Introduction

Undoubtedly, water is the most vital and precious element on Earth. It is nature’s gift and without it, almost no human survival activity would be achievable. Thus, battles and conflicts over accessing fresh, clean, and permanent water resources have been taking place since an immemorial time (Salameh et al. 2021). Climate change, growth of population, and contamination by industrialization made the water problem even worse. Consequently, countries have searched for alternative solutions to overcome the water shortage crisis. Desalination has played a key effective role in diminishing the gap between the desired amount of fresh water and the available resources. In 2020, It was reported that around 97.2 million m3/day of fresh water is supplied by desalination plants. Amongst the available desalination techniques, reverse osmosis (RO) is dominating the world market, where it accounts for 85% of the already existing plants (Zhao et al. 2021). RO is a membrane-based technology that is characterized over thermal desalination by its flexible processing, cost efficiency, and low energy requirement. Various parameters affect the performance of RO units including the selection of the right membrane material and its synthesizing method, the type of membrane module, and operating parameters (El-Gendi et al. 2017).

For many years, cellulose acetate (CA) asymmetric and polyamide thin film composite (TFC) have been the two dominant types of RO membranes. Even though polyamide-TFC membranes showed many pros over CA membranes, namely, high water permeation flux and better salt rejection, the desalination market still relies on CA membranes for their stable performance, durability, relatively low cost, simple processing, and tolerance to free chlorine attack. However, producing CA membranes with enhanced mechanical and thermal stability, salt rejection as well as water permeation flux is a priority to compete and cope with the development of TFC membranes and this may eventually affect the RO market in the near future (Ghaseminezhad et al. 2019; Liu et al. 2019b; Abdelhamid and Khalil 2019; Ali et al. 2021). Accordingly, membranologists have investigated two membrane modification approaches to overcome the disadvantages of polymeric membranes. The bulk membrane modification, such as the usage of multiple polymeric blends and impregnation of organic or inorganic particles. The other approach is by enhancing the membrane surface, for instance, grafting hydrophilic monomers to the membrane surface or introducing polar groups to alter the membrane chemical structure (El-Ghaffar et al. 2020; Elessawy et al. 2020; Elkady et al. 2020; Elkony et al. 2020; Batool et al. 2021).

The most common approaches for water treatment are blending two polymers and incorporating solid materials within the polymer matrix, or as called mixed matrix membranes (MMMs). Those approaches aim to obtain a new membrane with hybrid characteristics between those of its blended and/or incorporated components (Sivakumar et al. 2006; El-Aassar et al. 2016; Siddique et al. 2021). According to literature, the synthesis of blended polymers, with a selectivity better than the base polymer, is driven by the need to superimpose certain characteristics over those of the homopolymer. Accordingly, the properties, as well as the balance of the hydrophilic- hydrophobic membrane system, can be controlled when blending multi polymeric components(Nagendran et al. 2008; Arthanareeswaran and Kumar 2010; Moradihamedani and Abdullah 2017). Polyvinylidene fluoride (PVDF) is a polymeric membrane material that is strongly fabricated by phase inversion method and used in water treatment purposes for its chemical, thermal and incredible mechanical stability (Haponska et al. 2017; Lu et al. 2021; Shah et al. 2021). As previously stated in the literature, PVDF/CA blend membrane was prepared via phase inversion technique only in a few studies, where PVDF was the dominant polymer and CA was blended with minor concentration. The aim of CA addition was either for its hydrophilic properties to be applied at different applications rather than RO, such as sewage filtration (Hossein Razzaghi et al. 2014), juice clarification (Fitri and Widiastuti 2017), membrane bioreactor (Razzaghi et al. 2018) and ultrafiltration (Reza et al. 2022), or for its good conductivity to be used in lithium batteries (Cui et al. 2017). However, the blending of CA as a major matrix with PVDF to fabricate a completely hydrophilic membrane with suitable mechanical properties for RO application is considered the first investigation. On the other hand, metal–organic frameworks (MOFs) have attracted the attention in various applications for their marvelous properties. Amongst the prepared MOFs, UiO-66 as a hydrophilic filler was studied by many researchers for water desalination. It is known for its huge surface area and stability in different and harsh conditions (Kadhom et al. 2017).

As far as the authors know, there is no other research that deals with the fabrication of CA/PVDF blended polymeric membrane that is hybridized with UiO-66 nano MOF for RO application. Accordingly, this investigation seeks to boost CA performance by blending it with polymeric PVDF to obtain CA/PVDF blended RO membrane via phase inversion. In addition, the effect of impregnating UiO-66 nanoparticles on enhancing the RO performance of the new blended CA/PVDF polymeric membrane was investigated. The characteristics of the resultant hybrid polymeric membranes were tested by FTIR, XRD, TGA, SEM, and contact angle. Moreover, their RO performance including permeation and salt rejection will be discussed.

Experimental work

Materials

All chemicals, except sodium chloride (NaCl), were of analytical grade and used with no further purification or modification. Cellulose acetate (CA) with 100,000 g/mol average molecular weight and 39.8% acetyl content was supplied by Acros Organics, Polyvinylidene fluoride (PVDF) was obtained from Alfa Aesar. 1,4-dioxane (≥ 99.8%) and acetone (≥ 99.8%) were purchased from Fisher, methanol (CH3OH ≥ 99.8%) was delivered by Sigma Aldrich and acetic acid (glacial) was provided by Merck.

Fabrication of blank CA, blended CA/PVDF and CA/PVDF/UiO-66 hybrid membranes

Phase inversion method was used to obtain all flat sheet membranes under investigation. CA/PVDF blended membranes (CAPV-) were fabricated by preparing two separately homogeneous solutions of CA and PVDF then blending them overnight at 80 °C to avoid the formation of cloudy gel-like viscous solution(Li et al. 2013). Acetone was used as a common solvent for both PVDF and CA polymers. Moreover, dioxane was added as a solvent for the base CA polymer. The addition of acetic acid aimed to enhance the polymer mobility and structure. Methanol behaved as a non-solvent to stimulate the membrane water absorption and accordingly increase the membrane flow performance(Duarte et al. 2006; Abdellah Ali et al. 2020).For CA/PVDF blended polymeric membrane preparation, various amounts of CA were dissolved in a mixture of (dioxane/ acetone/ acetic acid/ methanol) with a ratio of (5.4/2/1/1.6), respectively. While certain amounts of PVDF were dissolved in a mixture of acetone and dioxane (2:1), sequentially. Eventually, the total polymer concentration in the homogeneous blended solution was kept at 14 wt% with the composition as indicated in Table 1. The increase of PVDF composition in the doping polymeric solution above 10 wt% of the polymer concentration resulted in a viscous solution that is hard to be casted. For this reason, the maximum blend polymer composition was chosen to be 9:1 of CA to PVDF, respectively. Later, 0.2 wt%, in respect to the weight of total blended polymers, of previously prepared UiO-66 nanoparticles(Koriem et al. 2021) were dispersed in the optimum mixed solvents for 2 h, then the optimum amounts of CA powder and PVDF solution were added to obtain CA/PVDF/UiO-66 (CPU) mixed matrix doping solution. The nanofiller was added to study its effect on the performance of the blended CA/PVDF membrane. The investigated amount of UiO-66 was selected as close to the best performance results obtained from literature (He et al. 2017; Kadhom et al. 2017).

Table 1 Composition and preparation conditions of the obtained membranes
Full size table

In order to obtain a membrane free from deficiencies caused by air bubbles, the prepared solutions were left before casting for 24 h. Then, blended polymeric solutions were casted on a glass plate at a thickness of 250 µm using an automatic film applicator with a speed of 50 mm/sec. The resultant casted films were left in the air to evaporate before immersing in a distilled water bath, where the solvent/nonsolvent exchange took place. The resulted membranes were annealed in a hot water bath. Finally, the annealed membranes were kept for 24 h in distilled water before conducting the RO performance tests.

Physicochemical characterization of blank CA, blended CA/PVDF and CA/PVDF/UiO-66 hybrid membranes

The obtained pristine CA, blended CA/PVDF and CA/PVDF/UiO-66 mixed matrix membranes were dried prior to characterization in a vacuum oven at 60 °C. Images of platinum coated top, bottom, as well as cross section of the membranes were taken using a JEOL JSM-6010 LV- SEMdevice to investigate their morphological characteristics. Bruker Vertex 70- FTIR equipmentwas used to identify and confirm the chemical structure of the fabricated membranes in a wave number range of 400–4000 cm−1. XRD was used to understand the crystalline nature of the fabricated membranes using a Shimadzu XRD-6100 device. TGA was used to understand the thermal stability of the prepared CA based membranes. In (TGA Q50) device, the temperature raised from 0° to 700 °C with a 10 °C/min raising step. The membrane hydrophilicity/hydrophobicity was measured using a contact angle measuring system (DSA 100, KRÜSS). Five random locations, on the surface of each sample, were selected to measure the contact angle to minimize the possible experimental measurement errors. The tensile strength of the pristine, blended and mixed matrix CA-based membranes was measured using a tensile testing device (4468, Instron). The membranes were cut into 10 × 50 mm and tested with an elongation rate of 3 mm/min.

Blank CA, blended CA/PVDF and CA/PVDF/UiO-66 hybrid membranesperformance Evaluation

  • In this investigation, the performance of the various prepared membranes was monitored via a CF042 crossflow membrane testing system (Sterlitech, USA). A schematic diagram of RO testing process can be seen in Fig. 1. Membranes were divided into rectangular shapes of 42 cm2 similarly to the cell active area. A 5000 ppm NaCl solution was prepared and fed to the system from a tank connected to a Hydra-cell high pressure pump at room temperature (25 ± 2 °C) by. The pressure was raised gradually until permeate water was observed. A sample from the permeate as well as the feedwater was collected, and the concentration of solute was measured by a TDS-meter (Adwa, AD32).

    Fig. 1
    figure 1

    Schematic diagram of RO membrane testing system

    Full size image

    Performance evaluation was made based on the permeate water flux and membrane salt rejection calculations as shown in Eqs. (1 and 2)(Ahmad et al. 2015; Ounifi et al. 2021).

    $$J = \frac{Q}{A*t}$$
    (1)

    where J is the permeate water flux (L/m2h), Q is the volume of permeate water (L), A is the active are of each membrane (m2), and t is the time (h).

    $$R\left( \% \right) = 1 - \frac{{C_{p} }}{{C_{i} }}$$
    (2)

    where R is the membrane salt rejection, Cp and Ci are the permeate and initial feedwater concentration, respectively.

Results and discussion

Blank CA, blended CA/PVDF and CA/PVDF/UiO-66 hybrid membranes Characteristics

  • FTIR analysis was used to identify the characteristic functional groups of the obtained CA membrane as well as, membranes fabricated by blending CA with PVDF with 5, 7 and 10 wt.%. As it can be seen in Fig. 2, the FTIR spectra of pure CA membrane indicated the carboxylic C–O, C–H bond of CH3, C=O, C–H stretching of CH2 and OH functional groups at 1036, 1370, 1731, 2949 and 3395 cm−1, respectively. Close results were previously reported (Liu and Bai 2005; Kamal et al. 2014; Aldalbahi et al. 2020). IR spectra of blended CAPV-5, CAPV-7 and CAPV-10 also showed bands corresponding to the main functional groups of CA. However, a slight shift in band 1036 cm−1 and the appearance of new peak at 874 cm−1 was observed after the blending of PVDF. This change could be attributed to the PVDF characterized C–C band stretching in β phase and amorphous phase band, respectively(Kang et al. 2016).

    Fig. 2
    figure 2

    FTIR spectra of blank PVDF, CA, blended CA/PVDF and hybrid CA/PVDF/UiO-66 membranes

    Full size image

    The previous observation gave an indication that there was no molecular interaction between PVDF and CA in the blended membranes. Further, after the impregnation of UiO-66 nanoparticles with CA/PVDF blended polymers, a more intense peak at 554 cm−1 was formed which could be assigned to the Zr-(OC) stretch (Zhang et al. 2019). These results confirm the interaction between UiO-66 MOF molecules and the hybrid CA/PVDF blending polymer. Additionally, as illustrated and marked in Fig. 2, the peak at 3382 cm−1, which was previously explained as the presence of OH group at CA polymeric matrix, has shifted to 3474 cm−1and its intensity decreased significantly after the addition of UiO-66 nanoparticles. This could refer to the interaction between the characterizing carboxylic group of UiO-66 and the present OH group of CA/PVDF blending via hydrogen bond (Sharma et al. 2018).In return, this could give an indication of the physical interaction between UiO-66 nanoparticles and CA/PVDF blended polymers. In conclusion, the previous observations indicated the presence of UiO-66 NPs in the prepared CPU membrane.

    XRD crystographic identification was chosen to investigate the effect of PVDF addition on the crystalline nature of pure CA membrane. Further, the presence of UiO-66 nanoparticles was proven by XRD patterns. The amorphous nature of pure CA membrane was clearly illustrated in Fig. 3, its characteristic peaks are relatively wide peaks at 2θ in the regions around 10° and 22° (Xu et al. 2016; Silva et al. 2017). In addition, a strong peak appeared at 21° after the PVDF addition and this could be attributed to the PVDF characteristic β phase(Ahmed et al. 2021). Furthermore, the successful impregnation of UiO-66 nanoparticles into CA/PVDF polymeric blended membrane was proved by the appearance of sharp characteristic peaks of UiO-66 in CPU membrane at 2θ = 7.5° and 8.6° (Schelling et al. 2018).

    Fig. 3
    figure 3

    XRD patterns of UiO-66 nanoparticles, blank CA, blended CA/PVDF and hybrid CA/PVDF/UiO-66 membranes

    Full size image

    SEM micrographs were used to investigate the morphological effect of polymeric PVDF blending with neat CA and UiO-66 nanoparticles impregnation on the blank CA membrane. Generally, the phase inversion fabricated anisotropic (asymmetric) CA membranes consist of a thin layer at the top that formed during the solvent evaporation step. This layer is followed by a thick, porous layer formed during the solvent/nonsolvent exchange. The thin layer is responsible for controlling both permeate water flux and salt rejection process, while the porous sublayer layer provided the required mechanical stability. (Strathmann et al. 1971; Suzana Pereira Nunes and Klaus-Viktor Peinemann 2006; Hołda and Vankelecom 2015; Ning 2015; Rana et al. 2015; Werber et al. 2017).

    It is illustrated from the cross-section images in Figs. 4c, 5c, and 6c) that all the fabricated CA-based membranes are composed of a dense top layer supported on a sponge-like porous layer, which agrees with the expected design investigated in the literature.

    Fig. 4
    figure 4

    SEM Images of neat CA membrane a top b bottom c cross section and d sponge layer

    Full size image
    Fig. 5
    figure 5

    SEM Images of a top b bottom and c cross section d sponge layer of the synthesized CAPV-5, CAPV-7 and CAPV-10 membranes

    Full size image
    Fig. 6
    figure 6

    SEM Images of a top b bottom and c cross section d sponge layer of CA/PVDF/UiO-66 membrane

    Full size image

Figure 4 displays top, bottom, and cross-section surfaces of pristine CA. As shown in Fig. 4c, the surface of the pristine CA membrane has a dense, groovy, and thin top selective layer, that is supported on a thick, porous sublayer. The sublayer is characterized by a tear or finger like structure supported on a sponge like pore structure(Elkony et al. 2020; Ali et al. 2021).As shown in Fig. 4b, the bottom surface of the pristine membrane contains pinholes as a result to the solvent-nonsolvent exchange across the interface(Morsy et al. 2016).

The change in the morphological structure of the pristine CA membrane with the addition of different amounts of PVDF is demonstrated in Fig. 5. Interestingly as illustrated in Fig. 5a, with increasing the PVDF concentration, the corrugated shape of the pristine CA membrane became smoother and the shape of the ridges and grooves on the surface decreased. Nevertheless, large agglomerations are detected on the surface of CAPV-10 membrane, which might cause defects in the membrane performance. Additionally, comparing to neat CA membrane, it was noticed from Fig. 5b that the number of pores was increased in the bottom surface. Moreover, in comparison to CAPV-0, it was found that the addition of PVDF resulted in a small increase in the selective dense top layer. This might explain the membrane’s enhanced salt rejection and the membrane’s decreased permeate water flux. However, as displayed in Fig. 5c, the thickness of the supportive sponge layer increase with increasing the amount of PVDF added. This might be explained by the delay of the demixing process during the membrane solidification in the coagulation bath (Ghaffarian et al. 2013; Struzyńska-Piron et al. 2014). Despite that, the blending of PVDF with CA does not have a noticeable change on the total membrane thickness, where all fabricated membranes thickness were found to be 85 ± 15 μm.

On the other side, the hydrophilic UiO-66 impregnation influenced the pore structure of the membrane. The presence of hydrophilic UiO-66 NPs in the dope solution increased the exchange rate of solvent and non-solvent. This led to a more uniform and increased number of a longer finger-like on a spongeous pore structure as well as a porous structure of the bottom surface, this can be seen in Fig. 6c and Fig. 6b, respectively. In addition, a thicker top selective layer than CAPV-7 was observed in Fig. 6c. The selective layer is responsible for the rejection and causes a resistance to the permeate water flux. Nevertheless, due to the high hydrophilic nature of UiO-66, the permeate water flux was enhanced. Also, white particles can be seen in Fig. 6a on the top surface of the membrane, and this is attributed to the successful distribution of UiO-66 NPs in the dope solution. The incorporated UiO-66 nanoparticles enhanced the permeability of water by providing an alternative path for the water molecules (Kim et al. 2019).

TGA was selected to express the membrane thermal decomposition by measuring the membrane weight change as a function of rising temperature. It is clearly indicated from Fig. 7 that each TGA curve has 3 degradation steps. At the beginning, around 4% of the blank CA sample weight was lost because of evaporating any residual amounts of absorbed water and/or volatilizing existing volatile components by increasing the temperature from 30 to 230° C. Secondly, almost 84% of the original blank CA weight was decomposed due to the decomposition of cellulose acetate when escalating the temperature up to 400 °C. Similar results were previously reported in literature (De Campos et al. 2013). On the other hand, a slight enhancement in the thermal stability of pristine CA membrane was noticed with the addition of PVDF and UiO-66. It is clearly seen that, at 400 °C only 76.2 and 78% of CAPV-10 and CAPV-7 sample, respectively, was decomposed comparing to 84% of the CAPV-0. Moreover, the addition of UiO-66 nano-MOF to CA/PVDF blended polymeric membrane has further improvement of the membrane thermal stability. It was found that CPU membrane started to decompose at around 332 °C rather than 290 and 296 °C for CAPV-0 and CAPV-7, consecutively.

Fig. 7
figure 7

TGA analysis of blank CA, blended CA/PVDF and hybrid CA/PVDF/UiO-66 membranes

Full size image

Contact angle was measured to evaluate the hydrophilicity of the various fabricated membranes. It gives an indication of the interaction between water and membrane surface. In general, when the measured contact angle (θ) is less than 90°, the membrane is considered hydrophilic. Consequently, the membrane is hydrophobic when its contact angle value is greater than 90°. It is also known that the value of θ has a reflection on the membrane surface roughness (Ba et al. 2019; Wu et al. 2020). As shown in Fig. 8, increasing the added amount of PVDF increased the value of contact angle from 43° to 63° (i.e., the membrane became more hydrophobic) in comparison to the pristine CA membrane. The findings showed that θ has increased from 43° for CAPV-0 to become 51° for CAPV-5. In addition, both CAPV-7 and CAPV-10 membranes showed the same trend, where θ value became 59° and 63°, respectively. However, the value of θ for all membranes was found to be below 90° (i.e., hydrophilic surfaces were still obtained after the PVDF addition). On the contrary, the impregnation of UiO-66 nanoparticles to CA/PVDF optimum membrane has decreased the value of θ to 49°.

Fig. 8
figure 8

Contact angle of the fabricated membranes

Full size image

Mechanical strength was tested to indicate the effect of PVDF and nanofiller addition into CA pristine membrane. The blending of PVDF with CA had an observable effect on the blank CA membrane mechanical stability. Figure 9 elaborates the improvement of mechanical strength with the increase of PVDF amounts. The mechanical stability of pure CA casted membrane is relatively low as it is considered a fragile polymer with a tensile strength equal to 2 MPa (Minhas et al. 2015; Asiri et al. 2022) compared with hard PVDF polymer that has 8.7 MPa tensile strength. Although, PVDF was added in small amounts with a maximum of 10 wt.%, it enhanced the tensile strength of CA up to 4.3 MPa.This could be explained by the PVDF entrapment in CA polymeric matrix as it has higher mechanical strength compared to CA. Moreover, the reduction of macro voids with the addition of PVDF (Hossein Razzaghi et al. 2014) may improve the mechanical properties of blended membrane. This also confirms the successful blending of both CA and PVDF. On the other hand, the incorporation of UiO-66 NPs at 0.2 wt% with CA/PVDF (CAPV-7) blended membrane slightly decreased the tensile strength of the membrane from 3.8 MPa without UiO-66 to 3.6 MPa. This behavior could be returned to the presence of UiO-66 decreased the thickness of the sponge supportive layer. In addition, the incorporation of UiO-66 NPs at this selected amount might have caused a nanoparticles aggregation that weakened the mechanical properties of the hybrid membrane.

Fig. 9
figure 9

Tensile strength of the fabricated membranes

Full size image

Blank CA, blended CA/PVDF and CA/PVDF/UiO-66 hybrid membranes performance evaluation

RO Performance of pristine CA, blended CA/PVDF and hybrid CA/PVDF/UiO-66 membranes was evaluated by calculating the membrane salt rejection and permeate water flux under a certain applied pressure. Figure 10 illustrates the comparison between permeate water flux and salt rejection for each membrane against the blank CA membrane at 8 bar. As it can be seen, blending PVDF with CA pristine membrane declined the permeate water flux along with a noticeable increase in the performance of salt rejection. Cellulose acetate (CA) is characterized by its hydrophilic nature while Polyvinylidene fluoride (PVDF) is known for its hydrophobic nature. Thus, the blending of two polymers contributed to reduce the membrane permeability as a result of decreasing the prepared membrane hydrophilicity. However, it was reported elsewhere that increasing the membrane surface hydrophobicity had a positive influence on enhancing the membrane salt rejection (Mansouri et al. 2018; Pang and Zhang 2018; Abdelhamid and Khalil 2019). Increasing the concentration of PVDF up to 7 wt% in the CA matrix resulted to a 40% enhancement of salt rejection and a 50% decline in permeate water flux. This might be explained as previously indicated at SEM images, that the addition of PVDF reduced the macrovoids in pristine CA membrane and escalated its small pores. However, increasing the concentration of PVDF above 7% led to a 7.5% decrease in salt rejection. This could be related to, above 7% PVDF a more viscous solution was obtained and in return a non-homogeneous membrane was obtained, and this had an effect on the membrane performance.

Fig. 10
figure 10

A comparison between salt rejection and permeate water flux for the fabricated blank CA, blended CA/PVDF and hybrid CA/PVDF/UiO-66 membranes

Full size image

As a result, 7 wt% of PVDF was found to be the optimum weight and thus 0.2 wt%, relative to the total polymer weight, from UiO-66 nanoparticles were added to this membrane for further enhancement of its RO performance. It is clearly seen that the impregnation of UiO-66 nanoparticles increased the salt rejection and permeate water flux over CAPV-7 membrane by 12.5% and 42%, respectively. This can be explained as the incorporation of zirconium based UiO-66 nano filler increasing the membrane surface roughness and hydrophilicity (Trinh et al. 2021). Also, UiO-66 was previously found to have a small aperture window size (Cavka et al. 2008; Liu et al. 2019a), this size is greater than water molecules, which explains the increased water selectivity and as a result water permeability. Further, the increase in salt rejection percentage might be attributed to the smaller aperture size that UiO-66 has compared to Na+ and Cl− ions (Kadhom et al. 2017; Ma et al. 2017).

Conclusions

The aim of the presented work was to develop and obtain a new hybridized casted RO membrane. The aim was successfully acquired by the polymeric blending of CA with PVDF and the impregnation of MOF nanoparticles of previously prepared UiO-66. FTIR and XRD patterns proved the successful preparation and impregnation of the blended CA/PVDF polymeric membrane and mixed matrix membrane of CA/PVDF/UiO-66. The optimum mixing ratio of the two polymers was selected depending on the membrane with highest salt rejection percentage. Thus, CAPV-7 was found to be the optimum membrane with a salt rejection of 80.24% and a permeate water flux of 4 L/m2h. Then, 0.2 wt% nano MOF of the total polymer weight was impregnated to the CAPV-7 membrane to further improve its salt rejection and permeate water flux. Higher salt rejection and permeate water flux of 90.4% and 5.7 L/m2h, respectively, was established by the novel hybridized CPU membrane.