Introduction

Membrane separation technology is widely used for getting potable and industrial grade water for various applications [1, 2]. Fouling of membrane affects the hydraulic permeability and forces frequent replacement of the fouled membrane thereby increasing the overall cost of the process [3, 4]. Therefore, the membrane fouling has been a subject of interest for researchers since long [5]. The biofouling of the membrane is initiated by the adhesion, growth, and multiplication of one or more bacterial species and/or flocs onto the membrane surface, which eventually leads to the formation of a cake layer on the membrane surface [6]. In recent studies, it has been concluded that the prime factor affecting the membrane fouling is its interfacial property [7] which controls membrane-foulant interactions hence, the adsorption of foulant on the membrane surface [7, 8]. Hydrophilicity and hydrophobicity are the prime factors behind fouling, and it has been recognized that hydrophilic membranes are less prone to fouling compared to hydrophobic membranes [9, 10]. Based on this information, strategies such as graft polymerization, chemical grafting, and surface coating have been considered to improve the surface hydrophilicity and modify membrane characteristics [2]. The main disadvantages of these approaches include weak interaction between the polymer and additives (blending) which affects the long-term durability or instability of the coated layer (surface modification) causing its release from the membrane [2].

Among various methods used, incorporation of inorganic particles into the polymer matrix to form a composite membrane is an effective approach to enhance membrane hydrophilicity and antifouling property [11]. Different inorganic nanomaterials, such as TiO2 [12, 13], SiO2 [14], Al2O3 [15], graphene oxides [16], and carbon nanotubes [17] have been widely used to fabricate composite membranes. However, nano-sized TiO2, due to its superior hydrophilicity, antibacterial property, and chemical stability has received much attention [18]. Addition of TiO2 nanoparticles improves the system’s antifouling property due to changes in hydrophilicity, porosity, zeta potential, and surface roughness [3].

Further, preparation of anti-biofouling membranes by coupling antimicrobial materials with polymeric matrix has also attracted the attention of many researchers in both academia and industry. There are two interesting approaches to kill bacteria before colonization: (1) release killing, and (2) contact killing [19]. Leaching of antibacterial agents on the membrane surface is termed as “release killing”, but there is continuous depletion of biomaterial that may cause environmental risks and significantly reduces the antibacterial efficiency. As a result, the need for developing composite membranes with the contact-killing surface and stable and long-term anti-biofouling activities has attracted the attention of researchers [19]. Inorganic metal oxides are being more preferred over organic antimicrobial agents due to their stability, robustness, and long shelf life [20]. Titanium dioxide (TiO2) has been the widely used nanomaterial because of its stability, hydrophilicity, nontoxic, and antifouling properties [21,22,23].

Most of the available published works have reported the experimental methods to characterize the antifouling property of the membrane. A systematic investigation of the interaction of nanoparticles in improving the antifouling property is rarely known. In this work, the efficacy of these membranes has also been investigated by using extended Derjaguin–Landau–Verwey–Overbeek (XDLVO) theory, for the first time. The experimental results are compared with the theoretical predictions approach for evaluating the antifouling property of the membranes and its quality changes by the addition of nanoparticles.

Further the effect of loading of TiO2 nanoparticles on the antifouling property of PVDF membrane, obtained by phase-inversion membranes, is evaluated by measuring the pure water flux and rejection efficiency of bovine serum albumin (BSA) and humic acid (HA). The bactericidal behaviour and antimicrobial effects are studied using Gram-negative bacteria (Escherichia coli).

XDLVO theory

The extended Derjaguin–Landau–Verwey–Overbeek (XDLVO) theory explains the exact role of hydrophilicity/hydrophobicity and all other physiochemical factors on the membrane surface energy (interfacial interaction between membranes and foulants) [3, 24]. The theory describes that the attachment of foulant with membrane surface is due to three types of interactions: (1) Lifshitz–van der Waals (LW), (2) polar or Lewis acid–base (AB), and (3) electrostatic energy (EL) [25,26,27] interactions.

The XDLVO theory considers the attachment of foulant to the membrane surface by evaluating the interaction of surface energies [28]. According to Oss [29], the total free energy of adhesion is the result of contribution of these three parameters and is expressed as:

$$U_{{{\text{mbc}}}}^{{{\text{XDLVO}}}}={\text{ }}U_{{{\text{mbc}}}}^{{{\text{LW}}}}+{\text{ }}U_{{{\text{mbc}}}}^{{{\text{EL}}}}+{\text{ }}U_{{{\text{mbc}}}}^{{{\text{AB}}}},$$
(1)

where \(U_{{{\text{mlc}}}}^{{{\text{XDLVO}}}}\) is the total interfacial free energy between the membrane and foulant, \(U_{{{\text{mbc}}}}^{{{\text{LW}}}}\), \(U_{{{\text{mbc}}}}^{{{\text{EL}}}}\), and \(U_{{{\text{mbc}}}}^{{{\text{AB}}}}\) represent individual components of the total interfacial energy, whereas m, b and c represent membrane, bulk liquid (water) and foulant (BSA), respectively. The individual energy is calculated using Eqs. (2), (3), and (4) as follows:

$$U_{{{\text{mbc}}}}^{{{\text{LW}}}}={\text{ }}2\pi \Delta G_{{{h_{\text{o}}}}}^{{{\text{LW}}}}h_{{\text{o}}}^{2}\frac{a}{h},$$
(2)
$$U_{{{\text{mbc}}}}^{{{\text{AB}}}}={\text{ }}2{\text{ }}\pi a{\text{ }}\lambda \Delta G_{{{h_{\text{o}}}}}^{{{\text{AB}}}}{\text{exp}}\frac{{({h_{\text{o}}} - h)}}{h},$$
(3)
$$U_{{{\text{mbc}}}}^{{{\text{EL}}}}={\text{ }}\pi \varepsilon a\left\{ {{\text{2}}{\zeta _{\text{c}}}{\zeta _{\text{m}}}{\text{ln}}\left( {\frac{{1+{{\text{e}}^{ - kh}}}}{{1 - {{\text{e}}^{ - kh}}}}} \right)+({\zeta _{\text{c}}}^{2}+{\text{ }}{\zeta _{\text{m}}}^{2}){\text{ln (1}} - {{\text{e}}^{ - 2kh}}{\text{)}}} \right\},$$
(4)

where a is radius of foulant (BSA), h is the separation distance between membrane and foulant, ho is minimum separation distance (0.158 nm), λ is decay length (value taken as 0.6 nm for aqueous solution) [30], ζc, ζm are the zeta potential of foulant and membrane, respectively, k is the inverse Debye screening length, whereas \(\Delta G_{{{h_{\text{o}}}}}^{{{\text{LW}}}}\) and \(\Delta G_{{{h_{\text{o}}}}}^{{{\text{AB}}}}\) can be calculated using Eqs. (5) and (6), respectively, as follows:

$$\Delta G_{{{h_{\text{o}}}}}^{{{\text{LW}}}}={\text{2}}\left( {\sqrt[2]{{\gamma _{{\text{b}}}^{{{\text{LW}}}}}} - \sqrt[2]{{\gamma _{{\text{m}}}^{{{\text{LW}}}}}}} \right)\left( {\sqrt[2]{{\gamma _{{\text{f}}}^{{{\text{LW}}}}}} - \sqrt[2]{{\gamma _{{\text{b}}}^{{{\text{LW}}}}}}} \right),$$
(5)
$$\Delta G_{{{h_{\text{o}}}}}^{{{\text{AB}}}}={\text{2}}\sqrt[2]{{\gamma _{{\text{b}}}^{+}}}\left( {\sqrt[2]{{\gamma _{{\text{m}}}^{ - }}} - \sqrt[2]{{\gamma _{{\text{c}}}^{ - }}} - \sqrt[2]{{\gamma _{{\text{b}}}^{ - }}}} \right)+2\sqrt[2]{{\gamma _{{\text{b}}}^{ - }}}\left( {\sqrt[2]{{\gamma _{{\text{m}}}^{+}}} - \sqrt[2]{{\gamma _{{\text{c}}}^{+}}} - \sqrt[2]{{\gamma _{{\text{b}}}^{+}}}} \right) - 2\left( {\sqrt[2]{{\gamma _{{\text{m}}}^{+}}}\gamma _{{\text{c}}}^{ - }+\sqrt[2]{{\gamma _{{\text{m}}}^{ - }}}\gamma _{{\text{c}}}^{+}} \right),$$
(6)

where \(~{\gamma ^{{\text{LW}}}}\), \({\gamma ^+}\), \({\gamma ^ - }\) are LW component, electron acceptor and donor parameter, in the order given. The surface tension of membrane (\(\gamma _{{\text{m}}}^{{{\text{LW}}}},\gamma _{{{\text{m}}}}^{ - }~,~\gamma _{{\text{m}}}^{+}\)) and BSA were quantified using extended Young’s equation. This equation shows the relationship between contact angle of a liquid and surface tension parameter of both liquid and solid surfaces and is evaluated using Eq. (7):

$$(1+\cos \theta )\gamma _{{\text{l}}}^{{{\text{TOT}}}}=2\left( {\sqrt[2]{{\gamma _{{\text{s}}}^{{{\text{LW}}}}}}\gamma _{{\text{l}}}^{{{\text{LW}}}}+\sqrt[2]{{\gamma _{{\text{s}}}^{+}}}\gamma _{{\text{l}}}^{ - }+\sqrt[2]{{\gamma _{{\text{s}}}^{ - }}}\gamma _{{\text{l}}}^{+}} \right),$$
(7)

where θ is the contact angle and \(\gamma _{{\text{l}}}^{{{\text{TOT}}}}\) is the sum of LW and AB and is calculated using Eq. (8):

$$\gamma _{{\text{l}}}^{{{\text{TOT}}}}={\gamma ^{{\text{LW}}}}+{\gamma ^{{\text{AB}}}}.$$
(8)

The polar component \({\gamma ^{{\text{AB}}}}\) of the material is expressed by Eq. (9):

$${\gamma ^{{\text{AB}}}}=2{({\gamma ^+}{\gamma ^ - })^{0.5}}.$$
(9)

Experimental

Materials

Polyvinylidene fluoride (PVDF), n-methyl-2-pyrrolidone (NMP), and TiO2 nanoparticles (Degussa P25 (25 nm), a hydrophilic metal oxide) were purchased from Sigma-Aldrich. Diiodomethane, glycerol, nutrient broth (NB) as media and nutrient agar (NA) used were obtained from High Media (Bombay), bovine serum albumin (BSA) and Humic acid (HA) were obtained from SDFCL (Bombay), and phosphate buffer saline (PBS) was obtained from Merck. Double distilled water (DD) used in all experiments was prepared in the laboratory.

Synthesis of PVDF/TiO2 composite membranes

The PVDF composite membranes were synthesized by a phase-inversion method. The PVDF pellets were dried in an oven at 90 °C for 24 h. 4 g of the dried PVDF pellets was dissolved in 20 mL of organic solvent NMP using a magnetic stirrer and a temperature of around 65–70 °C to achieve a homogeneous solution. Simultaneously different quantities of TiO2 were dispersed in 4 mL of NMP by sonication for 1 h. Each of these TiO2 suspensions was mixed separately with 20 mL of PVDF, and these were then thoroughly mixed together with continuous stirring for 8 h at 65–70 °C to achieve a uniform suspension. This solution was subsequently cast as a film on a glass plate with a clearance of 200 µm and exposed to air for 30 s for partial evaporation of the solvent. The plates coated with the polymer-TiO2 suspension were then placed in a water bath at ambient temperature (27 ± 1 °C) for 24 h for precipitation. The membranes were peeled off from the glass plates, washed and stored under water for further study. Membranes with different loading of TiO2 are named as M1 (0 g TiO2/g PVDF), M2 (0.005 gTiO2/g PVDF), M3 (0.01 TiO2/g PVDF) and M4 (0.02 TiO2/g PVDF).

Physiochemical properties of membranes

Contact angle measurement

The wetting characteristic of the membranes was evaluated by measuring the contact angle of the composite membrane surface with a contact angle goniometer [KRUSS, Germany] using the sessile drop method. 3 µL droplet of distilled water was used as the probe liquid at the ambient temperature.

Morphological characterization

The Smart Lab X-Ray Diffractometer (Rigaku Smart Lab Powder type, without χ-cradle) was used for X-ray diffraction studies. The equipment operated at 18 kW and used a Rotating Anode XRD (1200–1800 K temperature) containing CuKα radiation (λ = 1.5406 Å). The membrane surface was examined using a high-resolution scanning electron microscope (HRSEM, SUPRA 40, Zeiss). The membrane films were initially coated by sputtering with gold, and the photomicrographs were taken under very high vacuum. The scanning probe microscope (NTEGRA Prima) was used to detect the surface roughness of the synthesized membranes under the non-contact mode.

Antibacterial characteristics

Halo zone test

The antibacterial efficacy of PVDF/TiO2 membranes was evaluated experimentally using the halo zone method. Prior to the tests, all the materials used were autoclaved at 120 °C for 1 h to ensure sterility. Cultivation of E. coli was carried out for 24 h in a conical flask containing the sterilized solution of NB (1.3 g NB/100 mL water) placed in an incubator maintained at 37 °C. During this period, the bacteria were fully grown, and then 100 µL of the bacterial suspension was spread over the NB agar plate composed of 1.3 g of NB and 2.3 g of NA in 100 mL distilled water. The UV-treated membrane pieces of approximately 1 cm diameter were now placed on the agar plates and again incubated at 37 °C for 24 h. The extent of inhibition zone formation around the membrane pieces indicated the antibacterial behaviour of the membranes [31, 32].

Bacterial growth

All glass-wares, the nutrient broth solution, and all other solutions were autoclaved at 120 °C for 60 min. The membrane pieces of known area (1 cm × 1 cm) were first rinsed three times with 0.01 M fresh sterile phosphate buffered saline (PBS, pH 7.4) followed by sterilization through UV irradiation for half an hour. A bacterial suspension (10−3 cells/100 mL) was prepared in the sterile nutrient broth. The UV-treated membrane pieces were then immersed in the above suspension. Subsequently, the flasks containing pieces of different composite membranes and bacterial suspensions were incubated in a rotary shaker at 100–105 rpm at 37 °C in the absence of any light. The growth of E. coli was measured in terms of optical density at 600 nm (OD600) [33] at several incubation times (t = 0, 6, 12, 18, and 24 h) using a spectrophotometer (SYSTRONICS, PC Based Double Beam Spectrometer 2202). Each reported value was obtained by averaging ten individual measurements. Data of the exponential growth phases were fitted using the pseudo-first-order kinetics:

$$\ln \left( {{x_t}/{x_0}} \right){\text{ }}={\text{ }}\mu t,$$
(10)
$${\tau _{\text{d}}}={\text{ }}\ln 2/\mu ,$$
(11)

where x0 and xt represent values of OD600 for the bacterial suspensions at 0 and t h, respectively, µ is the specific growth rate (h−1), and τd represents the doubling time (h).

The adhesion and growth of E. coli on membrane surface were observed using a HRSEM (high-resolution scanning electron microscope).

Water permeation studies

The pure and composite PVDF membranes with the filtration area of 15.5 cm2 were used in a flow filtration cell to test the fouling behaviour using BSA as the model foulant (1 g/L, pH 7.5). Initially, the synthesized membranes were compacted at 0.2 MPa to achieve a stable flux, and then a pure water flux Qw (kg/m2 h) was measured at 0.1 MPa for 1.5 h and then it was replaced by the BSA solution. Permeation of BSA was recorded as QF and concentrations of feed and permeates were evaluated using UV spectrophotometer by measuring the absorbance at 280 nm [34].

$${Q_{\text{w}}}={\text{ }}M/(A \times t),$$
(12)

where M is the permeate mass (kg), A is the membrane area (m2), and t is the permeation time (h). The rejection percentage of BSA was calculated from Eq. (13):

$$R={\text{ }}\left( {1 - {\text{ }}{B_{\text{p}}}/{B_{\text{f}}}} \right){\text{ }} \times 100,$$
(13)

where Bp and Bf (mg/mL) are the concentrations of BSA in permeate and feed, respectively.

The fouled membranes were rinsed with water, and the water permeability was again measured as QWR. The regeneration test was carried out to analyze the recovery and self-cleaning ability of membranes which was calculated using Eq. (14):

$${\text{FRR }}={\text{ }}{Q_{{\text{WR}}}}/{Q_{\text{w}}} \times {\text{1}}00,$$
(14)

where QWR is the water flux after each cycle (kg/m2 h) and Qw is the initial pure water flux (kg/m2 h).

The filtration experiment with 5 mg/L HA at 0.2 MPa for 1.5 h was conducted to further evaluate the antifouling property of composite membrane, following the same procedure as described above by replacing BSA with HA. The water flux before and after contact with HA solution was measured, and flux decline was calculated in terms of relative fluxes.

Results and discussion

Physiochemical properties of the membrane

XRD analysis

The morphology of the membranes was studied using X-ray diffraction pattern as shown in Fig. 1. The PVDF has four different crystalline structures α, β, γ, and δ, where α-phase is the kinetically stable phase, and β-phase is the thermodynamically favoured phase [18]. At the diffraction angles 18.5° and 26.5°, the peak corresponds to α-phase, whereas at 20.5° for β-phase [19]. The additional two peaks for membranes M2, M3, and M4 with increasing intensity at 2 Ø equal 25.2° and 48.0° confirm the presence of anatase phase of TiO2 with (101) and (200) planes [20] which provides a better antifouling behaviour. The major peak intensity of PVDF (20.5°) in PVDF/TiO2 membranes is lower compared to those for pure PVDF membrane. This decrease in intensity is due to the intercalation of TiO2 nanoparticles. The introduction of nanoparticles results in an amorphous region.

Fig. 1
figure 1

XRD patterns of PVDF/TiO2 membranes

Full size image

AFM analysis

The AFM photomicrographs given in Fig. 2 show the effect of TiO2 loading on the membrane surface roughness. In a 3D AFM image, the bright portion represents the highest peak on the membrane surface, whereas the dark portion represents valleys. It is a known fact that smoother the membrane surfaces the lesser will be the fouling tendency [35]. In case of membranes with a rough surface, accumulation of foulants will occur within the “valleys” and such membranes will foul easily. Figure 2 depicts that membrane M1 (no TiO2) exhibits several large peaks and valleys, i.e., it has a very rough surface. The decrease in roughness parameter indirectly enhances the antifouling property of the polymeric membrane. Roughness is defined by three parameters: (1) mean roughness (Ra) defined as average deviation of the z values, which is half the average peak to valley depth, (2) root mean square roughness (Rq) which represents standard deviation of an entire z values, and (3) average of height (Rz) expressed as the difference of the largest positive and negative z values [3].

Fig. 2
figure 2

3D AFM image of different PVDF composite membranes

Full size image

The values of these roughness parameters for different composite membranes are listed in Table 1, (AFM scanning area of 5 µm × 5 µm). The results indicate that surface roughness decreases with increase in the concentration of TiO2, but the membrane M4 with the highest loading exhibits higher roughness; this is likely to be due to the presence of aggregated particles [36]. Excessive addition of TiO2 particles in the matrix results in bumps formation hence, increases the peak value. Similar results are also reported in the literature [37, 38].

Table 1 Roughness parameter of PVDF and its composite membrane
Full size table

Surface hydrophilicity and wetting ability of pure and composite membranes

The wetting ability of the polymeric membrane surface was determined by measuring the static contact angles. The observed decrease in the value of contact angle for PVDF/TiO2 membranes compared to pure PVDF membrane can be attributed to the high affinity of TiO2 for water [39]. The contact angle of pure PVDF membrane (86.7°) is higher than that of PVDF/TiO2 composite membranes and with an increase in TiO2 loading from 0 to 0.01 g TiO2/g PVDF it decreases from 86.7° (M1) to 72.6° (M3). These results are consistent with earlier reports [39, 40]. The wetting ability of membranes is a function of the surface hydrophilicity and roughness [41]. It was also observed that at high concentration of TiO2 (M4 membrane), this value increases further; this may be attributed to the particle aggregation resulting in non-uniform distribution of particles within the membrane matrix [41] as shown in Fig. 3. This value indirectly gives an idea that hydrophilicity of composite membranes is enhanced due to the presence of hydrophilic TiO2 particles hence, leads to an improvement in the antifouling property. Based on these data, the free energy and surface tension parameters were evaluated using equations defined in the XDLVO theory section.

Fig. 3
figure 3

SEM images of PVDF/TiO2 composite membrane

Full size image

XDLVO theory-based analysis of fouling behaviour

A better understanding of the fouling mechanism of the membranes can be obtained by estimating the physicochemical interactions between the foulant and composite membrane. The surface tension parameters of membranes were calculated using the Young–Dupré equation using the measured contact angles of three different probe liquids with known parameters (Tables 2, 3, 4). The results show that the electron donor component (\({\gamma ^ - }\)) value is always greater than the electron acceptor component (\({\gamma ^+}\)) value, indicating that all the membranes exhibit high electron donor mono-polarity that varies from 1.522 to 2.804 mJ/m2. The results are consistent with the reported values [42]. It was also found that membrane M4 has the highest values of (\({\gamma ^+}\)) and (\({\gamma ^{{\text{AB}}}}\)) compared to other membranes, indicating high polar properties.

Table 2 Contact angle value of three different probe liquids
Full size table
Table 3 Surface tension values (mJ/m2) of three probe liquids
Full size table
Table 4 Surface tension and surface free energy of membranes
Full size table

The high values of electron donor components (γ−) for TiO2-modified PVDF membranes indicate mono-polar hydrophilic surface due to the presence of hydrophilic TiO2 nanoparticles.

The increase in γTOT with increasing TiO2 loading in membranes means strong interfacial molecular affinity between the membrane and water molecules. A tightly bound formation of the steric-entropic barrier will restrain the interaction between the membrane surface and foulant, hence delays the fouling [43].

The individual surface free energy AB and LW parameters were calculated using Eqs. (5) and (6). The sum of these two energy components for any given material gives the free energy of cohesion (∆GTOT). It provides a quantitative perception regarding the hydrophilicity and hydrophobicity. The thermodynamic theory suggests that adhesion or attraction between two materials occurs when ΔGTOT is negative. Higher the negative value stronger is the hydrophobicity; intensity in the attraction between membrane and foulant, hence more severe will be the membrane fouling.

It is also observed that ∆GTOT is more negative for pure PVDF membrane, indicating it to be more hydrophobic and, hence, a stronger attraction between membrane and foulant [43]. The value increases on increasing the TiO2 loading, and is significant for membrane M3, indicating a less force of attraction between foulant (BSA) and the membrane. Further increase in TiO2 loading increases the agglomeration of the particle, resulting in a reduction in free energy of cohesion for PVDF/TiO2 membranes. These values were further used to calculate the individual interaction energy parameters, based on which total interaction energy \(U_{{{\text{mlc}}}}^{{{\text{XDLVO}}}}\) was evaluated. The plots of interaction energy as a function of separation distance are shown in Fig. 4. The figure explains the effect of interaction energy on membrane fouling. It is seen that as the separation distance increases the interaction energy gradually approaches zero. The figure also predicts that the foulant needs to overcome the repulsive force to attach to the membrane surface. The higher the repulsive barrier, the harder will be for the foulant to achieve attachment [3]. For the membranes used in this work the energy barrier followed the order M3 > M4 > M2 > M1. As a result, deposition or blocking of membrane surface for M3 is minimal. Thus it is seen that improving the membrane hydrophilicity by adding TiO2 will diminish the membrane fouling [19].

Fig. 4
figure 4

Variation of interaction energy between membrane and BSA

Full size image

These results provide a quantitative picture of the interfacial interactions between foulants and membrane based on XDLVO theory. From the contact angle values and the results of XDLVO theory, it can be said that fouling of membrane can be mitigated by strengthening the interface polarity and hydrophilicity.

Antibacterial property

Halo zone test

The halo zone tests were used to qualitatively investigate the antibacterial property of pure PVDF and TiO2-incorporated membranes against Gram-negative bacteria (E. coli) after 24 h of incubation at 37 °C. Figure 5 shows the strength of the inhibition zone, an area around membrane piece without any growth of bacteria after 24 h. These figures depict that pristine PVDF (M1) being hydrophobic does not exhibit inhibition zone hence has no antibacterial activity. All membranes with TiO2 exhibit inhibition zone formation, but the zones for M3 and M4 membranes are more prominent. Thus, it can be said that the inherent antibacterial property of TiO2 protects the membrane from fouling by not allowing bacteria to grow in the nearby region.

Fig. 5
figure 5

Inhibition zone formation for different PVDF composites membranes

Full size image

Bacterial growth

The Gram-negative (E. coli) bacteria were used to test the antimicrobial activity of pristine PVDF and PVDF/TiO2 membranes. The OD600 values during and after 24 h incubation are shown in Fig. 6. These figures indicate that membranes M3 and M4 exhibit significant inhibition capacity towards E. coli strains.

Fig. 6
figure 6

OD600 value of the M1 (control, no TiO2), M2, M3 and M4 incubated for 24 h

Full size image

The viability of bacteria on the pure and composite membrane surface before and after incubation for 24 h was determined by HRSEM. Figure 7 depicts the adhesion and growth of E. coli on pure PVDF and composite membranes. It is clear that the surface of membrane M1 is covered with a large number of bacteria while that of membrane M3 with the least number. This can be attributed to the disruption of the cell wall due to the presence of nanoparticles [44]. These results are consistent with the results of AFM and contact angle values since membrane M1 having a high value of roughness and being hydrophobic is more susceptible to fouling compared to the membrane M3 with the lowest roughness value as shown in the AFM result in Fig. 2 and the hydrophilic characteristics. The HRSEM images exhibit the growth behaviour of bacteria and it is observed that the trend for the growth of bacteria over the membrane surface is similar to that observed for the biocidal activities obtained from OD600 measurements.

Fig. 7
figure 7

HRSEM images of PVDF–TiO2 membranes before and after bacterial growth

Full size image

The cell proliferation of E. coli was studied for the exponential growth phase for 3–24 h to see the inhibition by nanocomposite membrane. The results reveal that specific growth rate (µ) of E. coli is very high in the presence of pristine PVDF and decreases dramatically with an increase in TiO2 loading. The increase in the doubling time τd (Fig. 8) was also observed. The optimal growth inhibition is not observed for the highest TiO2 loading (M4 membrane), due to agglomeration of particle which leads to the availability of less active sites for killing E. coli cells.

Fig. 8
figure 8

Antibacterial activity of different composite membranes

Full size image

At low concentrations, TiO2 does not cover the entire membrane surface as a result, some bacteria grow. The membrane M3 results in intimate contact between TiO2 particles and the bacteria which affects the bacterial cell wall readily and leads to enhancement in the antibacterial effect [45]. Antibacterial activity is reduced for membrane M4. At higher concentrations, due to particle agglomeration, the surface is not completely covered with TiO2 nanoparticles leaving some space for the bacteria to grow [46]. Our study has confirmed that TiO2 exhibits anti-bactericidal activity, and the result is consistent with that of Rahimpour et al. [45] as well as aggregation of TiO2 NPs results in the loss of available surface area for reducing fouling.

Permeate flux

Performance against BSA

A qualitative evaluation of the antifouling property of membranes was carried out by measuring the water flux with BSA as the model foulant. Incorporation of appropriate inorganic particles is an effective way to enhance the hydrophilicity and antifouling property. The water flux values before and after BSA filtration are shown in Fig. 9. An improvement in flux with an increase in TiO2 loading from 0 to 0.01 g TiO2/g PVDF is observed. This enhancement is due to the hydrophilic nature of TiO2. For membrane M4, flux has declined due to agglomeration of TiO2 particles [46, 47]. In the second set of runs after BSA filtration, it is observed that the decline in pure water flux is highest for pristine PVDF membrane compared to composite membranes. The same trend was followed for the third and fourth cycles with a small change in FRR suggesting the stability performance of the nanocomposite membrane. These results are in agreement with the results obtained from the XDLVO theory, where the energy barrier follows the order M3 > M4 > M2 > M1, so the deposition or blocking of membrane surface for M3 is minimal. The flux recovery ratio (FRR) of composite membranes is shown in Fig. 10a. The graph represents the increased value of FRR for modified membrane indicating better reuse, antifouling and cleaning property of the composite membrane.

Fig. 9
figure 9

Water flux value for different composite membranes before and after filtration of BSA bearing water

Full size image
Fig. 10
figure 10

FRR and % BSA rejection for different PVDF/TiO2 composite membranes

Full size image

The rejection percentage for various membranes is shown in Fig. 10b. This trend is expected because higher flux recovery ratio means less fouling of the membrane surface. Increase in nanoparticle loading reduces the roughness hence the foulant will not accumulate on the membrane surface rather it will be rejected easily resulting in a high flux.

Performance against HA

To further examine the antifouling property of nanocomposite membranes, ultrafiltration of HA was conducted. The variation of relative flux with time is shown in Fig. 11. The figure depicts a decrease in fouling tendency of PVDF/TiO2 membrane compared to pure PVDF. A maximum resistance against fouling was observed for M3 membrane having the lowest contact angle (Table 2) and least roughness (Fig. 2). The probability of adsorption decreases on the membrane surface with the addition of TiO2 nanoparticles which is a consequence of increasing membrane hydrophilicity. Adsorption of foulant is considered as the first step causing membrane fouling, and it is dependent on the physicochemical properties of both the membranes and foulants [48]. It is a well-known fact that hydrophobic surfaces have a higher tendency toward fouling, as a result, adsorption occurs relatively quickly for hydrophobic than hydrophilic surfaces [49]. The addition of TiO2 nanoparticles decreased the roughness, so further mitigated the fouling. Thus it became significantly important to fabricate membrane with less roughness and improved hydrophilicity to enhance the performance as well as antifouling ability. Hence, based on the results of BSA and HA filtration it may be concluded that TiO2-incorporated PVDF membranes possess better antifouling characteristic compared to pristine PVDF with the best result obtained for M3 membrane in both cases.

Fig. 11
figure 11

Flux ratio during filtration of HA solution

Full size image

Hence, it can now be concluded that TiO2 immobilized PVDF membranes exhibit good rejection and show the potential for water flux recovery. Also after deep investigation a comparison of the improved relative percentage of water flux, flux recovery ratio and contact angle for inorganic nanomaterial–PVDF hybrid membranes and PVDF/TiO2 membranes in this work is listed in Table 5. It is seen that TiO2–PVDF membrane presents a promising approach with the other PVDF composite membranes. Compared to other membranes reported in literature the cost of PVDF/TiO2 membranes can be lower due to the absence of other additives such as PVP or PES and low cost of TiO2 when compared to GO NPs and this may broaden the horizon for treatment of wastewater using membrane technology.

Table 5 Comparison of the performance of inorganic-polymer nanocomposite membranes prepared in this work and those reported in the literature
Full size table

Conclusion

The incorporation of TiO2 has enhanced hydrophilicity, reduced surface roughness and improved repulsive interaction energy barrier between the foulant and the membrane, thus resulting in a better antifouling ability of the membrane surface. The XRD spectra have confirmed successful incorporation of TiO2 in the membrane matrix. At higher loading, aggregation of TiO2 particles has resulted in increased surface roughness due to the formation of bumps leading to the reduction in their repulsive interaction and decrease in system’s antifouling characteristics. The composite membrane with optimal TiO2 loading (0.01 g TiO2/g PVDF) has shown an excellent flux recovery ratio (approx. 95%) compared to other membranes when subjected to four cycles of use, thus indicating that membranes with an optimum TiO2 loading can withstand fouling without affecting the flux. This membrane has exhibited the highest antifouling ability. Higher TiO2 loading has resulted in reductions in active surface sites available for the killing of bacteria.