Piezoelectric Photocatalytic BaBi₄Ti₄O₁₅@COFs /PVDF‐TrFE Membrane With Enhanced Built‐In Electric Field for Hydrogen Production

ABSTRACT

Piezoelectric photocatalytic water splitting hydrogen production technology has become a very promising solution to address the growth of global energy demand and the low‐carbon transformation. To overcome the challenges of powder catalyst recovery and secondary pollution, we designed a piezoelectric photocatalytic mixed matrix membrane (BBTNC‐MMM) using BaBi₄Ti₄O₁₅@COFs (BBTNC) S‐scheme heterojunction as the functional filler within the PVDF‐TrFE (PT) matrix membrane. The experimental results showed that the piezoelectric response output voltage and hydrogen production of BBTNC‐MMM were 3.8 V and 2072.3 µmol h⁻¹ m⁻² respectively, which were more than 3 times and 36 times higher than those of the pure PT membrane. COMSOL simulation showed that the BBTNC not only effectively enhanced the internal electric field, promoting the separation of photogenerated carriers, but also had abundant active sites on its surface. Those promoted the improvement of catalytic efficiency. Meanwhile, the porous membrane structure ensured that sufficient catalytic active sites were exposed to the aqueous environment and provided a path for the transport of reactants/products. This research successfully developed a piezoelectric photocatalytic membrane, providing a new direction for sustainable hydrogen production and piezoelectric response sensing applications.

The BaBi₄Ti₄O₁₅@COFs (BBTNC) S‐scheme heterojunction was embedded on the porous PVDF‐TrFE substrate membrane. BBTNC not only effectively enhanced the built‐in electric field strength, but also provided rich catalytic active sites. The constructed porous membrane provided the necessary interface and mass transfer conditions for catalytic reactions, achieving a significant increase in hydrogen production and piezoelectric output voltage.

1. Introduction

With the sharp growth of the economy and population, energy demand has risen in modern society [1, 2]. At present, most of the energy comes from the combustion of fossil fuels, which not only causes serious harm to the environment, but also increases people's concerns due to fossil fuel non‐renewability and irregular geographical distribution [3]. Therefore, it is vital to develop a sustainable, green, and cost‐effective alternative energy source to meet the growing energy demand [4, 5]. As a clean and renewable energy source with high energy density (140 MJ kg⁻¹), hydrogen (H₂) has emerged as a key solution to address today's global energy and climate challenges [6, 7]. In addition to traditional H₂ production methods involving coal gasification and steam reforming, water splitting H₂ production technologies (such as photocatalysis, photo‐electrocatalysis, electrocatalysis, thermal catalysis, and piezo‐photocatalysis) have been widely studied [8, 9]. Among them, the piezoelectric‐photocatalysis technology has become a very promising method, which can utilize mechanical energy and solar energy jointly to promote the occurrence of hydrogen evolution reaction (HER) [9, 10]. It has been demonstrated that localized polarization charges induced via the piezoelectric effect under mechanical force can significantly promote photogenerated carriers (e⁻ and h⁺) separation and enhance catalytic H₂ production [11, 12, 13].

Furthermore, the construction of heterojunctions is also an effective way to enhance catalytic efficiency [14, 15]. For instance, Guo et al. [16] constructed a heterostructure of BaTiO₃@MoSe₂ with the piezoelectric photocatalytic H₂ production of 2196 µmol h⁻¹ g⁻¹. Currently, Z‐scheme and S‐scheme heterojunctions, formed by two semiconductors with staggered bandgaps, have become research hotspots in the field of catalysis because they can enhance the separation efficiency of photogenerated carriers and retain high redox capabilities [17, 18]. The Z‐scheme, inspired by natural photosynthesis, operates by allowing photogenerated electrons from the conduction band (CB) of one semiconductor to recombine with holes from the valence band (VB) of the other [19]. This process leaves behind electrons and holes with stronger reduction and oxidation potentials, respectively [19, 20]. In contrast, the S‐scheme heterojunction composed of reducible photocatalyst (RP) and oxidized photocatalyst (OP), utilizes the built‐in electric field to facilitate the recombination of carriers, effectively retaining the highly reductive electrons in RP and highly oxidative holes in OP [21, 22].

In addition, with the recent development of porous framework materials, the application of covalent organic frameworks (COFs) have received extensive research [23]. COFs have become promising catalysts with advantages such as designable periodic ordered structure, suitable band structure, and high specific surface area [24, 25]. It is shown that the heterostructure between metal oxides and COFs can facilitate the separation of photogenerated carriers, thus improving the catalytic efficiency [26, 27]. Xu et al. [28] combined piezoelectric materials BiFeO₃ and COFs to form a Z‐scheme heterostructure for efficient piezoelectric photocatalytic water splitting. The obtained BiFeO₃@COFs exhibited the piezoelectric photocatalytic H₂ and O₂ production of 1416.4 and 708.2 µmol h⁻¹ g⁻¹.

Moreover, most of the research on piezoelectric‐photocatalysis focuses on piezoelectric powders with problems such as difficult recycling and environmental pollution, which greatly limit their practical application in piezoelectric‐photocatalysis [29, 30]. It is well known that organic polymers, polyvinylidenefluoride (PVDF) and its copolymers (such as poly(vinylidenefluoride‐trifluoroethylene) (PVDF‐TrFE) and poly(vinylidenefluoride‐hexafluoropropylene) (PVDF‐HFP)), are polymer materials with excellent piezoelectric properties, mechanical properties, flexibility, and chemical stability, which can be used as an excellent substrate material for piezoelectric‐photocatalytic membranes in the field of H₂ production [31, 32]. Su et al. [29] synthesized SnFe₂O₄/ZnIn₂S₄/PVDF piezoelectric‐photocatalytic membrane with the H₂ production of 1652.7 µmol h⁻¹ g⁻¹. Wei et al. [33] designed PVDF/Fe₃O₄@g‐C₃N₄ (PVFC) spiral micromotors with the H₂ production of 42.3 µmol h⁻¹. Although the inorganic–organic polymer membrane have shown good catalytic activity, these systems are still in their infancy [34].

Herein, we prepared a piezoelectric photocatalytic mixed matrix membrane (BBTNC‐MMM) with an S‐scheme heterojunction material as the functional filler and PVDF‐TrFE as substrate through phase inversion method to achieve water splitting H₂ production shown in Figure 1. Specifically, we designed a sheet‐like BaBi₄Ti₄O₁₅ (BBT) @COFs S‐scheme heterostructure (BBTNC), which leveraged inherent advantages of COFs. This not only greatly exposed the catalytic active sites but also enabled the constructed heterojunction to efficiently promote the separation of photogenerated carriers. Meanwhile, the simulation results confirmed that the construction of heterojunctions could significantly enhance the piezoelectric potential of the material. Subsequently, BBTNC was doped as a functional component into the PVDF‐TrFE substrate to construct BBTNC‐MMM with rich channels. This porous membrane structure ensured that sufficient catalytic active sites were exposed to the aqueous environment, while providing a smooth path for the efficient diffusion and transport of reactants/products, achieving efficient catalytic processes. The experimental results demonstrated that the BBTNC‐MMM exhibited significantly enhanced piezoelectric response and photocatalytic activity compared to the pure PVDF‐TrFE membrane. This provided a new strategy for designing high‐performance piezoelectric photocatalytic membranes.

FIGURE 1.

Preparation of BBTNC and BBTNC‐MMM materials.

2. Materials and Methods

2.1. Materials

1,4‐Dioxanewas (99.8%), triethanolamine (TEA, 98.0%), acetic acid (99.5%), N, N‐dimethylformamide (DMF, 99.8%), (3‐aminopropyl) triethoxysilane (APTES, 99.0%), methanol (99.5%), and ethanol (99.5%) were obtained from Beijing InnoChem Science & Technology Co., Ltd (Beijing, China). Bismuth oxide (Bi₂O₃, 99.8%), titanium dioxide (TiO₂, 99.8%), sodium chloride (NaCl, 99.5%), barium carbonate (BaCO₃, 99.0%), potassium chloride (KCl, 99.5%), barium chloride dihydrate (BaCl₂·2H₂O, 99.5%), 5,5‐dimethyl‐1‐pyrroline N‐oxide (DMPO, 97.0%), and p‐phenylenediamine (Pa, 99.0%) were provided by Shanghai Aladdin Biological Technology Co., Ltd (Shanghai, China). 1,3,5‐triformylphloroglucinol (Tp, 95.0%) was supplied from Chinese Academy of Sciences‐Yanshen Technology Co., Ltd (Jilin, China). Poly(vinylidene fluoride‐trifluoroethylene) (PVDF‐TrFE, 99.0%) was purchased from Dongguan Zhongcheng Plastic Products Co., Ltd (Guangdong, China). All the above reagents were used directly. Deionized water was produced by ultra‐pure water purification equipment from Chongqing Ever Young Enterprise Devel Opment CO., Ltd (Chongqing, China) with a conductivity less than 5 µS cm⁻¹.

2.2. Preparation of Sheet Barium Bismuth Titanate (BBT) Materials

As shown in Figure S1, the sheet BaBi₄Ti₄O₁₅ (BBT) was synthesized by the two‐step molten salt method according to the following formulas:

$$2\mathrm{B}{\mathrm{i}}_2{\mathrm{O}}_3+3\mathrm{Ti}{\mathrm{O}}_2\to \mathrm{B}{\mathrm{i}}_4\mathrm{T}{\mathrm{i}}_3{\mathrm{O}}_{12}$$ (1)

$$\mathrm{B}{\mathrm{i}}_4\mathrm{T}{\mathrm{i}}_3{\mathrm{O}}_{12}+\mathrm{BaC}{\mathrm{O}}_3+\mathrm{Ti}{\mathrm{O}}_2\to \mathrm{BaB}{\mathrm{i}}_4\mathrm{T}{\mathrm{i}}_4{\mathrm{O}}_{15}+\mathrm{C}{\mathrm{O}}_2\uparrow$$ (2)

First, TiO₂ and Bi₂O₃ powders (a molar ratio of 3:2) were mixed with NaCl‐KCl (a molar ratio of 1:1) molten salt. After being sintered at 1100°C for 2 h, the mixture was washed with deionized water and then dried in an oven at 120°C to obtain Bi₄Ti₃O₁₂ materials with bismuth lamellar structure (Figure S1a). Subsequently, the mixture of Bi₄Ti₃O₁₂, TiO₂, and BaCO₃ (a molar ratio of 1:1:1) was added into the BaCl₂·2H₂O‐KCl (a molar ratio of 1:1) molten salt. It was sintered at 1080°C for 2 h. After being washed with deionized water and then dried at 120°C in an oven, the sheet BBT material was finally obtained (Figure S1b).

2.3. Preparation of BBT@COFs Materials (BBTNC) and COFs Materials

The BBT@COFs materials were fabricated by solvothermal synthesis in Figure 1a. 500 mg of BBT was dispersed in 50 mL of ethanol. Then, 10 mL of APTES/ethanol solution (0.5 vol%) were slowly added into the dispersion and stirred at 25°C for 2 h. Subsequently, BBT‐NH₂ was obtained by washing with ethanol and vacuum drying at 60°C. BBT‐NH₂ (300 mg) and different masses of Pa (75 mg, 150 mg, 225 mg, and 300 mg) were dispersed in 72 mL of 1, 4‐dioxane, respectively. Then Tp (Tp and Pa with a molar ratio of 2:3) and acetic acid (12 mL) were added to the above dispersion, respectively. The dispersion was poured into a 100 mL Teflon high‐pressure reactor and heated at 90°C for 72 h in an oven. Subsequently, BBTNC materials were obtained by washing with ethanol and vacuum drying at 60°C. In addition, COFs materials were fabricated in the same way without BBT‐NH₂ in the solution.

2.4. Preparation of BBTNC/PVDF‐TrFE Mixed Matrix Membranes (BBTNC‐MMMs)

BBTNC/PVDF‐TrFE mixed matrix membranes were prepared by phase inversion method as shown in Figure 1b. For a typical procedure, PVDF‐TrFE (1.6 g) and BBTNC (100 mg) were dispersed in DMF (8.3 g), and then stirred for 2 h at 45°C. The obtained dispersion was coated onto a clean glass plate using a scraper, and then immediately immersed in deionized water to obtain the BBTNC‐MMM. By altering the amount of BBTNC (while keeping the quality of PVDF‐TrFE and the dispersion unchanged), a series of BBTNC‐MMMs with concentrations of 0.5, 1.5, and 2.0 wt% were prepared, respectively. Similarly, a series of BBTNC‐MMMs with PVDF‐TrFE concentrations of 14, 18, and 20 wt% were prepared using the same method, respectively.

In addition, BBT/PVDF‐TrFE mixed matrix membrane (BBT‐MMM), COFs/PVDF‐TrFE mixed matrix membrane (COFs‐MMM), and PVDF‐TrFE membrane (PT membrane) were fabricated by the same method. The detailed fabrication procedure could be found in the Supporting Information.

2.5. Catalytic H₂ Production Measurements and Samples Characterization

The piezoelectric‐photocatalytic H₂ production was performed using a 300 W xenon lamp simulating full‐spectrum solar light and an ultrasonic cleaner (40 kHz, 150 W power) as the mechanical vibration. 10 membranes about 0.6 cm² in size were directly immersed in 100 mL of solution. Prior to the reaction, the piezoelectric‐photocatalytic reactor was deoxygenated by blowing N₂ for 1 h. Finally, 200 µL gas from the reaction system was extracted every 1 h to determine the H₂ production by a gas chromatograph system (GC9700II). All of experiments were conducted 3 times and the average value was taken as the data of piezoelectric‐photocatalytic activity to ensure the accuracy of the data, while the reaction environment was maintained at a constant temperature by a water circulation system to eliminate the influence of the thermal effect. The detailed samples characterization could be found in the Supporting Information.

3. Results and Discussion

3.1. Different Physicochemical Properties of the Synthesized Samples

The crystal structure of the synthesized samples was analyzed via X‐ray diffractometer (XRD). As shown in Figure S2, the synthesized BBT was an orthorhombic structure with the space group A2₁am (JCPDS No. 04‐016‐4195) [35, 36]. In addition, no diffraction peaks corresponding to the raw materials (BaCO₃, Bi₂O₃, and TiO₂) were detected in the XRD pattern of the BBT, which confirmed that these raw materials had completely transformed into pure BBT phase and no residual impurities were detected. To study COFs stacking modes, three possible stacking modes (including AA, AB, and ABC) were constructed. As shown in Figure S3a–c, the experimental XRD pattern of COFs was consistent with the simulated pattern of the AA stacking mode, indicating that the AA stacking framework of COFs was successfully constructed. For COFs, diffraction peaks at 4.6°, 8.0°, and 26.8° were attributed to the (1 0 0), (1 1 0), and (1 1 1) planes in Figure 2a and Figure S3d, respectively [28]. The lattice parameters obtained by Pawley refinement were determined to be a = b = 22.81 Å, c = 3.53 Å, α = β = 90° and γ = 120° with refinement parameters of R _wp = 4.75% and R _p = 3.37% (Figure S3d). The XRD pattern of BBTNC exhibited the characteristic peaks of both COFs (at 4.6°) and BBT (at 30.0°, 32.8°, 56.7°, etc.), demonstrating the successful formation of the BBTNC shown in Figure 2a. Furthermore, compared the XRD pattern of PT membrane with that of BBTNC‐MMM, it showed that BBTNC was successfully added to PVDF‐TrFE.

FIGURE 2.

(a) XRD patterns, (b) Raman spectra, and (c) FT‐IR spectra of PT membrane, BBTNC‐MMM, BBTNC, BBT, and COFs. XPS spectra of (d) survey plot in BBTNC and BBT, (e) C 1s, and (f) N 1s in BBTNC, (g) Ba 3d, (h) Bi 4f, (i) Ti 2p, and (j) O 1s in BBT and BBTNC tested in darkness and under illumination.

The surface group of the prepared samples was analyzed by Raman spectrometer and fourier transform infrared (FT‐IR) spectrometer. As shown in Figure 2b, vibrational mode of BBT was observed the presence of Ba─O, Bi─O, and Ti─O, which indicated that synthesized BBT had the characteristic of an orthogonal structure [37]. Among them, vibrational mode located at wave numbers of 887.5 cm⁻¹ was related to the symmetrical stretching of TiO₆ octahedral [38]. Notably, Raman shifts below 200 cm⁻¹ were associated with perovskite A‐site cations (Ba²⁺ and Bi³⁺), reflecting the BBT ferroelectric properties [39]. The peak at 280.3 cm⁻¹ corresponded to O─Ti─O bending vibration [37, 38]. In addition, the spectra of BBTNC and COFs displayed three strong peaks at 1177.3, 1399.6, and 1612.8 cm⁻¹, homologous to the C‐Harom from benzene rings, D and G bands, respectively [40]. The vibrational mode of BBTNC located at wave numbers of 280.3 cm⁻¹ was related to Ti─O bonds, indicating the successful synthesis of BBTNC. Meanwhile, BBTNC‐MMM showed peaks similar to BBTNC, which were not observed in PT membrane, indicating that the BBTNC‐MMM was successfully prepared.

As illustrated in Figure 2c, the FT‐IR spectra of obtained BBTNC and COFs showed characteristic stretching bands of ─C═C and ─C─N at near 1590 and 1250 cm⁻¹, respectively, confirming that COFs was successfully loaded on BBT. Furthermore, the characteristic band at about 1590 cm⁻¹ on the FT‐IR spectrum of BBTNC‐MMM indicated that BBTNC was successfully incorporated into the PVDF‐TrFE membrane through the phase inversion method.

The atomic composition of the samples was revealed by X‐ray photoelectron spectroscopy (XPS). As displayed in Figure S4a, it could be seen that the XPS spectra of BBTNC‐MMM had very similar peaks with PT membrane, which was due to the significant influence of the PVDF‐TrFE substrate on the surface atomic composition of BBTNC‐MMM. PT membrane and BBTNC‐MMM showed four peaks at binding energies of 290.8, 288.2, 286.4, and 284.8 eV, which were identified as CF₂, CFH, CH₂, and CH_x groups, respectively (Figure S4b) [41]. The XPS survey spectrum of BBT exhibited the presence of Ba, Bi, Ti, and O, while that of the BBTNC showed the existence of C, N, O, Ba, Bi, and Ti (Figure 2d). The XPS peaks of C 1s at 288.9, 286.5, and 284.8 eV were designated to C═O, C─N, and C─C/C═C in BBTNC, respectively (Figure 2e) [42]. Figure 2f shows the deconvoluted peaks at 403.9, 400.4, and 400.0 eV were indexed into N─O, C═N, and C─N for the N 1s spectrum in BBTNC, respectively [28]. For the XPS spectra of BBT, the binding energies were located at 794.6, 779.3, 159.4, 164.7, 458.3, and 465.6 eV, matched with Ba 3d_7/2, Ba 3d_5/2, Bi 4f_7/2, Bi 4f_5/2, Ti 2p_3/2, and Ti 2p_1/2 in Figure 2g–i, respectively [37, 38, 39]. As illustrated in Figure 2j, the O 1s spectrum of BBTNC could be resolved into three peaks of 535.5, 533.2, and 530.9 eV, attributing to chemisorbed oxygen, oxygen vacancy, and lattice oxygen, respectively [28, 37]. It was worth noting that Ba 3d, Bi 4f, Ti 2p, and O 1s binding energies in BBTNC in darkness were shifted toward lower energy levels compared with those in pristine BBT. This indicated that the interface of BBT and COFs existed strong interaction and electrons were transferred from COFs to BBT [28, 43]. Under illumination, the binding energies of Ba 3d, Bi 4f, Ti 2p, and O 1s in BBTNC were positively shifted. This indicated that under illumination, the photogenerated electrons transfer from BBT to the COFs [44]. The electron transfer indicated the formation of the S‐scheme heterostructure between BBT and COFs, which could promote the formation of built‐in electric field, thereby enhancing catalytic activity [43, 44, 45].

To further illustrate the generation of the built‐in electric field in BBTNC, the surface charge densities of BBT and BBTNC were detected by Kelvin probe force microscopy (KPFM) in Figure 3. As shown in Figure 3c, the surface potential of BBTNC was about 120 mV, which was more than twice that of BBT (≈ 50 mV). The built‐in electric field intensity of the material increases proportionally with its surface charge density [46]. A stronger built‐in electric field was formed inside BBTNC, significantly promoting the separation of photogenerated carriers and thereby enhancing the catalytic efficiency [47].

FIGURE 3.

The surface KPFM potential images of (a) BBT and (b) BBTNC, and (c) the surface potential curves of BBT and BBTNC.

In addition, Figure S5 shows the thermogravimetric analysis (TGA) result of the samples. It declared that BBT, COFs, and BBTNC all maintained excellent thermal stability up to 410°C. Subsequently, the N₂ adsorption–desorption isotherms of prepared materials were measured. BBT exhibited a nearly horizontal and hysteresis‐free isotherm, confirming its non‐porous materials in Figure S6a,d. As depicted in Figure S6b,c, the curves of BBTNC and COFs conformed to the type‐I isotherm, indicating that there were interconnected micropores in the structures of BBTNC and COFs [42]. The pore diameters of both COFs and BBTNC were ≈ 1.8 nm, as shown in Figure S6e,f. Consistent with this, pure BBT showed the minimal BET surface area (0.47 m² g⁻¹), and due to the introduction of non‐porous BBT, BBTNC exhibited a lower surface area (225.17 m² g⁻¹) than COFs (424.92 m² g⁻¹), as detailed in Table S1. This indicated that the introduction of COFs was conducive to increasing catalytic activity of materials [28].

3.2. Morphological Observation of Samples

The morphological characteristics of prepared BBT and BBTNC powders were analyzed respectively by SEM and TEM. As illustrated in Figure S7, prepared BBT had a smooth surface and maintained a sheet‐like morphology with a size of 3–10 µm. It could clearly find that BBT was wrapped by a COFs shell layer with a thickness of approximately 120 nm, while the surface of BBT also displayed a large number of flower cluster‐like COFs from the SEM and TEM images of BBTNC (Figure 4a–c and Figure S8). As shown in Figure S9, the surface of BBTNC was rougher, with a Ra value of 28.8 nm, while the surface of BBT was smooth, with a Ra value of 7.8 nm. Moreover, the interplanar spacing of 0.27 nm corresponded to the (2 0 0) lattice planes of BBT in Figure 4d.

FIGURE 4.

(a) TEM, (b) TEM‐EDS, (c) enlarged TEM, and (d) HR‐TEM images of BBTNC, (e) SEM surface and (f) SEM cross‐sectional images of PT membrane, (g) SEM surface and (h) SEM cross‐sectional images of BBTNC‐MMM, and (i) TEM, (j) TEM‐EDS, and (k) HR‐TEM images of BBTNC‐MMM.

Subsequently, the surface and cross‐sectional morphologies of BBTNC‐MMM and PT membrane were analyzed by SEM. As shown in Figure 4e, the surface of the PT membrane was wrinkled without notable micropores. As illustrated in Figure 4g, there were plate‐structured particles on the membrane surface, indicating that the BBTNC particles have been successfully added into PT substrate membrane. This could also be obtained from the SEM‐EDS of the membrane cross‐section (Figure S10). In Figure 4f,h, the cross‐section images of PT membrane and BBTNC‐MMM revealed the membrane thickness of approximately 90 µm. In addition, the interiors of both the PT membrane and BBTNC‐MMM were porous structures, with finger‐shaped pores on one side and reticular pores on the other side. This porous structure allowed the BBTNC integrated into the membrane to play a crucial role in the subsequent piezoelectric photocatalytic water splitting H₂ production. Moreover, the surface morphology of BBTNC‐MMM were observed by TEM. Similar to the SEM images of BBTNC‐MMM, Figure 4i,j shows that BBTNC had been successfully incorporated into PT substrate. As shown in Figure 4k, the interplanar spacing of 0.30 nm corresponded to the (1 1 9) lattice planes of BBT.

3.3. Piezoelectric Response Properties of the BBTNC‐MMM

As shown in Figure 5a,b, piezoelectric response sensors were constructed using BBTNC‐MMM and PT membrane with a surface area of about 10 cm² to prove the piezoelectricity of membranes, respectively. Specifically, copper foil tape was used as the electrode and attached to both sides of the membrane. Then, copper wires were connected to both ends, and finally, it was wrapped with PET tape. Subsequently, the piezoelectric response characteristics of BBTNC‐MMM and PT membrane were detected using the equipment built by the laboratory (Figure S11). The electrical signal outputs of the prepared BBTNC‐MMM and PT membrane were tested respectively by tapping with the same force. The voltages of PT membrane and BBTNC‐MMM were approximately 1.2 and 3.8 V, respectively (Figure 5c,d). Doping of BBTNC significantly improved the piezoelectric response performance of the membrane. As stress concentrators, BBTNC could effectively convert mechanical energy into charges under external pressure, thereby enhancing the built‐in electric field strength of the membrane [48]. Hence, the output voltage of the BBTNC‐MMM had increased by more than 3 times compared to the PT membrane. This indicated that the incorporation of BBTNC could significantly improve the piezoelectric performance of PT membrane, making BBTNC‐MMM a potential candidate for piezo‐catalytic applications.

FIGURE 5.

(a) The design schematic diagram and (b) the physical diagram of the piezoelectric response sensors based on BBTNC‐MMM and PT membrane, and generated voltage of devices based on (c) PT membrane and (d) BBTNC‐MMM.

3.4. The Piezoelectric‐Photocatalytic H₂ Production Performance of Mixed Matrix Membranes and PT Membrane

The piezoelectric‐photocatalytic H₂ production performances of BBTNC‐MMMs in triethanolamine solution were tested. Figure 6a shows the influence of COFs concentrations on the efficiency of BBTNC‐MMMs piezoelectric‐photocatalytic H₂ production. When the mass ratio of Pa to BBT was 0.5:1, BBTNC‐MMM had the highest piezo‐photocatalytic efficiency and the H₂ production reached 928.4 µmol h⁻¹ m⁻². However, with the increase of COFs concentrations, the H₂ production efficiency of BBTNC‐MMMs decreased, which might arise from effects related to excess COFs, such as shielding and unnecessary recombination [28].

FIGURE 6.

(a) Piezo‐photocatalytic H₂ production of BBTNC‐MMM with different concentrations of (a) COFs, (b) BBTNC, and (c) PVDF‐TrFE, (d) H₂ production of BBTNC‐MMM with different coating thicknesses of membrane, (e) H₂ production of PT membrane, BBT‐MMM, COFs‐MMM and BBTNC‐MMM, and (f) Cycling runs of BBTNC‐MMM for piezo‐photocatalytic H₂ evolution.

Meanwhile, a series of BBTNC‐MMMs with different BBTNC concentrations (0.5, 1.0, 1.5, and 2.0 wt%) were fabricated. The H₂ production performance was shown in Figure 6b. As the concentration of BBTNC increased from 0.5 wt% to 2.0 wt%, the H₂ production of BBTNC‐MMMs increased from 928.4 to 1335.9 µmol h⁻¹ m⁻², then remained basically unchanged, and finally decreased to 933.3 µmol h⁻¹ m⁻². It was possible that the high concentration of BBTNC caused a large accumulation of BBTNC in the membrane, which limited the transport of reactants and products, thereby leading to a decline in catalytic efficiency [29]. In addition, since the H₂ production was basically the same when the BBTNC concentration was 1.0 and 1.5 wt%, considering the economy, the concentration of BBTNC in the subsequent membrane production process was selected as 1.0 wt%.

Subsequently, the influence of the concentration of PVDF‐TrFE on the piezoelectric‐photocatalytic H₂ production performance of membranes was explored, as shown in Figure 6c. With the increase of PVDF‐TrFE concentration, the H₂ production remained basically unchanged (approximately 1300 µmol h⁻¹ m⁻²) at first and then decreased. This might be because as the concentration of PVDF‐TrFE increased, the catalytic active sites within the membrane decreased, eventually leading to a decline in performance [29]. Furthermore, the influence of membrane thickness on performance was also studied. As shown in Figure 6d, when the coating thickness of the membrane was 500 µm, the production of H₂ reached its peak, which was 2072.3 µmol h⁻¹ m⁻².

To sum up, the optimal preparation process of the membrane was as follows: the BBTNC concentration was 1.0 wt%, the PVDF‐TrFE concentration was 16 wt%, and the coating thickness of the membrane was 500 µm. BBT‐MMM and PT membrane were fabricated by the above‐mentioned process and the H₂ production of BBTNC‐MMM, BBT‐MMM and PT membrane were compared. As shown in Figure 6e, the H₂ production of membranes followed the order of PT membrane (56.3 µmol h⁻¹ m⁻²) < BBT‐MMM (497.0 µmol h⁻¹ m⁻²) < COFs‐MMM (694.5 µmol h⁻¹ m⁻²) < BBTNC‐MMM (2072.3 µmol h⁻¹ m⁻²). Importantly, the H₂ production of BBTNC‐MMM was 36.8 times that of the PT membrane, which suggested that the incorporation of BBTNC into PT membrane could enhance the piezoelectric‐photocatalytic activity.

To illustrate the synergistic effect of piezoelectric photocatalysis, the piezoelectric catalytic and photocatalytic H₂ production of BBTNC‐MMM were evaluated using the same method. As shown in Figure S12, the piezoelectric catalytic and photocatalytic H₂ production of BBTNC‐MMM were 446.6 and 1265.3 µmol h⁻¹ m⁻², respectively, which were much lower than its piezoelectric photocatalytic H₂ production. The reason for this phenomenon was that the built‐in electric field generated by the piezoelectric effect could effectively promote the separation of electron‐hole pairs, thereby leading to an increase in catalytic efficiency.

In addition, the piezoelectric photocatalytic H₂ production of BBTNC‐MMM was systematically evaluated in different sacrificial agents and water media in Figure S13. In the 20 vol% methanol aqueous solution, the H₂ production of BBTNC‐MMM reached the highest of 2330.2 µmol h⁻¹ m⁻², significantly higher than that in the 20 vol% TEA aqueous solution (2072.3 µmol h⁻¹ m⁻²). This might be due to the smaller methanol molecules penetrating the membrane pores more easily to reach the catalyst active sites. Moreover, when the TEA concentration was reduced to 10 vol%, the H₂ production still remained at 2040.1 µmol h⁻¹ m⁻², which was very close to that of BBTNC‐MMM in a 20 vol% of TEA. This indicated that BBTNC was the main reason for the enhanced H₂ production of the membrane rather than TEA concentration. It is worth noting that BBTNC‐MMM showed the H₂ production of 882.6 µmol h⁻¹ m⁻² in pure water, while due to the presence of ions, the H₂ production slightly increased to 903.3 µmol h⁻¹ m⁻² in seawater. This discovery provided an important reference for the application of BBTNC‐MMM in actual water environments like seawater.

Finally, the piezoelectric‐photocatalytic cyclic stability of BBTNC‐MMM was evaluated. As shown in Figure 6f, BBTNC‐MMM maintained excellent stability throughout 18 hours of cycling, indicating strong adhesion between BBTNC and the PVDF‐TrFE substrate. The morphology of BBTNC‐MMM after the catalytic reaction was analyzed using SEM and TEM. As shown in Figure S14a and Figure 4g, the surface morphology of the membrane showed no significant difference from that before the reaction, indicating that the BBTNC inside BBTNC‐MMM had not aggregated, detached, or dissolved. Similar results could also be observed from the TEM and TEM‐EDS images of BBTNC‐MMM in Figure S14b and Figure 4j. Furthermore, nearly identical the XRD spectra of BBTNC‐MMM before and after reaction demonstrated that the crystal structure of the BBTNC‐MMM did not change after reaction (Figure S15). Compared with the FT‐IR spectra before the reaction, it showed that the spectrum of BBTNC‐MMM after the reaction displayed the characteristic peak of TEA at 1030 cm⁻¹, while other peaks remained unchanged, indicating that the composition of BBTNC‐MMM did not change after the reaction (Figure S16). The above results proved that the BBTNC‐MMM had excellent mechanical and chemical stability.

3.5. Piezoelectric‐Photoelectrochemical Characterizations of BBT and BBTNC

The charge carrier transport properties of BBTNC were investigated by a series of piezoelectric‐photoelectrochemical measurements. As presented in Figure 7a, the transient piezoelectric‐photocurrent response results showed that BBTNC existed the higher piezoelectric‐photocurrent density than BBT, which meant that the charges in the BBTNC were more easily separated and transferred. This could also be obtained from the photoluminescence (PL) spectra of the membranes (Figure S17). Compared with BBT‐MMM and COFs‐MMM, the PL intensity of BBTNC‐MMM was significantly quenched, indicating that the constructed heterojunction could effectively suppress the recombination of photogenerated carriers [28]. In addition, the piezoelectric‐photocurrent densities of BBTNC and BBT were both higher than their respective photocurrent densities and piezo‐current densities. This demonstrated the synergistic effect of ultrasound and light stimulation can enhance the catalytic efficiency [49].

FIGURE 7.

(a) The transient current curves of BBT and BBTNC, the Mott‐Schottky plots for (b) BBT and (c) COFs with 500 Hz, 1000 Hz, and 2000 Hz, (d) UV–vis of BBT, COFs, and BBTNC, and the band‐gap energies of (e) BBT and (f) COFs.

The charge transfer resistance performance of BBTNC was investigated by the electrochemical impedance spectroscopy (EIS) in Figure S18. The Nyquist plots demonstrated that the charge transfer resistance of BBTNC was lower than that of BBT, which proved that the heterojunction could facilitate the charge migration process in the catalytic process [28, 50]. Furthermore, in order to demonstrate the influence of the piezoelectric effect, the charge transfer resistance of BBTNC under ultrasound and without ultrasound was also investigated. The results showed that the piezoelectric effect could facilitate the effective separation of charges, which was beneficial for the subsequent catalytic reaction.

Mott‐Schottky (M‐S) plots for BBT and COFs were shown in Figure 7b,c. The conduction band (CB) of BBT and COFs were acknowledged via M‐S measurements under three different frequencies. It could be obtained that the flat band positions (E _fb) of BBT and COFs were ‐0.68 and ‐0.79 eV vs the saturated silver chloride electrode. Consequently, the CB of BBT and COFs were estimated to be ‐0.48 and ‐0.59 eV (calculation formula: E _{CB, NHE} = E _{fb, Ag/AgCl} + 0.197 V) [51].

In addition, the light absorption characteristics of BBT and BBTNC were measured using UV–vis. As shown in Figure 7d, the COFs and BBTNC exhibited relatively wide absorption bands with absorption edges of ≈550 nm, while the absorption edge of BBT was approximately 335 nm. This indicated that the heterostructure constructed by BBT and COFs could collect more visible light, thereby generating more photogenerated electron–hole pairs. This was also one of the reasons why the H₂ production of BBTNC‐MMM was higher than that of BBT‐MMM.

Moreover, the bandgap energy (E _g) was calculated via Tauc's equation (Equation 3) [52]:

$$\alpha h\upsilon =B{\left(h\upsilon -E_{\mathrm{g}}\right)}^n$$ (3)

where α denotes the absorption coefficient, h is the Planck constant, υ indicates the frequency, B is the constant, and n represents a constant with values of 2 and 1/2. Thus, E _g of the materials could be determined from the tangent intercepts of (αhv)^1/2 and (αhv)² vs photon energy (hv) curve [52]. Since BBT and COFs are usually indirect bandgap semiconductors and direct bandgap semiconductors, the values of n are 2 and 1/2, respectively [25, 53]. As illustrated in Figure 7e,f, the bandgap energies (E _g) of BBT and COFs were respectively 2.77 and 2.08 eV. According to the results, the valence band (VB) of BBT and COFs were 2.29 and 1.49 eV vs NHE, as shown in Figure S19.

Although BBT had piezoelectric properties, its wide bandgap (2.7 eV) severely limited its light absorption range, and the rapid charge recombination further reduced its photocatalytic efficiency [35]. In addition, although COFs had the relatively good light capture ability, their charge separation efficiency was insufficient, resulting in the relatively low catalytic performance [25, 54]. In contrast, the BBTNC heterostructure effectively overcame these limitations by synergistically integrating the characteristics of the two components, thereby achieving high catalytic efficiency through the coupling of piezoelectric and photocatalytic mechanism [28].

The charge transfer process of BBTNC was further demonstrated by electron paramagnetic resonance (EPR) using DMPO as the probe. As displayed in Figure S20a,b, under lighting conditions, only the DMPO‐*OH signal of BBT was detected, while only the DMPO‐*O₂ ⁻ signal of COFs was detected. Meanwhile, only the DMPO‐* O₂ ⁻ signal was detected in the BBTNC, but no DMPO‐*OH signal was detected. This might be due to the fact that the COFs shell covered the BBT, making it difficult to capture the signal. Together with PL and XPS data, these results strongly supported the formation of step‐scheme (S‐scheme) heterojunctions in BBTNC [55, 56].

3.6. Piezoelectric‐Photocatalytic Mechanism via the S‐Scheme Heterojunction

The piezoelectric potentials generated by BBT and BBTNC under the same force were analyzed by COMSOL Multiphysics simulation. As shown in Figure 8a,b, under the action of force, the potential difference generated by BBTNC was approximately 18.9 µV, which was 63 times higher than that of BBT (0.3 µV). It demonstrated that the construction of heterojunction effectively enhanced the built‐in electric field of the material [33], and this result was also consistent with the test results of XPS and KPFM. In addition, BBT had a uniform potential distribution (Figure S21), while BBTNC generated a nanoscale strong potential gradient at the interface. This not only promoted the spatial separation and directional migration of photogenerated carriers, thereby significantly inhibiting their recombination, but also generated local active sites, ultimately achieving a significant improvement in catalytic performance.

FIGURE 8.

The piezoelectric potential distribution of the simulated (a) BBT and (b) BBTNC, and the formation mechanism of the BBTNC heterostructure with under (c) the stimulation of light and (d) the combined stimulation of light and ultrasound.

Under light conditions, BBT and COFs respectively absorbed light energy to generate electron–hole pairs. Due to the formation of S‐scheme heterojunction, the electrons on BBT CB flowed towards COFs VB and combined with the holes generated on COFs VB in Figure 8c [57]. This S‐scheme heterojunction promoted the separation of photogenerated carriers while retaining strong reducing (electrons of COFs CB) and oxidizing (holes of BBT VB) capabilities [22, 58]. Subsequently, electrons on the COFs CB promoted the occurrence of the reduction reaction and generated H₂, while the holes in BBT VB oxidized triethanolamine (TEA), thereby enhancing the photocatalytic hydrogen production efficiency [56, 57]. Meanwhile, TEA could promote the generation of H⁺ in the solution, further enhancing the catalytic efficiency [59].

As shown in Figure 8d, the BBT in the heterojunction established the piezoelectric polarized built‐in electric field and induced band bending under simultaneous light and ultrasonic vibration, which worked in synergy with the S‐scheme heterojunction to enhance the transfer of photogenerated electrons from BBT to COFs, further achieving separation of photogenerated carriers [9, 60]. On the other hand, ultrasound increased transport of mass in porous membranes, enhancing the catalytic efficiency [9]. This research provided an important theoretical foundations and practical guidance for developing novel and highly efficient piezoelectric‐photocatalytic synergistic systems.

4. Conclusion

In conclusion, this work successfully synthesized the sheet‐like BaBi₄Ti₄O₁₅@covalent organic frameworks (BBTNC) by solvothermal method, which made full use of inherent properties of COFs, including high specific surface area and suitable band structure. COMSOL simulation showed that the piezoelectric potential difference of BBTNC was 63 times that of BaBi₄Ti₄O₁₅ and BBTNC generated a nanoscale strong potential gradient at the interface. This not only greatly exposed the catalytic active sites, but also the constructed S‐scheme heterojunction had a higher built‐in electric field intensity, which could effectively promote the separation of photogenerated carriers. Subsequently, the mixed matrix membrane (BBTNC‐MMM) was prepared by incorporating BBTNC into PVDF‐TrFE through the phase inversion method. The BBTNC‐MMM exhibited the significantly enhanced piezoelectric response, generating an output voltage of 3.8 V under tapping mode, which was more than 3 times higher than that of the PT membrane (1.2 V), indicating its potential for piezoelectric catalytic applications. Meanwhile, BBTNC‐MMM had excellent piezoelectric photocatalytic activity, with the H₂ production of 2072.3 µmol h⁻¹ m⁻² and good cycling stability for 18 h. Overall, this study provided new possibilities for the application of piezoelectric‐photocatalytic membranes in water splitting H₂ production.

Author Contributions

Ran Lei:—methodology and writing – original draft. Yan Zhu: —formal analysis. Shengpeng Han: —software. Piqi Song: ‐ investigation. Guicheng Jiang: —funding acquisition, writing – review & editing. Danqing Liu: —funding acquisition, writing – review & editing. Bin Yang: —funding acquisition, writing – review & editing.

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Supporting File: smll73916‐sup‐0001‐SuppMat.docx.

Data Availability Statement

The data that support the findings of this study are available from the corresponding author upon reasonable request.