photo-activation and piezoelectric activation. The strong piezoelectric current and voltage signals in Supplementary Fig. 22a, b and the obvious performance advantage of PVDF/CSP0.14 membrane catalyst than the non-ferroelectric composite membrane (polyurethane (PU)/VSP0.14) in Supplementary Fig. 23 provide the direct evidences to this conclusion.Panel water-splitting reaction system and photocatalytic mechanismSubsequently, a homemade panel water-splitting reaction system was fabricated as Fig. 5a Alkaline water (pH = 14.0) flowed into the panel reaction module embedded membrane catalyst (16 cm × 23 cm) at a flow rate of 20 mL/min under driving of peristaltic pump. When the panel reaction module is filled, it was irradiated by a xenon lamp to drive water splitting. Water-gas mixture flowed into a gas-liquid separation bottle from panel reaction module, and gaseous product was dried in a drying bottle containing concentrated sulfuric acid and subsequently entered a gas collecting bottle as Fig. 5b. As shown as the inset of Fig. 5a and supplemented Supplementary Movie 1, it can be observed from the panel reaction module that a large number of bubbles are continuously evolved on the surface of membrane catalyst under SSL-irradiation. With continuous reaction, bubbles were released in drying bottle (Fig. 5c), and supplemented Supplementary Movie 2 provides direct evidence for this phenomenon. After 3 h of reaction, gas sample was extracted from the sampling port of gas collecting bottle for gas chromatographic detection, and chromatographic peaks belonging to hydrogen and oxygen appeared as Fig. 5d, e, indicating the successful progress of water splitting process in this panel reaction system. The ignition experiment of collected gas in inset of Fig. 5e and Supplementary Movie 3 further provides the evidence for this conclusion, and 0.05% of STH efficiency was achieved on this panel reaction system for water splitting. Due to the high stability of this membrane catalyst, no active catalyst lost from the panel reaction module even suffering long-term mechanical forces of fluid flow and bubble collision. By increasing the number of panel modules, the scale of panel reaction system can be expanded as Supplementary Fig. 24, so that water splitting can be achieved under strong outdoor solar-irradiation. Of course, there are still many engineering problems to be solved in actual operation. Based on our study, the catalytic mechanism of this membrane catalyst for HTH conversion is proposed as Supplementary Fig. 25. Owing to the excellent broadband light-harvesting capacity of inorganic catalyst and sensitive piezoelectricity of PVDF matrix, the membrane catalyst in panel reaction module can be double-activated under stimulations of light energy and mechanical energy, thus generating high-density excitons (e− + h+) to drive HTH conversion (1). Due to the presence of hydroxyl groups on the surface of SiO2 nanolayer and the influence of a large number of OH− ions in aqueous phase, a low chemical potential conducive to the transfer of photogenerated h+ from CdS NRs to the surface of SiO2 nanolayer is formed, and related oxidation reactions can be achieved on the surface of the catalytic material, effectively avoiding the impact of corrosion on the internal CdS microstructure. The relevant processes are shown as follows: (1)~(4). In this catalytic process, OH− ions consume most of the photogenerated h+ (4), thus better promoting the half-reaction of hydrogen production (3).$${{{{\rm{Cat}}}}}_{{{{\rm{Membrane}}}}}+{{{\rm{h}}}}{{{\rm{\nu }}}}\to {{{{\rm{Cat}}}}}_{{{{\rm{Membrane}}}}}({{{{\rm{e}}}}}^{-}+{{{{\rm{h}}}}}^{+})$$
(1)
$${{{{\rm{H}}}}}_{2}{{{\rm{O}}}}+{{{{\rm{Cat}}}}}_{{{{\rm{Membrane}}}}}\to {{{{\rm{H}}}}}^{+}+{{{{\rm{OH}}}}}^{-}$$
(2)
$${{{{\rm{H}}}}}^{+}+{{{{\rm{e}}}}}^{-}\to {{{{\rm{H}}}}}_{2}$$
(3)
$${{{{\rm{OH}}}}}^{-}+{{{{\rm{h}}}}}^{+}\to {{{{\rm{O}}}}}_{2}+{{{{\rm{H}}}}}_{2}{{{\rm{O}}}}$$
(4)
Fig. 5: The panel water-splitting reaction system and photocatalytic mechanism.a Digital photos of laboratorial panel reaction system based on the membrane catalyst: b gas collecting bottle and (c), bubble display in gas drying bottle. d Hydrogen standard curve obtained on a gas chromatograph (FULI, GC 9790II, CN). e Gas chromatogram of gaseous product from this panel reaction system.In summary, in order to overcome the drawback of particulate photocatalysts in flat-panel HTH conversion, a highly active CdS@SiO2-Pt composite with rapid CdS-to-Pt electron-transfer kinetics and restrained photoexciton recombination kinetics was prepared to process into an organic-inorganic membrane by compounding with organic ferroelectric PVDF. This PVDF networked membrane catalyst with synergistic organic-inorganic interface displays high photostability and excellent operability, achieving improved SSL-driven alkaline (pH = 14.0) HTH activity (213.48 mmol m−2 h−1) following a 0.68% of STH efficiency. No obvious variation in its appearance and micromorphology was observed even being recycled for 50-times, which considerably outperforms the existing membrane photocatalysts. Subsequently, a homemade panel reaction system was fabricated to achieve alkaline water-splitting to obtain a 0.05% of STH efficiency under SSL-irradiation. This study opens up a prospect for practical application of panel photocatalytic hydrogen production with organic-inorganic interface networked membrane technology.