Synthesis and performance of ultrafiltration membranes incorporated with different oxide nanomaterials: experiments and modeling

Ultrafiltration (UF) is a pressurized membrane filtration process for the removal of various macromolecules, colloids, and suspended particles from various solutions and is used in various industrial processes, such as water and wastewater treatments, chemical manufacturing, and food processing. Therefore, membrane contamination, called membrane fouling, is a big problem in UF (Garcia-Ivars et al. 2014). Fouling membrane depends on the properties of the membrane surface, including morphology, porosity, and hydrophilicity (Vatanpour et al. 2012).

As use of the UF process is growing, efforts to enhance the efficiency of UF-processes become increasingly relevant (Garcia-Ivars et al. 2014). Due to their exceptional hydrolytic and thermal stability in addition to strong mechanical properties and chemical stability, sulfone-containing polymers (such as polyethersulfone [PESU] and polysulfones [PSFU]) are commonly used for the preparation of nanofiltration (NF) and UF membranes for water application (AbdulKadir et al. 2018). The fouling tendencies of PESU and PSFU in addition to the general polymer membranes are, however, the main concern for real applications as they result in lower fluxes (an increase of the overall energy demand), reduce the life of the membrane, and cause changes in the efficiency of separation (Luque-Alled et al. 2020). The intrinsic hydrophobicity of the membrane material has been identified by several researchers as one of the primary causes of fouling. Industrial wastewater that needs to be treated contains organic foulants with a biological origin, which are generally attracted to hydrophobic membrane surfaces and holes and hence absorb them (Bassyouni et al. 2019).

The principal approach to reducing polymeric membrane fouling is to prevent excessive absorption or adhesive processes on the membrane surface since this process prevents or at least slows down colloid build-up (Ng et al. 2013). The mixing of polymers and inorganic nanoparticles has recently attracted further attention due to the convenience and mild syntheses required for membrane preparation (Shen et al. 2012). It is considered a possible strategy to render higher permeability and antifouling characteristics by incorporating hydrophilic groups into the membrane. The addition of hydrophilic groups to the membrane increases the permeability and antifouling features of the membrane. The incorporation of hydrophilic groups into the membranes usually prevents the deposition of hydrophobic foulants and increases the water permeate flux by creating a surface water layer that facilitates hydrogen bonding with water molecules (Khan et al. 2018).

In this study, the influence of the addition of different nanomaterials into the membrane polymer was investigated. Graphene oxide (GO), tin oxide (SnO₂), aluminum oxide (Al₂O₃), and titanium oxide (TiO₂) were used, and their effects on water flux and humic acid fouling-resistance properties were studied.

GO was purchased from MKnano (Canada). Aluminum oxide (Al₂O₃), tin oxide (SnO₂), and titanium oxide (TiO₂) were obtained from Acros Organics (Belgium). PESU (PESU3000, molecular weight 62,000–64,000) from Solvay Advanced Polymers was used to fabricate membrane substrates. PVP (average molecular weight 40,000 Da, Loba Chemie) was used as the additive in the casting solution. Humic acid was obtained from Loba Chemie (India).

All the membranes were fabricated using a phase inversion technique. Membrane compositions are shown in Table 1. Each membrane's composition was stirred at 70 °C until it formed a homogeneous solution, which was then cooled to room temperature. In an airtight bottle, the dope solution was degassed for 24 h. Using an Elcometer 4340 motorized film applicator with a traverse speed of 70 mm/s, the dope solutions were then cast on a clean glass plate to form a uniform film with a thickness of 150 μm. The film formed on the glass plate was immersed in a 25 °C tap water coagulation bath. The water in the coagulant bath was changed three to four times to remove any excess solvent.

Using the following equations, membrane porosity and mean pore radius were calculated (Mu et al. 2019):

and are the wet and dry membrane weights, respectively, is the water density (g/m³), A is the membrane are (m²), and l is the membrane thickness (m).

The humic acid solution flux was calculated using the amount of water that permeated the membranes at 5 bar for 2 h. After humic acid solution filtration, the fouled membranes were washed with distilled water for 20 min after which the water flux of cleaned membranes J₂ was measured once more.

A higher FRR value typically indicates a better UF membrane antifouling performance.

The resistance that develops during the filtration process can be an indication that the membrane is fouled. When a cake/gel layer forms on the membrane's surface, fouling happens as a result of adsorption or because of a build-up of contaminants within the membrane's pores.

Scanning electron microscopy (SEM) images of the membrane's surface and cross-section were captured with a JSM-7100 F instrument (JEOL, USA). Humic acid concentration was measured using Lambda 365 ultraviolet/visible (UV/Vis) Spectrophotometer.

Several mathematical models have been developed to describe the causes of pore blocking and cake layer fouling induced by impurities present during the filtration process (Huang et al. 2020). Four traditional single models were proposed by Hermans and Bredée (Hermans 1936). Hermia's models (Hermia 1982) are based on classical constant-pressure dead-end filtration equations. These equations consider four primary types of membrane fouling and include complete, intermediate, and standard blocking and formation of a cake layer. These models can be modified to consider a crossflow configuration (Corbatón-Báguena et al. 2015). Table 2 displays the linear expressions of the four classical single models.

Table 2

Membrane fouling models for the constant-pressure UF process

Model .	Expression .	Parameter .
Cake layer		(s/m2)
Complete blocking		(s−1)
Intermediate blocking		(m−1)
Standard blocking		(m−1)

Model .	Expression .	Parameter .
Cake layer		(s/m2)
Complete blocking		(s−1)
Intermediate blocking		(m−1)
Standard blocking		(m−1)

View Large

Membrane porosities and mean pore sizes are summarized in Table 3. The porosity increased from 80.34% for the pure membrane PES-1 to 83.52, 81.61, 83.13, and 81% for the PES-2, -3, -4, and -5 modified membranes, respectively, while the mean pore size increased from 6.25 nm for PES-1 to 8.06, 6.86, 7.58, and 6.42 nm for PES-2, -3, -4, and -5, respectively. The results showed that the addition of nanoparticles enhanced the mean pore size and porosity (Zhu et al. 2018). During the phase inversion process, hydrophilic polymeric nanoparticles cause a significant increase in the exchange rate of both the solvent and non-solvent, leading to an increase in the mean pore size and a notable boost in membrane permeability. At the same time, the nanoparticles reorganize favorably on the surface, which leads to an increase in hydrophilicity and thus facilitates an increase in the membranes' antifouling properties.

As can be observed in Figure 6, all of the modified membranes displayed a higher water permeability than the pure membrane PES-1; specifically, the PES-2 membrane displayed the highest value of 143.6 L/m² h bar followed by the PES-4 membrane at 92.12 L/m² h bar and 83.68, 75.43 L/m² h bar for PES-3, and PES-5 membranes, respectively. In most cases, the pure water flux increases noticeably due to pore formation and hydrophilicity enhancement. This finding agrees with the sequential order of contact angle obtained for Al₂O₃–SnO₂–TiO₂–GO modified membranes. Thus, the effect of adding metal–oxide particles on promoting pore formation may be dependent on the particle density close to the membrane surface (María Arsuaga et al. 2013; Liu et al. 2018; Esfahani et al. 2019).

Fouling parameters, such as R_ir, R_r, R_t, and F_rr, of membranes were calculated to obtain information about their antifouling properties. A summary of these parameters for non-composite and nanocomposite membranes is shown in Table 4. Since the modified membranes have lower R_t values, this finding indicates that they have higher water flux values and less fouling after UF of the humic acid solution.

Two types of fouling exist: (1) reversible and (2) irreversible. Irreversible fouling is important in membrane filtration performance because the corresponding permeation flux caused by this type of fouling cannot be recovered by cleaning, whereas in the case of reversible fouling, the contaminant is only loosely attached to the surface of the membrane, and it is possible to remove contaminants by washing the membrane with water.

Irreversible fouling (R_ir) of the pure PES membrane (PES-1) was found to be 54.11%, while for the nanocomposite membranes, PES-2–5 had irreversible fouling percentages of 32.87, 32.5, 20.57, and 26.23%, respectively. In the reversible fouling case, when compared with the R_r of a pure membrane (10.24%), the R_r values of modified membranes, PES-2 (23.65%), PES-3 (25.01%), PES-4 (30.95%), and PES-5 (28.03%) are all noticeably higher.

Incorporating nanomaterials has been shown to significantly mitigate membrane fouling, even the most intractable forms of irreversible fouling. As can be seen, the improved antifouling performance can be attributed to the increased hydrophilicity of the membrane on its surface in addition to the surface pore size and cross-sectional structures that are preferable (Xu et al. 2017; Liu et al. 2018).

It is commonly known that a higher flux recovery ratio (F_rr) value indicates greater membrane-associated fouling resistance. As compared to other membranes, PES-1 has a lower FRR of 45.89%. Due to its highly hydrophilic and smooth surface contact with the humic acid solution, PES-4 had the highest FRR of any of the other fabricated membranes (79.4%).

Table 5 displays the values of the fit parameters (k-values) in addition to the coefficient of determination (R²) for membranes.

Two single models (cake layer and standard or intermediate blocking) showed good linear fitting (R² > 0.95) compared to the other two models (cake layer and complete blocking). This finding suggests that many different fouling mechanisms were taking place at the same time when viewed from a statistical point of view.

Membrane fouling intensity may be studied using the physical principle of Hermia's fitted parameters (k-values), namely, those membranes with higher k-values are more likely to become fouled (Heidari et al. 2019). In Table 4, a comparison of Hermia's fitted parameters reveals that PES-1 (pure membrane) has the lowest value. This finding indicates that the membrane's propensity for fouling is reduced when nanoparticles have been incorporated into it.

The incorporation of nanomaterial particles into a membrane substrate can produce an improvement in membrane performance. GO, aluminum oxide (Al₂O₃) tin oxide (SnO₂), and titanium oxide (TiO₂) were used in this study and incorporated into PESU. The water flux of the modified membranes improved when compared with the membrane without a nanomaterial. Increases were seen in PES-2 (Al₂O₃), PES-3 (TiO₂), PES-4 (SnO₂), and PES-5 (GO) from 65 L/m² h bar for the pristine membrane (PES-1) to 143.6, 83.68, 92.12, and 75.43 L/m² h bar, respectively. The integration of nanoparticles was found to have a major and direct impact on the hydrophilicity of the membrane's surface.

The fouling parameters, including ⁠, ⁠, ⁠, and were applied to determine the fouling tendency of the membranes. The results showed that membrane propensity for fouling is reduced when nanoparticles have been incorporated into it.

The results show that the cake layer and standard and intermediate blocking mechanism models best explain the experimental results.

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

The authors declare there is no conflict.