Antifouling polymeric nanocomposite membrane based on interfacial polymerization of polyamide enhanced with green TiO₂ nanoparticles for water desalination

Materials

Titanium tetra-isopropoxide (C₁₂H₂₈O₄Ti) was used as a starting material (Sigma Aldrich, Co. LLC, Germany). Aqueous plant extract of Thyme plant from Wadi Hassn, which grows abundantly in mountainous areas of Jordan was used as a capping and reducing agent (Ahmadi & Jafarizadeh-Malmiri, 2021) in TiO₂ NPs synthesis. According to a previous study in Jordan for the same location, the thyme was of Thymus bovei, with a preferable aromatic flavor and odor (Kherissat & Al-Esawi, 2019). Thyme and its constituents contain phenolic compounds such as thymol and carvacrol, which are naturally antimicrobial and antioxidant sources used in veterinary medicine, agriculture, food, and medicine. The polyethersulfone (PES) granule (Mw 58,000 g/mol, Goodfellow, England), polyvinylpyrrolidone (PVP) (Mw 58,000 g/mol, Acros Organics, Germany) as pore former and 1-methyl-2-pyrrolidinone (NMP) (Thermo, ACROS, USA) were used to fabricate the PES substrate. Methylene blue (MB; Sigma-Aldrich, Inc., St. Louis, MO, USA). Sodium Chloride (NaCl), Magnesium Chloride (MgCl₂), and Calcium Chloride (CaCl₂) were from (Sigma Aldrich, Darmstadt, Germany). M-phenylendiamin (MPD), Trimesoyl Chloride (TMC) from Sigma Aldrich (St. Louis, MO, USA). n-hexane 96% (Sigma Aldrich, Darmstadt, Germany) were used for PA thin film preparation. Bovine serum albumin (BSA) from Sigma Aldrich (St. Louis, MO, USA) was used as a foulant in the antifouling test.

Nutrient agar, meat extract, peptone, casein peptone, and soybean peptones were purchased from Liofilchem (Roseto, Italy); nutrient broth (Oxoid, Mumbai, India), Sodium chloride (extra pure, Lobachemie, Mumbai, India), D-glucose (JHD, Guangdong, China), lecithin, and di-kaliumhydrogenphosphate from Riedel de- HaenAG Seelze-Hannover, Germany; tryptone soya agar (casein soya bean digest medium) (Oxoid, Hampshire, UK); Tween 80 (Polysorbate-80; ICI, London, UK); yeast extract agar (Mast Group, Merseyside, UK); agar (Merck, Branchburg, NJ, USA); and tryptone powder (Bio Basic, Markham, ON, Canada). Furthermore, Gram-negative Escherichia coli (E. coli) and Gram-positive Staphylococcus aureus (S. aureus) bacteria ATCC no. 8739 and 25913, respectively, were used.

Preparation of thyme plant extract

The thyme plants were washed several times with deionized water before air dried in a clean environment for three days. The dried plant was ground into a fine powder using a blender. To obtain a small size free of impurities, a sieve with a mesh number of 20 µm was used. 3g of the powder was boiled in 100 mL of deionized water for 30 min using the method described by Arabi et al. (2020). The resulting mixture was then filtered with Whatman filter paper, and the extracted extracts were kept in the refrigerator at 4  °C to preserve, avoid putrefaction and maintain the active ingredient properties for future use.

Synthesis of TiO₂ NPs

The titanium tetra-isopropoxide (C₁₂H₂₈O₄Ti) was used as a precursor for titanium. A mixture of 77% (v/v) deionized water and 7.7% (v/v) of the precursor, and 15% (v/v) aqueous extract, was added dropwise to a total volume of 100 ml. The solution was stirred in reflux at 50 °C for 4 h. The produced solution was centrifuged at 5500 rpm for 20 min. The product was dried for one hour at 60 °C. The final product has been calcined for 3 h at 500 °C to obtain the anatase crystal type, which has the highest photocatalytic and antimicrobial activity among the crystal structures as mentioned by Ripolles-Avila et al. (2019).

Preparation of substrate and thin film nanocomposite

The casting solution contained 18% of PES, 1% of PVP and 81% of NMP as a solvent at 60 °C with stirring for 24 h and then gas bubbles were removed by desiccator. Two steps in a Spin-Coater were used to deposit the casting solution on the glass plate, with a spinning speed of 200 rpm for 10 s and then 1500 rpm for 20 s, followed by the phase inversion method of immersing the glass plate in DI water.

TFC and TFN membranes were achieved through interfacial polymerization (IP) on fabricated PES membranes as support. Interfacial polymerization is a kind of growth polymerization where such polymerization takes place at the interface of two immiscible phases (usually two liquids), leading to a polymer that is transformed on the interface. To begin the IP process, the PES membrane was placed in a frame and allowed to contact with an aqueous solution containing MPD (2 wt/v%). After 2 min of retention on the PES membrane surface, the aqueous solution was discarded. The same membrane surface was then in contact with a pouring of 0.2 wt/v% TMC in an n-hexane solution to react with the previous monomer MPD for 1 min, then the organic solution was removed. The composite material membrane has been cured at 60 °C for 5 min until stored in RO water for further use. The whole membrane refers to the thin film composite (TFC) membrane. The fabrication for thin-film nanocomposite (TFN) membranes was the same as for TFC membranes, except for introducing 5% TiO₂ nanoparticles into the aqueous solution using an ultrasonic bath.

Analysis of the synthesized TiO₂ nanoparticles and the fabricated membranes

Ultraviolet–visible spectroscopy (UV–VIS) (Double beam UV-2450; Shimadzu, Kyoto, Japan) measurements were used to analyze the sample optical characteristics for the TiO₂ NPs, Fourier-transform Infrared Spectroscopy (FTIR) (Bruker-Avance III, Billerica, USA) was used for functional group characterization.

For the crystalline phases, Crystallinity and average crystalline size, the synthesized TiO₂ were evaluated using X-ray Diffraction (XRD) (Rigaku Ultima IV, Rigaku company, Tokyo, Japan). In addition, Debye Scherer relationship was used to calculate the average crystallite diameters (D) as follows: (1)${\displaystyle D=\left(k.\lambda \right)/\left(\text{ß}.\cos \left(\Theta \right)\right)}$

where k is the 0.9 crystal shape factor, λ is the X-ray diffraction wavelength of 0.15406 nm depending on the X-ray source, ß is the peak width at half maximum height, and Θ is the Bragg diffraction angle.

Analyzing elemental composition was done using X-ray fluorescence (XRF) (+NEXQC; Rigaku, The Woodlands, TX, USA). The surface hydrophilicity of the blend membranes was evaluated by a drop shape analyzer using a contact angle goniometer (Attension Theta lit; Biolin Scientific, Stockholm, Sweden). A scanning electron microscope (SEM) was used to examine the surface morphology of the nanoparticles and the membranes (Quanta FEG 450, Netherland), in addition to atomic force microscopy (AFM) for topography of membrane and roughness. dynamic light scattering (DLS; Theta Lite, USA) was used for the size distribution and stability of the nanoparticles. Thermogravimetric analysis (TGA) (Netzsch TG 209 F1 Iris) was used to confirm the purity of TiO₂NPs prepared by green approach and to investigate the effects of membranes modification on their decomposition behavior, and the percentage of TiO₂NPs within TFN membranes. The analysis was done at a heating rate of 10 °C/min from room temperature to 800 °C.

The catalytic performance of TiO₂ nanoparticles was investigated using the Methylene blue (MB) dye as a model pollutant and was used for the photocatalytic treatment of polluted water and wastewater under the source of UV light source at 365 nm (UVP C-70G chromate-Vue Cabinet, Ultraviolet Sources: 254 nm UV of 1290 µW/cm2 and 365 nm UV of 2100 µW/cm2), and under sunlight simulator (Peccell’s sunlight simulator; PEC-L15, Peccell Technologies, Inc., Yokohama, Japan) with a sunlight irradiance of 100 mW/cm2. Antimicrobial efficiency was tested also against gram positive and gram-negative bacteria.

Membrane performance

Membrane filtration performance was tested in a dead-end filtration system (HP4750; Sterlitech, Auburn, WA, USA) with a 14.6 cm² effective area. At first, the membrane was cut and placed into the cell according to its size. The membrane was then compressed for 30 min at 2.5 bars with deionized water.

The permeation flow (J₀) is calculated at 2 bars using the equation: (2)${\displaystyle \text{Flux}\left(J_0\right)=W/\left(A\ast \Delta T\right)}$

where W is the weight of the permeated solution in gram, A is the membrane surface area (m²), and ΔT is the duration of the experiment (hour). All calculations were repeated three times, and the average results were reported.

The feed solutions utilized to evaluate the membrane’s performance were 1000 ppm NaCl, MgCl₂, and CaCl₂ aqueous solutions and 2000 ppm NaCl to compare between the TFC and TFN that simulate brackish water composition. The tests were carried out at a pressure of 2 bar and a temperature of 25 °C. The concentration of the feed solutions and the concentration of the permeate were used to determine the solute rejection. The following equation was used to determine the rejection: (3)${\displaystyle \text{Rejection}\left(\text{\%}\right)=\left(C_{\mathrm{F}}-C_{\mathrm{P}}\right)/C_{\mathrm{F}}\ast 100\text{\%}}$

where C_F and C_P are the feed solution and permeate concentrations, respectively.

Fouling resistance

The membrane’s antifouling properties were assessed utilizing conventional BSA as foulant. Model foulants for the TFC and TFN were 0.5g/L BSA. The permeated flux was measured before, and after adding foulant over time. The flux of solution with BSA foulant (J₀) (kg/m² h) was calculated at 2 bars and for 90 min. The membranes were cleaned with deionized water after filtering the BSA solution. After rinsing the membranes, they were submerged in deionized water for 20 min. After that, the water flux of cleaned membranes (J_w) was calculated. The following formula was used to determine the flux recovery (F_R) (Wang et al., 2019): (4)${\displaystyle F_{\mathrm{R}}\left(\text{\%}\right)=\left(J_w/J_0\right)\ast 100}$

where the J₀ is water flux before fouling, J_w is the water flux after fouling. The Eq. (5) was used to calculate the total fouling ratio (R_t): (5)${\displaystyle R_t=\left(1-Jp/J_0\right)\ast 100\text{\%}}$

where the J_p is the flux with fouling. The following Eqs. (6) and (7) were used to define and determine percent of reversible fouling (R_r) and irreversible fouling (R_ir). (6)${\displaystyle R_r=\left(\left(J_w-J_p\right)/J_0\right)\ast 100}$ (7)${\displaystyle R_{\mathrm{ir}}=\left(\left(J_0-J_w\right)/J_0\right)\ast 100.}$

Antimicrobial activity

The antimicrobial property of NPs was investigated using the liquid medium method against various pathogens bacteria including E. coli, and Staphylococcus aureus. (S. aureus) the pathogens cells tested were prepared from a (0.5) McFarland turbidity standard (5 × 10⁷ cell mL ¹). The testing was incubated at 37 °C for 24 h according to Khashan et al. (2021).

The bacteria’s inhabitation rate was investigated by inoculating the broth with 0.2 mL of bacterial strains after the addition of TiO₂NPs at various doses of 100, 500, and 1000 ppm. The optical density (O.D.) of the bacteria at a wavelength of 600 nm was used to estimate its growth, while the inhibition ratios were calculated using Eq. (8). (8)${\displaystyle \text{Inhibition of Efficiency}\left(\text{\%}\right)=\frac{\left(\text{Control OD}-\text{Tested OD}\right)}{\text{Control OD}}\ast 100\text{\%}}$

where the control O.D is a result from bacteria without NPs and tested O.D is bacteria with nanoparticles

The antimicrobial activity of the membranes was assessed using the standard “ISO 22196:2007 Plastics—Measurement of antibacterial activity on plastics surfaces” as described in our paper previously (Alnairat et al., 2021).

Briefly, 13.0 g of nutrient broth in 1.0 L of DW was used. 28.0 g of nutrient agar was mixed with 1.0 L of DW to make Plate Count agar. Lecithin and polyoxyethylene sorbitan monooleate was used to prepare Soybean Casein Digest Broth (SCDLP broth).

Phosphate-buffered physiological saline was prepared by dissolving 8.5 g of NaCl in 1.0 L of DW and diluting 800 times. Autoclaving was used to sterilize all the media. Three TFN samples were compared to sex specimens of TFC membranes with dimensions of (2.2 × 2.2) cm².

Half of the TFC was washed with 10.0 mL of SCDLP buffer. A dilution of 100 L was taken and cultured on plate agar. The remaining pieces were washed with SCDLP, diluted, cultured, and incubated in the same way.

The number of colonies from each dilution was counted after incubation. The following equation was used to calculate the number of viable bacteria found: (9)${\displaystyle N=\left(100\times C\times D\times V\right)/A}$

where N is the number of viable bacteria recovered per cm² of the tested membrane samples. C: plate count average for duplicate or triplicate plates D: the counted plate dilution factor V: the volume of SCDLP added to the samples (mL). When no colonies are recovered in any of the agar plates for a dilution series, the colonies are counted as “V” (where V is the amount of SCDLP added to the membrane in milliliters). The sterilization ratio was calculated using the following (Eq. (10)): (10)${\displaystyle R=\left(A-B\right)/A\times 100}$

where R denotes the sterilization ratio. A: the number of viable bacteria recovered per cm² of the TFC membrane. B: the number of viable bacteria recovered per cm² of the treated TFN membrane.

Titanium oxide nanoparticles

Ultraviolet–visible spectroscopy (UV–VIS)

The absorption spectra of TiO₂ NPs were analyzed to further investigate the optical properties and confirm the formation of TiO₂ NPs. Figure 1A shows the absorption spectra of TiO₂ NPs, which displayed strong and prominent absorption peak at 260 nm. This indicates that the TiO₂ nanoparticle suspension has a high absorption capacity for ultraviolet light as reported by Gouda & Aljaafari (2012). The band gap energy of a TiO₂ NPs describes the amount of energy required to excite an electron from the valence band to the conduction band. An accurate determination of the band gap energy is critical in predicting the optical properties. In 1966, Tauc proposed using optical absorption spectra to calculate the band gap energy. The Tauc technique assumes that the energy-dependent absorption coefficient can be represented by the following (Eq. (11)) (Makuła, Pacia & Macyk, 2018). (11)${\displaystyle {\left(\alpha h\nu \right)}^{1/\gamma }=B\left(h\nu -E_g\right)}$

where h denotes the Planck constant, v: the frequency of a photon, E_g: the band gap energy, α: absorption coefficient equals 2.302 * absorbance, and B: a constant equal to 1240. The γ factor varies according to the nature of the electron transition and is equal to 1/2 for indirect transition band gaps and the result is shown in Fig. 1B. The energy band gap equals 3.17 eV confirm TiO₂ NPs formation, the literature value for anatase is estimated to be between 3.10 and 3.20 eV (López & Gómez, 2012).

Fourier-transform infrared spectroscopy (FTIR)

Figure 2 illustrates the FTIR spectra of the synthesized TiO₂ NPs and the plant extract. Ti–O stretching and Ti–O–Ti bridging stretching modes correspond to an absorption peak between 500 and 700 cm⁻¹ as reported by Goutam et al. (2018). The Ti-O-Ti bond is responsible for the observed peaks at 665 cm⁻¹ and the peak appeared in the range of 1600–1700 cm⁻¹ are ascribed to the -C =O-Ti bond from the residual of plant, and the amount of anatase infiltrative cavities may boost the number of hydroxyl groups that convert to hydroxyl radicals, increasing the rate of attack on organic compounds. FTIR spectra of the plant extract at 1640 cm⁻¹ is related to the C = C which is related to the thymol and the stretching of H-bonded O-H groups in hydration water causes the absorption band at 3300 cm⁻¹ (Ahmadi & Jafarizadeh-Malmiri, 2021).

X-ray diffraction (XRD) and X-ray fluorescence (XRF)

Figure 3 represents the synthesized TiO₂ NPs XRD patterns, which show sharp peaks at 25.53°, 38.1°, 48.23°, 54.15°, 62.94°, 68.28°, and 75.44°, which correspond to the anatase phase’s (1 0 1), (0 0 4), (2 0 0), (1 0 5),(2 0 4), (2 2 0), and (2 1 5) planes, respectively (JCPDS No. 01-071-1166) (32). The average crystallite size was calculated using Scherrer equation (Kashale et al., 2017) to be around 8 nm, with sharp peaks indicating a strong crystallinity of 92.15%.

Figure 4 shows the energy dispersive X-ray fluorescence (EDXRF) spectra for the prepared TFC, TFN membranes and green TiO₂ NPs. The results of nanoparticles spectra indicated the presence of Ti elements with Ti-K α and Ti-K β signals (4.51 keV and 4.93 keV, respectively) and showed a high concentration of Ti. Our results agree well with the results published by Archana et al. (2010) and Murali et al. (2016).

Scanning electron microscope (SEM)

Figure 5 depicts SEM image for the prepared TiO₂ NPs. The result showed a spherical form of the nanoparticles with a size distribution ranging from 7 to 5 nm. Also, SEM images indicated the presence of irregular and aggregated particles in range of 100 nm.

Dynamic light scattering (DLS)

Surface charges of the TiO₂ NPs in DI water and size distribution are shown in Fig. 6. A high positive or negative zeta potential causes flocculation and agglomeration resistance. If the zeta potential values of the particles are low, there is no strength to keep them from colliding and flocculating (Alnairat et al., 2021). Zeta potential (ζ) values of around ±30 mV in TiO₂ NPs indicate that either negatively or positively charged colloidal particles have reasonable dispersion stability (Ceballos-Chuc et al., 2022). A negative zeta potential value of −33.1 mV was observed in Fig. 6A, which represents the strength of electrostatic repulsion in dispersion between charged particles and solvent. A high zeta potential represents good stability to molecules and particles of sufficient size that the solution or dispersion will resist aggregation and close to 35 mV finding (Ceballos-Chuc et al., 2022). Figure 6B shows peak at 18 nm, which is close to SEM image results and peak at 100 nm, also in consistence with the irregular aggregation shown in the SEM. It is well known that DLS measures the hydrodynamic radius and that a small number of larger particles (which could be aggregates or impurities) usually have a larger population of smaller particles. The polydispersity index (PDI) was 0.3 signifies a homogeneous population while implying a uniform sample in terms of particle size and suitable for the DLS test.

Thermogravimetric analysis (TGA)

TGA was used to confirm the purity of TiO₂ NPs prepared by green approach. The results showed a weight loss in the temperature below 150 °C related to moisture trapped in the of NPs sample (Kashale et al., 2017), as shown in Fig. 7. In addition, the reduction in weight at temperature in the range of 250 to 670 °C. represents the decomposition of organic compounds.

There was a small weight reduction, which is shown in the range of 500–700 °C, possibly due to the loss of residual carbon from the plant (Saranya et al., 2018). The residual was equal to almost 95% of the total weight which indicated highly pure TiO₂ NPs.

Photocatalysis activity of TiO₂ NPs

A total of 25 mg of the prepared TiO₂ NPs was mixed with 10 mg of methylene blue in 100 mL of distilled water under a UV and other samples under one sun simulator. In a dark room, the mixture solution (Fig. 8C) was constantly stirred. There was no change in color intensity after 30 min. However, as shown in Figs. 8A and 8B there was a slight decrease in peak at the UV–VIS test at 664 nm due to equilibrium sorption and desorption. The photodegradation result is very active under UV and the sun simulator due to MB degradation of and a decrease in MB concentration caused by the photoactivity of TiO₂ NPs and capacity the free radical’s capacity to oxidize organic MB. The rate of reaction was calculated using the corresponding kinetic equation, (Ln (C_t/C₀) = Kapp * t), where C_t denotes the concentration at time (t), C₀ denotes the concentration at time zero, and kapp indicates the apparent rate constant (Haleem et al., 2020; Zahid et al., 2022). Figure 8D depicts the graphical representation of the rate of reaction. The slopes of these plots were used to calculate the values of kapp. Under the same set of reaction conditions, the apparent rate of reaction for the reduction of MB under the sun simulator and UV was found to be −0.00919 and −0.0164 min⁻¹, respectively, which reflects the higher degradation rate of MB under UV.

Features and performance of PES, TFC and TFN

Fourier-transform infrared spectroscopy (FTIR)

FTIR was used also to verify the introduced thin film and NPs onto the nanocomposite membrane. Figure 9 shows PES, TFC, and TFN FTIR spectra. The PES and PA have bonds around 3100 cm⁻¹ for main aromatic C–H stretches, also at 1500 cm⁻¹ benzene ring stretches, and the S =O bands (1300 and 1150 cm⁻¹) can be noticed. The main characteristic peaks for the PA was observed in TFC and TFN, absorption band at 3300 cm⁻¹ correspond to the N-H stretching, the absorption band at 1480 cm⁻¹ for the N-H bending, and at 1630 cm⁻¹ for the amide C =O stretching bond. An extra peak related to Ti =O NPs around 500 nm is shown in the TFN spectra, which confirm the incorporation of the NPs within the PA polymer matrix (Ahmadi & Jafarizadeh-Malmiri, 2021).

X-ray Diffraction (XRD) and X-ray Fluorescence (XRF)

Figure 10 shows the energy dispersive X-ray fluorescence (EDXRF) spectra for TFC, TFN membranes. Based on qualitative comparison, the differences between TFC and TFN membranes spectra was the presence of titanium (previously examined alone as powder; Fig. 4), which confirmed the incorporating the TiO₂ NPs within the PA active layer. Both spectra show Ti-K α and Ti-K β signals and non-indexed peaks in XRF patterns are associated with membrane materials (PES and PA) as shown in Fig. 9, this contributed to the successful fabrication of nanocomposite thin film membrane surfaces. The energy peaks in both spectra were around 2.310 and 2.960 related to the distinctive S K α and Ar K α X-ray emission lines respectively; the presence of Ar K α in the spectra was attributed to the amount of Ar in the atmosphere. The presence of S on the PES membrane could be the cause of the measured S, according to an elemental analysis of PES (Chen et al., 2022). It is worth mentioning that some elements such as hydrogen, carbon, and oxygen, cannot be detected by XRF due to their weak fluorescent X-ray. However, polymers are organic compounds containing mainly hydrocarbons, and an XRF cannot be used to analyze the pure polymer (Alqaheem & Alomair, 2020).

Scanning electron microscope (SEM)

SEM was used also to investigate the microstructure of the deposited thin film and the effect of TiO₂ nanoparticles on the PA thin film microstructure and morphology.

Figures 11A, 11B and 11C show the PES, TFC, and TFN PES top surface, respectively. The tidy PES substrate membrane displayed the featureless, reasonably smooth shape of the surface. The TFC surface displays a typical shape of the ridge-and-valley structure after the interfacial polymerization reaction. Meanwhile, the topography of TFN with 5% TiO₂ NPs indicated a reduction in the ridge-and-valley surface shape, due to filling the ridge and valley with nanoparticles according to Asadollahi et al. (2020). An identical bottom SEM images for the PES, TFC, and TFN membrane were clearly shown in Figs. 11D, 11E and 11F; all images show the same porous structure with nearly the same pores and shape due to the fact that all of them have the same substrate (PES) at the bottom surface and the other layers (TFC and TFN) were only added on the top surface of PES. Membranes cross-sections of PES, TFC, and TFN are depicted in (Figs. 11G, 11H and 11I). All membranes are expected to have a characteristic asymmetrical structure. The active layers on top surface in the TFC membranes are distinguished with high roughness in comparison to the PSF membrane and TFN. All cross sections have a sponge-type structure of PES membrane support.

Thermogravimetric analysis (TGA)

TGA was used also to investigate the effects of membranes modification on their decomposition behavior, and the percentage of TiO₂NPs within TFN membranes. Figure 12 depicts the TGA characteristics of PES, TFC, and TFN membranes. The zone that appeared below 400 °C is stable in all samples. The concentration and distribution of TiO₂ in the polyamide thin film are probable to be the most significant factors that influence the composite products’ properties. It is hypothesized that the inhomogeneous dispersion of TiO₂ NPs at relatively high concentrations may be attributed to the disordering of polyamide chains, thereby minimizing interfacial adhesion between NPs and the polyamide network (Rajaeian et al., 2013).

The preliminary area in weight loss for all samples seemed to be below 100 °C, which can be attributed to the evaporation of the residual organic solvent and adsorbed water in composites after drying. After 400 °C, the TFN had less weight loss in the entire temperature range, which can be attributed to the partial dehydroxylation of the TiO₂ NPs (Rajaeian et al., 2013). The drop appears at around 500 °C, which is the decomposed heat of PES as the main phase in the membranes. It can be seen that all PES, TFC, and TFN membranes go through primary thermal degradation; at around 500 °C for PES and around 450 °C for TFC and TFN, which could be attributed to the sulfonic group in polymer chains; furthermore, another weight loss stage occurred above 600 °C, which corresponds to the main polymer backbone splitting in the presence of the S =O group, which can be bonded with H₂O, causing the formation of intermolecular molecular hydrogen bonds (Kotp et al., 2017). Weight reduction in the TFC and TFN is more constant below 600 °C, as demonstrated by a reduction in mass loss slope associated with carbon molecule detachment above 600 °C. TFN was found to be more stable above 600 °C due to the existence of TiO₂ NPs and higher residual at 700 °C (Kotp et al., 2017). The apparent negative mass loss in the TGA thermograph (mass gain above 100%), can be attributed to the fact that the precision and accuracy of the TGA depend on different factors, such as the weight disturbances at the beginning of heating, continuing weight oscillations, miscellaneous observations related to disorders in gases, and erratic buoyancy disturbances. Thus, the first weight deviation of a TG curve typically begins slightly but noticeably before the recorded temperature increases (Czarnecki, 2015).

Atomic Force Microscope (AFM)

Figure 13 shows the AFM images topography of PES, TFC, and TFN. Table 1 summarizes the average roughness (Ra). The results indicated that polymerized thin film composite (105.8°) membrane has rougher morphologies compared with the PES support membrane (13.5°). For TFC, a uniform distribution of ridge-to-valley morphology can be observed. According to the roughness quantitative data, the TFC membrane has a rougher surface, which is consistent with SEM observations and explains the higher rejection than TFN (43.9°). The SEM and the AFM results are consistent with each other. The same profile of the distribution of the TiO₂ NPs in the SEM images are shown in the AFM images. The SEM showed that there was a good distribution of the nanoparticle’s aggregates, which means it is in a poor dispersion, but with good distribution and with the homogenous distribution of the aggregates shown from the AFM images.

Drop shape analyzer

The contact angles of the blend membranes’ surfaces were determined using the sessile drop method. Figure 14 depicts the contact angles of the PES, TFC, and TFN membranes. The PES composite support has the highest water contact angle of 91° so it has the lowest hydrophilic surface when compared to other thin film polymerized membranes (TFC at 79°, and TFN of 72°) which can be attributed to the presence of strong hydrophilic polar amide functional groups inserted onto the PES which increase the hydrophilicity of the PES surface. Because TiO₂ nanoparticles are hydrophilic, incorporating NPs within the thin film layer provides a slightly increase in the hydrophilicity of the TFN membrane compared to TFC membrane. The contact angleis in consistence with the roughness results from the AFM, the surface roughness decreased the contact angle for a droplet on a hydrophilic surface and increased the contact angle for a droplet on a hydrophobic surface, according to Wenzel’s equation (Wenzel, 1936), and led to an enhancement of polar interaction with water droplets; thus, a reduction in water contact angle. The results were in consistence with the results of Rajaeian et al. (2013), which can be attributed to the filling of TiO₂ NPs in the ridges and valleys; thus the membrane’s surface become smoother.

Membrane performance

The effect of using different salt and different concentrations

The rejection experiments were carried out using different salts (MgCl₂, CaCl₂, and NaCl) at a concentration of 1,000 ppm to evaluate the PA membrane’s salt rejection capabilities and they were almost the same in the range of 80–85% rejection (Fig. 15A). Different theories (preferential sorption, electrical) can explain salt transport through a thin film composite membrane (Bhattacharya & Ghosh, 2004). Water preferentially sorbs at the membrane solution interface and then moves through the capillaries of the membrane via viscous flow, with the flux being directly proportional to the effective pressure, according to the preferential sorption theory. Ions move through membrane pores via convection, diffusion, and the electric potential gradient, according to the electrical theory. From a chemical point of view, the formation of -COOH functionality in the cross-linked polyamide is essential for repelling salt and polar compounds. According to Dai et al. (2002), and Joshi, Singh & Bhattacharya (2011), the membranes reject high-valence anions and low-valence cations more than low-valence anions and high-valence cations. Finally, the unexpected higher rejection of sodium chloride for the TFC membrane compared to MgCl₂ and CaCl₂ is most likely due to diffusion-driven permeation. When the solution reaches a high ionic strength, charge repulsion is screened and molecular size and diffusion become the dominant ion rejection mechanisms which is in consistent with results for Tajuddin et al. (2019); also, the unexpected increase of monovalent ion rejection than divalent ions may be due to their larger hydrated ionic radii, the Mg²⁺ and Ca²⁺ ions have high diffusivity of through the membrane according to Gallab et al. (2017). Figure 15B illustrates the effect of different NaCl concentrations. Increasing the salt concentration decreases the rejection. Salt concentration has little effect on flux in the range of 0 to 2,000 ppm and has a greater impact in the range of 5,000 ppm. These findings revealed differences in flux as NaCl concentration increased. Concentration polarization explains the increase in flux values caused by an increase in flow velocities. The mass transfer effect caused by concentration polarization is small at low salt concentrations and easily prevents this effect by a water flow, resulting in no significant change in flux. This is what causes the relatively large variances in flux found in high salt concentrations and small variances in low-range salt concentrations. Our findings agree with those of Krieg et al. (2005) and Koyuncu & Topacik (2003).

Flux and salt rejection

Figure 16 illustrates the flux and salt rejection of TFC and TFN membranes. The hydrophilicity of the TFN increased with high TiO₂ NP concentration, but NaCl rejection marginally decreased. The primary factor for immigrating TiO₂ nanoparticles on the membrane surface relates to the presence of carboxyl and amide groups following polycondensation which increases the hydrophilicity of the thin film consistently with increasing contact angle. A slight decrease in salt rejection for TFN is assumed to be due to defects formed at the interface of inorganic materials and polymer matrix because of high loading, which is consistent with findings of Kim et al. (2016). With 5% wt TiO₂ NPs, flux increased because of the hydrophilic nature of TiO₂ NPs, while salt rejection decreased slightly due to the trade-off relationship between water flux and salt rejection (Huang et al., 2013). After the filtration experiment, the mechanical strength of the membranes decreased with critical concentration, resulting in easier peeling of the PA-TiO₂ layer from the PES substrate. These findings could be explained by the fact that at high TiO₂ concentrations, the interference of interfacial polymerization by TiO₂ nanoparticles becomes significant, resulting in a lower degree of polymerization of PA (Lee et al., 2008; Huisman, Prádanos & Hernández, 2000). TFN has higher reversible fouling by 31.15% and lowers irreversible of 16% compared to 28.53% and 38% for TFC. The recovery rate was also increased in TFN upon addition of nanoparticles from 61.77% to 83.2%, while the total fouling ratio was reduced from 66% to 47% due to improving the antifouling properties.

Antifouling performance

Water flux recovery after fouling with BSA solution was used to analyze the antifouling performance of the pristine TFC and TFN. Figure 17 and Table 2 depict the results for pristine membrane, and TFN. The flux recovery value of a TFC membrane was only 61.77%, indicating its ability for antifouling properties. The flux recovery percentage of the TFN membrane was 83.2%, which was related to a 5% TiO₂ membrane. Electrostatic force (Coulomb force), hydrogen-bonding force, hydrophobic force (entropy effect), and Van der Waals forces are the forces that cause proteins to adhere to solid surfaces in an aqueous solution, the TFN demonstrated better antifouling properties due to dissociation by the surface of negative membranes (Yang, Xu & Dai, 2006). The strong electrostatic repulsion force between both negatively charged BSA molecules and the membrane surface improved antifouling effectiveness (Huisman, Prádanos & Hernández, 2000). TFN has higher reversible fouling by 31.15% and lowers irreversible of 16% compared to 28.53% and 38% for TFC. The recovery rate was also increased in TFN upon addition of nanoparticles from 61.77% to 83.2%, while the total fouling ratio was reduced from 66% to 47% due to improving the antifouling properties.

Antibiofouling performance

Figure 18A shows the optical density of bacterial culture growths in the presence of TiO₂ NPs in a plate reader at 600 nm. The optical density decreased as TiO₂ NP concentrations increased in all types of tested patterns. The inhibition rate is shown in Fig. 15B and reaches 72% and 93.2% with 1,000 ppm TiO₂ NPs in E. coli and S. aureus respectively. The effect of TiO₂ NPs against gram-negative bacteria was less than for gram- positive bacteria due to the difference in membrane structures of bacteria. This effect may be according to mechanism of releasing the free radical according to Khorshidi et al. (2018). Or may be attributed to the fact that metal oxide NPs have a positive charge whereas microorganisms have a negative charge; this produces an electromagnetic attraction among microorganisms and metal oxide NPs, resulting in oxidation and death of the microorganism (Khorshidi et al., 2018; Khashan et al., 2021). The inhibition rate increases with increased NPs concentration, the breakdown of the bacterial surface membrane by reactive oxygen species (ROS), especially hydroxyl radicals (OH), which results in phospholipid peroxidation and ultimately cell death, is the cause of the antibacterial action of TiO₂ NPs (Khashan et al., 2021). These ROS can break down organic materials and stop cellular activity with higher activity in gram-positive (Ripolles-Avila et al., 2019).

According to the results shown in Table 3, the antibiofouling activity of TFN membrane against Gram-negative (E. coli) and Gram-positive (S. aureus) bacteria was significant with high sterilization ratio of 99.9% TFC has an antimicrobial effect after 24 h of incubation and cultivation according to the nature of polyamide similar the result of Zhai et al. (2020) and Khoo et al. (2021). This could be attributed to TiO₂ NPs’ major antibacterial activities. The generation of reactive oxygen species (ROS) is thought to cause bacterial membrane damage and, as a consequence, cell death (Al Mayyahi, 2018). TiO₂ affect the cell wall and membrane by damaging DNA and ceasing their replication since bacteria is known to have natural antioxidants and enzymatic antioxidant defense systems that inhibit lipid peroxidation and the effects of ROS radicals. When OH2^⋅ − and OH^⋅ are exceeded, a set of redox reactions can lead to the death cell by the alteration of different essential structures such as the cell wall and DNA (López de Dicastillo et al., 2020; Abu-Dalo et al., 2019b). Table 4 shows previous research using TiO₂ with polyamide membranes, as well as recent studies to solve the problem of fouling. The method of preparing nanoparticles differs in this study, as does the use of a spin coater in the preparation of a thin film substrate for the first time, when compared to traditional by-knife casting, the spin coating process increased solvent evaporation and reduced coat production time as well as reduce the hand person error through casting according to Burmann et al. (2014). When compared to titanium prepared in various ways, we find that the properties are similar in terms of increasing the flow of water and maintaining or slightly decreasing salt rejection. As shown in Table 4, nanoparticles such as silver and copper have recently highly contributed to improving fouling resistance by up to 90%.