Antifouling BaTiO₃/PVDF piezoelectric membrane for ultrafiltration of oily bilge water

Ballast and bilge waters are the major wastewaters generated on the ships, which are hazardous for the marine environment. Bilge waters are particularly dangerous due to the carcinogenic petroleum components in them (Karakulski & Gryta 2017). With regards to this, the MARPOL 73/78 Convention has been made by the International Maritime Organization (IMO) to prevent the discharge of oily wastewaters directly into the sea (Ulucan & Kurt 2015). The Annex 1 of MARPOL 73/83 requires that the oil concentration of discharged wastewater cannot exceed 15 ppm (Iduk & Nitonye 2015).

Bilge water is usually a complex composition of wastewater, sea water, oil, surfactant, sludge, metals, and other chemicals (Bian et al. 2019). Especially due to the presence of surfactant, the major part of the oil in bilge water exists in form of emulsified oil, which can be dispersed well and kept stable for a long time. Conventional oily wastewater treatment methods include gravity (Varjani et al. 2020), centrifugation (Cambiella et al. 2016), dissolved air flotation (Bian et al. 2019), de-emulsification, coagulation and flocculation (Zhao et al. 2021), which have several disadvantages such as low efficiency, high operation costs, corrosion and recontamination problems. Ultrafiltration process for oil/water emulsion wastewater treatment has attributed to its advantage of effective oil rejection due to its suitable membrane pore sizes (generally within 2–50 nm range) (Huang et al. 2015; Luo et al. 2015), and mild operating conditions, i.e., low energy consumption (Padaki et al. 2015). Membrane fouling caused by oil molecules deposition on the membrane surface or changes in the membrane pore size resulted in a decrease in permeate flow. Fouling is a filtering process that is constantly deteriorating in terms of efficiency and expense. Fouling prevention would improve membrane filtering performance while also lowering operating costs and extending the membrane's lifespan. Membrane cleaning (Shi et al. 2014), membrane surface modification (Adib & Raisi 2020; Aktij et al. 2020; Wang et al. 2020), membrane vibration (Ullah et al. 2020), backflush, and ultrasound (Qasim et al. 2018; Aktij et al. 2020) are some of the fouling mitigation approaches used. Membrane cleaning is a common approach for preventing membrane fouling, and it involves both physical and chemical cleaning of the fouled membrane. Backwashing and other physical cleaning procedures are effective at limiting reversible fouling. However, to eliminate irreversible fouling, physical cleaning is usually performed in conjunction with chemical cleaning. Chemical cleaning is an effective way to recover the permeate flow of membrane. On the other hand, frequent chemical cleaning using cleaning agents such as bases, oxidants, acids, and surfactants, may cause damage to the membrane structure and integrity. It is necessary to replace the damaged membrane. In comparison to other fouling reduction strategies, surface modification is very feasible and effective. It was discovered that membranes with a more hydrophilic surface had better antifouling properties (Liu et al. 2015, 2022). One of the major modifying ways to decrease oil fouling is to increase the hydrophilicity of the membrane surface. Attaching functional hydrophilic groups to the surface of the membranes, covering the membrane with hydrophilic polymers or nanoparticles, and incorporating hydrophilic materials inside the membrane are all frequently used methods for improving the hydrophilicity of the membranes (Ang et al. 2020). Nevertheless, the modified hydrophilicity of the membrane surface is unstable over time, reducing the antifouling capabilities of the membranes. The nondurable hydrophilic membrane has hampered its widespread use in wastewater treatment (Tang et al. 2021). Vibration-assisted unsteady-shear enhanced process (VSEP) is an efficient way to reduce membrane fouling (Zouboulis et al. 2019). However, one of VSEP's drawbacks is its high energy consumption, which has been estimated to be as high as 500 W/m² (Zamani et al. 2015). Furthermore, excessive vibration not only raises energy costs, but it also has the potential to harm membranes. Ultrasound wave provides a cost-effective and efficient alternative to standard cleaning methods in fouling mitigation; nonetheless, it is limited to remove the external fouling at the membrane surface. As a result, the internal fouling of membrane needs to be removed using chemical cleaning. In comparison to bench-scale, scaling up the ultrasound wave technique in practical use is difficult. As the ultrasound waves pass through the membrane module, ultrasonic waves are less efficient in fouling mitigation for membrane filtration of oily wastewater. Ultrasound wave method is difficult to scale up in practical application compared to the bench-scale. Especially, ultrasound wave is less effective in fouling mitigation for membrane filtration of oily wastewater as the waves pass through the membrane module. Increases in ultrasonic strength that are too high may cause membrane damage (Andrés et al. 2020). In contrast to external mechanical vibration generated by vibratory shear-enhanced processes and ultrasound waves, the piezoelectric membrane can generate internal vibrational (Le et al. 2022), which can effectively mitigate membrane antifouling. Ultrafiltration membranes based on piezoelectric materials to generate in-situ molecular vibration show strong antifouling performance in an alternating electric field. Because of its superior chemical, biological, and mechanical resistance, polyvinylidene fluoride (PVDF) is an attractive polymer for membrane production. Moreover, PVDF is a semi-crystalline polymer that has piezoelectric characteristics. The all-trans (β) phase is the primary contribution to PVDF's piezoelectric capabilities among the four known crystalline structures (α, β, γ, δ) of PVDF polymer (Huang et al. 2021; Zhang et al. 2019). At the present time, piezoelectric membranes based on piezoelectric materials such as lead zirconate titanate and β-PVDF have been produced for membrane separation processes. The filtering performance of these piezoelectric membranes enhanced under the application of AC signal. Mao et al. (2018) prepared a porous lead zirconate titanate membrane by dry pressing and sintering at 950 °C to mitigate fouling during oil-in-water emulsion separation by the application of an alternating voltage. Cao et al. (2020) prepared β-PVDF flat sheet membranes to reduce membrane fouling in anaerobic membrane bioreactor by applying a given frequency and alternating voltage. Chen et al. (Chen & Pomalaza-Ráezb 2019) used a piezoelectric PVDF film to separate the kaolin suspension and slow the membrane fouling by changing the voltage and frequency. However, β-PVDF has low piezoelectric coefficient to limit the properties of antifouling. The hazardous nature of Pb-based piezoelectric materials like lead zirconate titanate raises severe concerns about human and environmental health (Habib et al. 2020). Furthermore, the preparation of porous piezoelectric ceramic membrane requires sintering at high temperature. Although a perovskite-type BaTiO₃ is an outstanding piezoelectric material owing to its low-cost, excellent mechanical properties and inherent piezoelectric attribute (Siddiqui et al. 2016; Chen et al. 2017; Yang et al. 2017), perovskite-type BaTiO₃ is difficult to fabricate porous PZT ceramic filtration membrane. Hence, in this work, a piezoelectric membrane consisting of BaTiO₃ nanoparticle and PVDF matrix was prepared to separate bilge water. The piezoelectric membrane was investigated for filtration function and antifouling performance.

Perovskite-type BaTiO₃ (<100 nm, metal basis <99.9%) and PVDF powders (Mw = 534,000) were supplied from Sigma-Aldrich (Shanghai, China). The other reagents, e.g. dimethylacetamide (DMAC, >99%, reagent), polyvinyl pyrrolidone and ethanol were purchased from Shanghai Macklin Biochemical Co., Ltd. All other chemicals and reagents were used without further purification.

BaTiO₃/PVDF membrane sheet was prepared by the phase inversion method. BaTiO₃ powders (2%, by weight of PVDF) were fully dispersed in DMAC by ultrasonic treatment for 1 h. Then, PVDF powders (20%, by weight of the solution) and polyvinyl pyrrolidone (3%, by weight of the solution) as pore-foaming agent were added in the stirring mixed solvent. Vigorously stirring was carried out for 24 h to form a homogeneous casting solution, and then left still at 25 °C for 12 h to remove air bubbles. An incipient membrane formed with the degassed solution being uniformly casted onto a glass substrate by a casting knife with gap thickness of 250 μm. Subsequently, the casting film with its glass substrate was immersed immediately into a coagulation bath of distilled water and ethanol (40% v/v) to remove the residue of diluent at 25 °C, complete the phase inversion process. Once peeled off from the glass plate, the solidified membrane was rinsed thoroughly in deionized water at room temperature for completely removing the residual solvent. Then, the BaTiO₃/PVDF film was electrically poled for piezoelectric activation by being sandwiched between two copper plates under an electric field of 50 MV·m⁻¹ provided by a high DC voltage supply in air at 90 °C for 4 h. The poled BaTiO₃/PVDF piezoelectric membrane sheet was used for ultrafiltration of oily bilge water. The average pore size and porosity were determined by mercury intrusion porosimetry and presented in Table 2.

The bilge water used in the study was collected from a shore settling tank for oily wastewater accumulated and treated by the sedimentation in Weizhou harbor (China). The more details of the oily bilge water have been listed in Table 1 as well.

The BaTiO₃/PVDF piezoelectric film was cut into 45-mm diameter to fit the membrane cell. To determine the filtration performance of the BaTiO₃/PVDF piezoelectric membranes, the membrane was mounted in a crossflow membrane module filtration system. As shown in Figure 1, the piezoelectric membrane was sandwiched between two stainless steel mesh electrodes, which were connected to the anode and cathode of the functional signal generator (DG1022, Rigol, China). An alternating current (AC) signal with frequency ranging from 1.0 Hz to 15.0 MHz and voltage up to 200 V was produced by signal generator. The vibration signals of the membrane were collected by a hydrophone (RHS-10, Maihuang, China), then were processed using a digital oscilloscope (DS1202Z-E, Rigol, China).

The morphologies of the BaTiO₃/PVDF composite film were observed with scanning electron microscopy (JSM-IT300, JEOL, Japan). The crystalline structure of the samples was determined via by D8 Advance diffractometer (Bruker AXS, Germany) using Cu K-α radiation (40 kV, 40 mA). The pore size distributions of the membrane were measured by the gas-liquid replacement method employing a membrane pore size distribution apparatus (3H-2000PB, Beishide Instrument Technology Co., Beijing, China). Contact angle was measured by a contact angle measuring instrument (XG-CMB3, Sunzern Instrument Co., Ltd, Shanghai, China).The particle size distribution of oil droplets in the bilge water was determined with a dynamic light scattering (DLS) (DynaPro NanoStar, Wyatt, USA). Oil content was measured in the samples using Lambda 750 S ultraviolet-visible spectrophotometer (PerkinElmer, USA). The pH value of bilge water was adjusted with HCl and NaOH solution, and then determined by PHSJ-5 pH meter (Shanghai Precision & Scientific Instrument CO. Ltd, China). Total organic carbon (TOC) was analyzed by Shimadzu TOC-VCPH analyzer (Shimadzu Corp., Japan). Zeta potential was measured by Nano Plus Zetasizer (Micromeritics Instrument (Shanghai) Ltd, China).

Figure 2 shows the X-ray diffraction patterns for PVDF, BaTiO₃ and PVDF/BaTiO₃ piezoelectric composite film. The peaks were observed at 22.53°, 31.86°, 39.24°, 45.39°, 51.19° and 56.29°, 65.96° corresponding to (100), (110), (111), (200), (210) and (211) planes of the barium titanate, confirmed the cubic perovskite crystal form of BaTO₃ (Martins et al. 2014; Karthik et al. 2019; Yang et al. 2020), the peak at 20.28° corresponding to (110) plane proved the existence of β-crystal phase of PVDF (Mallick et al. 2020). The results confirmed the existence of piezoelectric β-phase PVDF and BaTiO₃ in composite films.

SEM images of the top surface, bottom surface and cross-section morphologies of fabricated BaTiO₃/PVDF membrane are illustrated in Figure 3. Figure 3 exhibits a typical morphology character of ultrafiltration membrane. As shown in Figure 3(a), the morphology of the top surface displayed a relatively dense skin with sparsely distributed nanopores. The nanopores on the surface would alleviate oil droplets trapping by the surface structures. Furthermore, the pore diameters of the bottom surfaces for the membrane were obviously larger than those of the corresponding top surfaces, as shown in Figure 3(b). The top surface was relatively smooth and no micrometer size particles were observed. This indicated that no large BaTiO₃ particles clustered on the top surface. The cross-section SEM image showed that a long finger-like macrovoids structure with sponge-like sublayer extended toward the bottom surface (Figure 3(c)), which is beneficial for water molecules to quickly permeate through the membrane (Chen et al. 2020).

The pore size and contact angle significantly affect the membrane separation process. As shown in Table 2, the average pore size of unpoled PVDF membrane was about 296 nm and the contact angle was about 136.1°. After being poled, the average pore size of the membrane was 302 nm and the contact angle nearly maintained the same size. There is no obvious change in the average pore size and contact angle of the membrane after poling.

Vibration performance of the BaTiO₃/PVDF piezoelectric membrane excited by AC signal was investigated. Vibration diagrams of the excitation piezoelectric membrane with different voltages at frequency of 220 kHz and the effects of the applied voltages on the vibration amplitude of the piezoelectric membrane are displayed in Figure 4. It is obviously observed that when the voltage of the applied AC signal was raised, the amplitude of vibration for the piezoelectric membrane grew significantly. The vibration did not increase linearly with the voltage of the applying AC signal (Zhang et al. 2019). The vibration amplitude of the membrane reached 12.5 mV when it was activated with a 20 V AC stimulus.

The effects of the applied frequency on the vibration amplitude of the BaTiO₃/PVDF piezoelectric membrane excited by AC signal at a constant voltage of 10 V are shown in Figure 5. The amplitude of vibration did not depend substantially on the frequency of the applied AC signal, which is similar to what has been seen in other piezoelectric materials (Darestani et al. 2013; Shivashankar & Gopalakrishnan 2020). The membrane's highest vibration response, with vibration amplitude of 10.6 mV, was seen at a frequency of 220 kHz, which was the resonance frequency of the BaTiO₃/PVDF piezoelectric membrane. Different piezoelectric material has different resonance frequency. A piezo material's resonance frequency is determined by its material composition, shape, size, and volume (Gao et al. 2017). The resonance frequency of a thicker piezo sheet will be lower than that of a thinner sheet of the same shape and material composition.

In order to evaluate the effect of the applied AC voltage on the ultrafiltration performance of the BaTiO₃/PVDF piezoelectric membrane, the permeate flux as a function of time was investigated using the membrane excited at the resonant frequency of 220 kHz with AC signals for various AC voltages, during 180 min ultrafiltration of the bilge water of 397.3 ppm at pH of 7.1 and a crossflow of 500 mL·min⁻¹, as shown in Figure 6.

During ultrafiltration operation for 21 min, the permeate flux for the piezoelectrical membrane activated by AC signal voltage ranging from 5 to 25 V suffered a fall in varied degrees, then the permeate fluxes progressively stabilized. For contrast, following a 21-min ultrafiltration operation, the permeate flux for the non-excited membrane dropped dramatically from 57.5 kg·m⁻²·h⁻¹ to 33.8 kg·m⁻²·h⁻¹, before eventually stabilizing at a much lower level. The results indicate that the non-excited membrane experienced a much more serious fouling than that of the piezoelectrically excited membrane, the excitation of the piezoelectric membrane dramatically reduced membrane fouling. The insert of Figure 6 shows the final stationary flux versus the AC signal voltages. The strong increase of the final stationary flux with amplitude of the applied AC voltages is obviously observed from the insert of Figure 6. When compared to the non-excited membrane, the membrane stimulated by AC signal with a voltage of 20 V produced the largest increase of 63.4 % in the final stationary flux during ultrafiltration of the bilge water. The maximum final stationary flux was caused by the excitation voltage of 20 V, which was close to 85.9% of its initial value in this study. It is self-evident that the vibration of the membrane is enhanced with the increase in the amplitude of the excitation voltage. However, even when the excitation voltage was raised to 25 V, the ultimate stationary flux remained unchanged when compared to the excitation voltage of 20 V. Based on these findings, the optimal excitation voltage for operating the piezoelectric membrane was determined to be 20 V. It is still unknown how piezoelectric vibration mitigates membrane fouling. One probable explanation is that piezoelectric vibration disrupts the initial concentration polarization layer, causing the creation of the filter cake layer on the feed side to be more effectively alleviated.

Despite the fact that piezoelectric materials with a wide variety of uses have been developed, most of them are not suited for generating filtration membranes due to the filtration membrane's unique requirements and characteristics. The main piezoelectric materials to be employed for producing the antifouling ultrafiltration membrane are piezoelectric PVDF and lead zirconate titanate (PZT). Table 3 summarizes the antifouling performance of this piezoelectric membrane in comparison to other piezoelectric membranes. The antifouling properties of the piezoelectric membrane are clearly affected by the piezoelectric material, excitation AC signal, and foulants. Furthermore, the filtration membranes which are made up of the same piezoelectric material excited with different AC signals show different antifouling performances for different foulants. However, the increase in the permeance of the BaTiO₃/PVDF piezoelectric membrane was found to be higher than that of the other piezoelectric membrane in Table 3. The reason for this is that the perovskite-type BaTiO₃ is an outstanding and sensitive material with a high piezoelectric coefficient, and when the BaTiO₃/PVDF piezoelectric membrane is excited by an AC voltage, strong vibration induced by the piezoelectric material increases the final stationary permeate flux. BaTiO₃/PVDF piezoelectric membranes offer better piezoelectric properties than PVDF piezoelectric membranes and are easier to fabricate than inorganic piezoelectric membranes like lead zirconate titanate piezoelectric membranes. In comparison to previous piezoelectric membranes, the BaTiO₃/PVDF piezoelectric membrane shows a lot of potential for antifouling performance.

The effect of the excitation frequency on the permeate flux and ultrafiltration performance of the piezoelectric membrane for the ultrafiltration of bilge water was investigated. The results are shown in Figure 7. Overall, the piezoelectric membrane that was excited by an AC signal produced larger permeate fluxes than the membrane that was not excited by an AC signal. The piezoelectric membrane demonstrated the greatest permeate flux at an excited AC signal frequency of 220 kHz, followed by 160 kHz. In the insert of Figure 7, the ultimate stationary fluxes of the membrane vs excitation frequency are displayed. From 50 kHz to 220 kHz, the final stationary flux dropped with increasing frequency, peaked at 220 kHz, and thereafter fell with increasing frequency. The frequency of the AC signals had no significant effect on the final stationary flux. The results show that the membrane with the highest frequency does not have the best antifouling properties (Darestani et al. 2013). The reason for this is that the highest frequency cannot provide the greatest amplitude of vibration. On the basis of these results, it was determined that 220 Hz was the optimum frequency for operating piezoelectric membranes.

To investigate the effect of the flow rate on the ultrafiltration performance of the piezoelectric membrane, the filtration experiments for bilge water were carried out. Figure 8 shows the results of the experiment operated at the cross-flow rate of 300, 500 and 800 mL/min for the non-excitation situation and excitation situation, respectively. The results illustrated in Figure 8 show that at all crossflow rates, the permeate flux maintained by the membrane excited by 10 V AC signal was greater than those for the non-excition membrane. An average of more than 30% increase in the permeate flux was obtained for the excitation membrane over the non-excitation. Moreover, the increase in the final stationary flux of the excitation membrane is different from that of the non-excitation membrane by increasing the crossflow rate. The insert of Figure 8 displays the effect of piezoelectric vibration on the final stationary flux was dramatic; when the piezoelectric membrane was excited with AC signal voltage of 10 V, the highest final stationary flux of 50.6 kg·m⁻²·h⁻¹ at a cross flow rate of was obtained, which was about 1.6-fold higher than the final stationary flux of 32.1 kg·m⁻²·h⁻¹ at the flow rate of 300 mL·min⁻¹. By increasing the crossflow from 300 to 800 mL·min⁻¹, a 59.3% increase in the final stationary flux was observed. For comparison, 30.3% improvement in the final stationary flux for the non-excitation membrane was achieved. The reason can be ascribed to that increasing the flow rate yields a thinner laminar boundary layer with a higher membrane-surface shear stress, which promotes fouling removal and diminishes fouling deposition on the surface of the membrane (Elcik et al. 2016). Higher shear stresses are also generated by the membrane mechanically vibrating. In fact, the in-situ vibration of the membrane being excited by AC signal is taken as a special case of generating shear stress. Consequently, turbulence generated by the piezoelectric membrane surface vibration worked synergistically with the shear forces created by the crossflow to mitigate membrane fouling.

The effect of pH value on the permeate flux and antifouling performance of the BaTiO₃/PVDF piezoelectric membrane for filtration of bilge water is shown in Figure 9(a). It was found that the final stationary fluxes increased with pH and AC voltage application. When the system was operated for filtration of bilge water at AC signal voltage of 10 V with a pH of 10.2, 7.1 and 3.4, the corresponding final stationery flux was 50.6, 39.6, 13.4 kg·m⁻²·h⁻¹, respectively. In case of the system was operated for filtration of bilge water without applying AC signal at pH of 10.2, 7.1 and 3.4, the corresponding final stationery flux was 25.5, 22.4, 10.6 kg·m⁻²·h⁻¹, respectively. In other words, whether the ultrafiltration system was excited by AC signal or not, the lowest final stationery flux of the BaTiO₃/PVDF piezoelectric membrane at a pH of 3.4 was observed compared with that of at pH values of 10.2 and 7.1. The phenomenon can be attributed to the electrostatic effect caused by the double layer (EDL) near the membrane and oil droplets surfaces. The double layer (EDL) near the membrane and oil droplets surfaces is consist of charged ions or molecules. The ζ potential is primarily determined by the amount and species of charged ions or molecules as well as the detailed structure of electrostatic double layer. pH value is the most important factor that remarkably affects ζ-potential. As shown in Figure 9(b), the isoelectric point of the BaTiO₃/PVDF piezoelectric membrane was approximately 6.7, which meant the membrane had negative surface charge density when the pH value of bilge water was over 6.7. The zeta potential of oil droplets in bilge water at pH value of 6.7 was determined as about −67.4 mV. Therefore, at pH values over 6.7, the membrane and oil droplets of bilge water both negatively charge and repel each other when their surfaces are within 100 nm. The repulsion reduces the blocking of the membrane pores by oil droplets, which results in the oil droplets adhering on the surface of the membrane are easily removed by vibration and increases in the final stationary flux, as shown in Figure 9(a). On the other hand, at pH value below 6.7, the membrane surface possessed positive charge density and EDL exhibited an attraction for oil droplets. The oil droplets in bilge water gathered on the surface of the membrane and blocked the membrane pore, which resulted in the final stationary fluxes decreased, as shown in Figure 9(a). Therefore, pH value above isoelectric point is important to avoid the fouling occurred on the surface of the membrane during the ultrafiltration of bilge water.

The permeate flow stability of the BaTiO₃/PVDF piezoelectric membrane for the ultrafiltration separation of bilge water excited by an AC signal was investigated through long-term experiments. The bilge water was continually fed through the piezoelectric membrane during the 50-hour separation. It became clear water, implying that the membrane was highly and consistently available in bilge water. Figure 10 summarizes the results about the 50-h ultrafiltration of bilge water with the BaTiO₃/PVDF piezoelectric membrane excited by AC signal at 220 kHz and 15 V, monitoring the flux stability, oil content, TOC content and TOC removal rate in the outflow. As can be seen from Figure 10(a), the permeate flux of the membrane had no obvious change for 50-h ultrafiltration of bilge water, suggesting that the piezoelectric membrane has outstanding antifouling performance.

Figure 10(a) and 10(b) show the oil removal of the piezoelectric membrane as well as the TOCs in the collected filtrates, respectively. The greatest oil content in the permeates from the piezoelectric excitation membrane for 50-h ultrafiltration of bilge water was 13.5 ppm, which was less than 15 ppm and so met the MARPOL 73/78 regulations for marine discharge (oil concentration less than 15 ppm). As shown in Figure 10(b), the TOC concentrations in the outflow from the piezoelectric excitation membrane during 50-h ultrafiltration of bilge water were monitored to be in the range of 42.8 to 72.4 mg·L⁻¹, while the initial TOC concentration of bilge water was 1,354.4 mg·L⁻¹, resulting in a rejection coefficient for TOC of the piezoelectric membrane of 94.7–96.8%. The experimental results show that the membrane is readily available for treating oily wastewater.

An antifouling BaTiO₃/PVDF piezoelectric membrane was fabricated and applied in bilge water separation successfully. The effects of the in-situ vibration generated by the piezoelectric membrane on the ultrafiltration performance of bilge water were investigated through the amplitude of the excitation AV and its frequency, bilge water pH, flow rate.

The vibration amplitude tests showed that the vibration amplitude of the membrane was enhanced with the increase in the AC signal voltage. The final stationary flux increased with increasing in the applied AC signal voltage. It was found that the final stationary flux of the membrane excited by 20 V of AC signal showed the highest membrane flux, which was close to 85.9% of its initial value in this study. In-situ vibration of the piezoelectric membrane excited by AC signal dramatically mitigated membrane fouling for the ultrafiltration of bilge water. The extent of fouling decreased with increasing the applied AC signal voltage. The frequency of the applied AC signal also affected the filtration performance of the excitation membrane for the ultrafiltration of bilge water. But, high frequency does not always bring about the high final stationary flux. The piezoelectric membrane excited by AC signal at 220 kHz of frequency produced the highest final stationary flux of 50.6 kg m⁻² h⁻¹. High flow rate always resulted in high permeate flux of the excitation membrane.

The stable permeate flux of the piezoelectric membrane excited by 15 V of AC signal voltage at 220 kHz for the 50-h filtration of bilge water was observed without remarkable decline. The application of the BaTiO₃/PVDF piezoelectric excitation membrane for the ultrafiltration treatment of bilge water allows to substantially reducing the content of oil to less than 14 ppm, which met the marine discharged level set by MARPOL convention. The membrane for the ultrafiltration of bilge water exhibits high rejection coefficient for TOC greater than 94.7%.

This work was supported by National Key Research and Development Plan of China (Grant No. 2018YFC1407404).

All relevant data are included in the paper.