A charged nanocomposite membrane via co-deposition of gallic acid and polyethyleneimine-silver for improving separation and antibacterial properties

The continued presence of pharmaceuticals and personal care products (PPCPs) in water environments, as well as their potential negative effects on ecology and human health, has recently become one of the environment's research hotspots (Fatta-Kassinos et al. 2011; Fabbri et al. 2017; Mille et al. 2020). With the increase in urbanization, the content of PPCPs in the environment is getting higher and higher, such as in surface water, drinking water, and groundwater (Murthy & Gupta 1997). Researchers have found that the concentration of PPCPs has reached the level of ng/L in the water environment (McClellan & Halden 2010; Jelic et al. 2011). One way in which the aquatic environment is contaminated by PPCPs is through the direct discharge of drug-containing wastewater, and another source is excretion by patients treated with drugs (Bowen & Mohammad 1998a; Afonso et al. 2001). PPCPs in trace quantities have been proven to have possible harmful effects on human and animal lives, including disruption of several biological processes as well as development and reproduction inhibition (Ma et al. 2018; Ren et al. 2021). Herein, it is imperative to seek a highly efficient way to remove PPCPs.

Membrane separation technology plays an important role in solving the problems of water scarcity and pollution (Wang et al. 2018; Shokrollahi et al. 2020). Currently, membrane filtration has been used for the effective removal of individual PPCPs (Yang et al. 2017). Yoon et al. (2007) found that the nanofiltration (NF) membrane with 43–90% rejection for PPCPs was higher than the ultrafiltration membrane whose retention was about 40%. NF membrane technology has been widely used in water softening, brackish water desalination, wastewater treatment, dye purification, food processing, and organics removal (Lau et al. 2015; Mohammad et al. 2015). Also, surface coating is one of the methods to prepare the NF membrane. Han et al. (2014) utilized 2-hydroxypropyltrimethyl ammonium chloride chitosan with a positively charged character and good membrane-forming ability to fabricate the functional layer of the composite NF membrane. The composite membrane showed a classical positively charged membrane character which had higher rejection to multivalent cations.

Although membrane treatment technology has considerable advantages, there are many problems that need to be solved in practical applications, such as membrane biofouling (Guo et al. 2012). The key to preventing membrane biofouling is to prevent initial cell adhesion and its growth on the membrane surface (Rahaman et al. 2014). Therefore, the design of functional membranes with antibacterial properties is the most durable and effective method to prevent membrane biofouling (Jiang et al. 2013). Silver nanoparticles (AgNPs) have high thermal stability, chemical stability, and catalytic activity. It was often used in antibacterial preparation due to its ability to inhibit the growth of microorganisms (Basri et al. 2011). Researchers have successfully prepared various membranes containing AgNPs and proved their antibacterial properties (Yu 2003; Chou et al. 2005; Slistan-Grijalva et al. 2008; Taurozzi et al. 2008; Andrade et al. 2015). Also, research shows AgNPs can effectively inhibit the growth of bacteria because it can interfere with the synthesis of ribonucleic acid (Zhang et al. 2000).

Nevertheless, membranes are still facing great challenges in achieving the highest filtration efficiency due to the trade-off effect between permeability and rejection, as well as the need to introduce antibacterial properties in the membrane (Ayaz et al. 2019). Gallic acid (GA) is a natural polyphenolic compound containing carboxyl that is widely found in grapes, tea, and other plants in nature (Sileika et al. 2013). It has been widely used in food, biology, medical chemistry, and other fields due to its antioxidant (Rajalakshmi et al. 2001), water-soluble, antibacterial, and other characteristics. Besides, its carboxyl and hydroxyl groups provide good hydrophilicity, which could improve the permeation flux of the membrane. Also, in order to improve the separation performance, it is an effective method to improve the charge. Polyethyleneimine (PEI) is a water-soluble polymer with high cation density and strong adsorbability. Active primary and secondary ammonia on the PEI molecular chain can react with the epoxy group, carboxylic acid, and isocyanate (Dick & Ham 1970). Thus, it can be used to prepare a charged NF membrane (Chen et al. 2020).

Herein, this study aims to construct a positively charged composite NF membrane with natural GA and charged PEI via surface coating. This novel membrane contained a loose pore structure and excellent hydrophilicity, which could improve the permeate flux. The positively charged membrane had a stronger electrostatic repellent to both divalent cation and positively charged pharmaceuticals, which was mainly caused by the amine groups of PEI in the surface active layer. The characteristics of the membrane were analyzed by scanning electron microscope (SEM), atomic force microscope (AFM), water contact angle, and zeta potential, respectively. Meanwhile, the optimum conditions for membrane preparation were investigated. In addition, a GA/PEI-Ag antibacterial membrane was prepared by in situ reduction of AgNO₃ on the membrane surface to improve the antibacterial property. The stability and antibacterial properties of the antibacterial membrane were studied.

The polyacrylonitrile ultrafiltration membrane (PAN, MWCO = 50 kDa) was purchased from Ande Membrane Separation Technology Engineering Co., Ltd (China). PEI was provided by Nanjing Lattice Chemical Technology Co., Ltd (China). GA was supplied by Shanghai Balinwei Chemical Technology Co., Ltd (China). Silver nitrate (AgNO₃) was obtained from Nanjing Chemical Reagent Co., Ltd (China). Atenolol (ATE), Carbamazepine (CBZ), Amlodipine (AML) and Ibuprofen (IBU) provided by Shanghai Yuanye Biotechnology Co., Ltd (China) were used in this study.

The GA/PEI-Ag composite antibacterial membrane with a high antibacterial rate was prepared by using AgNO₃ solution as the Ag⁺ source, using the binding force of phenolic hydroxyl group in GA, the in situ reduction of Ag⁺, the complexation of silver with –NH₂ and –NH on PEI molecular branch chain, and uniformly loading silver nanoparticles on the membrane surface with the assistance of ultraviolet reduction. Firstly, 8 mM of AgNO₃ was added to 0.3% PEI and stirred sufficiently at room temperature to form a composite solution of PEI-Ag ⁺. Then, 0.2% GA was added and stirred sufficiently to form a uniform composite solution of GA–PEI-Ag ⁺. Second, the PAN-H membrane was immersed in GA–PEI-Ag⁺ complex solution for 6 h with agitation to produce a GA–PEI-Ag antibacterial membrane. The surface of the PAN-H membrane was fully dried under the protection of water bath temperature and nitrogen blowing at 40 °C. Finally, the dried GA/PEI-Ag antibacterial membrane was irradiated with 200 W ultraviolet lamp for a period of time to assist in the reduction of silver ions to an elemental substance.

The chemical structure of the membrane surface was characterized by Fourier transform infrared spectroscopy (Thermo Fisher). The surface and section morphology were observed by SEM (NEC Electronics Corporation, Japan), and the roughness was studied by AFM (Bruker, Germany). At the same time, the contact angle was measured by the contact angle analyzer (Bruker, Germany) to reflect the hydrophilicity of the membrane. Then, the zeta potential of membranes was characterized by an electrokinetic analyzer (Anton Paar Gmbh, Austria).

The separation experiment of fabricated membranes was tested in the cross-flow filtration membrane module. Membranes were pre-operated using pure water at 7 bar for 30 min to obtain a stable flux prior to the test. In the separation experiment, salts (MgCl₂, MgSO₄, Na₂SO₄, NaCl, Na₂CO₃, and CaCl₂) with different charge properties and charge numbers of 0.5–10 g/L were selected. The rejection of membranes was determined under a pressure of 5 bar at 25 °C.

The GA/PEI composite membrane was used to intercept 50 μg/L of ATE, 500 μg/L of IBU, 5 μg/L of AML, and 5 μg/L of CBZ raw material solution, respectively. 100 mL of leachate was collected and the concentration of PPCPs in leachate was measured by the solid-phase extraction high-performance liquid chromatography (SPE-HPLC).

The PAN-H membrane was cut into three identical small diaphragms with a diameter of 20 mm. Three AgNO₃ solutions with different concentrations of 6, 8, and 10 mM were prepared, and three antibacterial membranes with different concentrations of silver were prepared by the in situ reduction method, respectively. Thereafter, Ag⁺ loading, Ag⁺ release rate, flux, and Ag⁺ release were tested to evaluate the stability of the antibacterial membrane.

As illustrated in Figure 2(b), with the increase of PEI concentration from 0.05 to 0.3%, the water flux decreased from 31.05 to 14.33 L·m⁻²·h⁻¹, while the MgCl₂ retention rate increased from 74.40 to 83.93% at the beginning and then decreased to 78.84%. This was due to the moderate cross-linking of GA and PEI with a moderate concentration, forming a moderate pore size with high selectivity. However, when PEI was excessive, some positions may not form cross-linked structures, thus forming larger pores (Zhang et al. 2014). For this reason, 0.3% PEI concentration was selected as the best concentration.

As exhibited in Figure 2(c), when the reaction time was 3 h, the pure water flux was relatively high, reaching 51.91 L·m⁻²·h⁻¹ at 0.5 MPa, but the rejection rate of salt was lower than 40%. With the increase of reaction time, the flux began to decrease and the salt rejection rate rose rapidly. That was because the membrane's mean effective pore size decreased and its membrane thickness grew as the reaction time increased (Zhang et al. 2016). Because the reaction between PEI and GA had likely reached a state of dynamic equilibrium by 9 h, the pure water flux was still maintained at 36.94 L·m⁻²·h⁻¹, and the rejection rate had also improved to some extent. As a result, when the reaction time was greater than 9 h, the flux and the rejection rate almost did not change. Therefore, considering flux and rejection rate, the optimal reaction time was determined to be 9 h.

As shown in Figure 2(d), experiments demonstrated that using a PEI molecular weight of 600 Da of the NF membrane, the pure water flux was very high, but the interception of Na₂SO₄ and MgCl₂ solution rate was relatively low. It was because the amount of amino groups with high relative molecular weight of PEI was relatively scarce, leading to higher reactivity, which increased the hydrophilicity of the basement membrane by forming a layer of smooth surface reaction layer (Weng et al. 2016). When the molecular weight of PEI was 1,800 Da, the reaction activity increased, the thickness of the reaction layer increased, and the pure water flux decreased slightly. However, the rejection rate of the composite membrane for Na₂SO₄ and MgCl₂ solutions increased significantly because of the increase in the number of amino groups leading to the increase of effective charge. When the molecular weight of PEI was 10,000 Da, the number of reactive amino groups decreased due to the steric effect, leading to reduced reaction activity. Moreover, the macromolecular reticular structure was more prone to entanglement, which may lead to uneven local coating, resulting in greater roughness and reduced hydrophilicity. Thus, the flux of pure water decreased obviously, but the increasing trend of the rejection rate of salt solution slowed down (Ariza et al. 2002). Therefore, after comprehensive consideration, PEI with a molecular weight of 1,800 Da was adopted.

Figure 3(b) indicates the variation trend of the static contact angle of each composite membrane under different reaction times and different molecular weights of PEI. It can be seen from the figure that the contact angle of all composite membranes was less than 50°, indicating that both PAN-H membranes and composite membranes at different times and different molecular weights of PEI were hydrophilic. The contact angle of the PAN-H membrane as the base membrane was 48.31°. After coating on the surface of GA/PEI, the contact angle decreased to some extent. This was as a result of the membrane's porosity shape and the hydrophilic groups, including carboxyl, hydroxyl, amino, and methylamino, in the GA/PEI selection layer (Dye & Izake 2016). In addition, it was obvious that the contact angle of the composite membrane M3 was 33.19°, which greatly improved the hydrophilicity compared with the PAN-H membrane.

From Figure 3(c) it can be inferred that the variation of zeta potential on the surface of the PAN-H membrane and the GA/PEI composite membrane with pH 3–10. The zeta potential of the PAN-H membrane was about −49.67 mV under neutral conditions, and the surface of the PAN-H membrane was negatively charged in the pH range of 3–10. After GA and PEI surface coating, composite membrane zeta potential greatly improved because the positive charge carried by the amino group in PEI made the membrane surface positive (Wu et al. 2014). A more positive charge on the surface made the GA/PEI composite membrane improve the removal rate of divalent cation and positively charged PPCPs. However, the negative charge on the membrane surface was usually caused by protonated sulfonic acid or carboxylic acid groups at neutral pH. Due to the dissociation of functional groups, pH had an effect on the charge of the membrane, and the zeta potential of the membrane would also become more and more negative with the increase of pH and the deprotonation of functional groups (Tanninen & Nystrom 2002).

Figure 4(b) reveals that the MWCO of the composite membrane was 958 Da, which belonged to the category of the NF membrane (interception molecular weight is 200–1,000 Da). According to Equation (3), the effective pore size of the composite NF membrane can be calculated as 0.69 nm.

Figure 7(b) reflects the change in the rejection rate of the GA/PEI composite membrane for MgCl₂ and NaCl at different concentrations. As can be seen from the figure, with the increase in salt solution concentration, the rejection rate of the GA/PEI composite membrane for salt solution showed a decreasing trend. However, it can still maintain a relatively high retention rate for high concentration salts. On the one hand, an increase in Cl⁻ concentration reduced electrostatic repulsion by balancing the positive charge on the membrane surface, thus reducing the rejection rate of cations (Song et al. 2017). On the other hand, the Debye length of salt will decrease with the increase in salt concentration. However, the electrostatic repulsion is obvious only at a distance shorter than the Debye length, which leads to a reduction in the rejection rate.

Table 1 summarizes the comparison of water permeability and separation performance to PPCPs by different NF membranes in the previous literature. Compared with the membranes previously reported, the GA/PEI composite membrane had good water flux, good removal effect on ATE and AML, which had potential application value.

Table 2 describes the fixed Ag⁺ density and adsorption rate of the antibacterial membrane after the optimal reaction time. The results proved that with the increase in AgNO₃ concentration, the Ag⁺ density attached to the surface of the antibacterial membrane gradually increased as well as the adsorption rate. When the concentration of AgNO₃ increased from 6 to 8 mM, the Ag⁺ loading on the membrane increased significantly, and the latter was almost twice as much as the former. When the concentration of AgNO₃ increased to 10 mM, the density of Ag⁺ fixed on the membrane only slightly increased compared with that of the GA/PEI-Ag (8 mM) antibacterial membrane. This phenomenon can be explained as the silver ions on the membrane surface tended to saturate when the concentration of AgNO₃ was 8 mM. So when the concentration increased again, the adsorption rate will not be greatly improved. Therefore, 8 mM was selected as the optimal concentration of antibacterial membranes.

The variation in the amount of Ag⁺ released from three antibacterial membranes with different concentrations is examined in Figure 9(b). The concentration of silver ion in water was almost unchanged after 24 h of deionized water immersion, and the order of silver ion release was as follows: GA/PEI-Ag (10 mM) > GA/PEI-Ag (8 mM) > PEI-Ag (6 mM). It can be seen that the amount of silver ion release increased with the increase in concentration. Table 2 showed the silver ion density and static release rate after the release of three different concentrations of antimicrobial membranes. According to the data in the table, the order of the static release rate was as follows: GA/PEI-Ag (10 mM) > GA/PEI-Ag (6 mM) > PEI-Ag (8 mM). This was because 8 mM AgNO₃ fixed more silver ions, and the release values of GA/PEI-Ag (6 mM) and GA/PEI-Ag (8 mM) silver ions were consistent, so GA/PEI-Ag (8 mM) had a lower release rate. Therefore, 8 mM was determined to be the optimal concentration of the antimicrobial membrane.

After the filtration of pure water, the pure water flux, AgNO₃ concentration, and dynamic release rate of the three antibacterial membranes are shown in Table 2. It can be seen from the table that in the antibacterial membrane, the flux increased as the concentration of AgNO₃ decreased, showing a slow downward trend. The pure water flux in GA/PEI-Ag (8 mM) antibacterial membrane was slightly lower than that in the GA/PEI-Ag (6 mM) antibacterial membrane. However, the pure water flux of the GA/PEI-Ag (10 mM) antibacterial membrane decreased significantly, indicating that the AgNPs loaded on the surface of the GA/PEI-Ag (8 mM) antibacterial membrane were the most stable among the three kinds of membrane. Both pure water flux and release rate verified the rationality of choosing 8 mM AgNO₃ concentration to prepare the antibacterial membrane from the side. While this study found that the silver ion concentration in the static release experiment was almost unchanged after 24 h of immersion in deionized water. However, the membrane in dynamic experiment did not undergo such a long immersion.

In summary, this study successfully prepared a positively charged composite membrane for removal of PPCPs through surface coating technique. The success of the modified membrane was confirmed using FTIR. Also, the optimal membrane was determined to be the one with a GA and PEI concentration of a 0.2 and 0.3%, respectively. The optimal reaction time was 9 h and the optimal PEI molecular weight was 1,800 Da. SEM and AFM were used to study the surface microstructure of the optimal membrane. The optimal membrane showed a high PWP coefficient of 74.21 L·m⁻²·h⁻¹·MPa⁻¹ and MWCO of 958 Da. The optimal membrane also evaluated good repellent performance to various inorganic salts and PPCPs. After AgNPs were loaded on the modified membrane, the obtained GA/PEI-Ag antibacterial membrane possessed a great antibacterial ability. At the same time, the bacteriostatic rate of the GA/PEI-Ag antibacterial membrane was much higher than that of the GA/PEI composite membrane. Therefore, this study provides a new strategy for the preparation of NF membranes with effective PPCPs removal and antibacterial effect.

All relevant data are included in the paper or its Supplementary Information.

The authors declare there is no conflict.