The major scope of this study is the fabrication and development of a substrate and polyamide rejection layer for an efficient thin-film hydrophilic composite forward osmosis (TFC-FO) membrane. Fabrication of a thin-film nanocomposite forward osmosis membrane employing interfacial polymerization and modification of substrate characteristics using titanium dioxide (TiO2) nanoparticles as additives (TFNC-FO) are studied. Characterizations of the prepared TFC-FO and TFNC-FO membranes were determined. The morphologies of cross-section, upper and bottom surfaces for the TFC-FO and TFNC-FO membranes were studied using scanning electron microscopy (SEM). Energy-dispersive X-ray (EDX) spectroscopy was used to examine the compositions of different elements for both membranes. The hydrophilicity of the prepared TFC-FO and TFNC-FO membranes was investigated using the measurement of the contact angle test. A Fourier Transform Infrared (FT-IR) spectrophotometer was used to observe the existing functional groups of the TFC-FO and TFNC-FO membranes. The thermal stability of the membrane was evaluated via thermogravimetric analysis (TGA). The overall performance of TFC-FO membranes was evaluated with and without adding TiO2 nanoparticles through different parameters, such as membrane flux, initial feed concentration, draw solution concentrations, reverse solute fluxes, membrane permeabilities, and finally, the effect of FO membrane orientations. FO membrane performance was successfully enhanced by adding different concentrations of TiO2 nanoparticles from 0.5 to 1.5 wt%. The findings indicated that an increase in the concentration from 0.5 to 1 wt% leads to a clear increase in both the porosity and hydrophilicity of the nanocomposite substrate and consequently, an increase in the water flux. However, further increasing the concentration of TiO2 nanoparticles to more than 1 wt% affects the membrane performance.

  • Preparation of a TFNC-FO membrane.

  • Incorporating TiO2 nanoparticles into polysulfone membrane matrix.

  • Evaluation of FO performance.

  • FO membrane permeability.

  • Characterization of the synthetic FO membranes.

Graphical Abstract

Graphical Abstract
Graphical Abstract
     
  • Am

    effective membrane area (m2)

  •  
  • Ap

    permeability of pure water (l/m3 · h. bar)

  •  
  • Bs

    solute permeability (g/m3 h1. bar)

  •  
  • Ct

    salt concentration of the water feed (mg/l)

  •  
  • i

    Van't Hoff Constant (dimensionless)

  •  
  • Ε

    membrane porosity (%)

  •  
  • Ρ

    solution density (kg/m3)

  •  
  • π

    osmotic pressure (atm)

  •  
  • Js

    reverse solute flux (g/m2 h.)

  •  
  • Jv

    pure water flux based on FO (l/m2 · h1)

  •  
  • V

    permeate volume (m3)

  •  
  • L

    membrane thickness (m)

  •  
  • Ww

    weight of wet membrane (g)

  •  
  • Wd

    weight of dry membrane (g)

  •  
  • M

    molarity of the aqueous solution (mole/l)

  •  
  • AL

    active layer

  •  
  • CVM

    compact video microscopy

  •  
  • EDX

    energy-dispersive X-ray

  •  
  • ICP

    internal concentration polarization

  •  
  • FO

    forward osmosis

  •  
  • IP

    interfacial polymerization

  •  
  • DS

    draw solution

  •  
  • FS

    feed solution

  •  
  • PA

    polyamide

  •  
  • PSF

    polysulfone

  •  
  • PVP

    polyvinylpyrrolidone

  •  
  • RO

    reverse osmosis

  •  
  • FO

    forward osmosis

  •  
  • UF

    ultrafiltration

  •  
  • MF

    microfiltration

The desalination process is critical in addressing worldwide water scarcity issues (Charcosset 2009; Razmjou et al. 2013; Ahmed et al. 2021). While current desalination technologies like thermal distillation and membrane technologies have greatly contributed to the success of sea/brackish water desalination, their high energy consumption remains a source of concern for many, especially when the majority of the energy consumed comes from non-renewable sources (Charcosset 2009).

Forward osmosis (FO) desalination utilizing membrane technology has piqued the interest of membrane scientists and manufacturers in recent years as a possible desalination technique to incorporate with reverse osmosis (RO) in the future (Cath et al. 2006). The difference in osmotic pressure between the two solutions, feed and draw, generates a critical performance index of FO due to the flux flow, and so there is no need for an external driving force. Furthermore, when compared to the RO process, which is a pressure-driven process, FO provides great contamination rejection and lower membrane fouling (Cornelissen et al. 2008). The power consumption at FO is significantly reduced as a result of this working paradigm. Another advantage of using low (Emadzadeh et al. 2014) operating pressure in the FO process is that it lowers fouling, which reduces the cleaning procedure's periodic time and, as a result, enhances the membrane's life (Lee et al. 2010; Zhao et al. 2012; Emadzadeh et al. 2014).

Despite the significant advantages that the FO membrane gives in the desalination process, scientists are still working on ways to improve the characteristics of thin-film composite (TFC-FO) membranes, particularly in the upper active skin layer (Ma et al. 2012) The attention has shifted to a new idea known as ‘thin-film nanocomposite (TFNC),’ which displays a breakthrough in the way super-hydrophilic nanoparticles (NPs) are synthesized and integrated into the polyamide (PA) thin layer (Hoek et al. 2020). Several researchers used this new generation of thin films to insert hydrophilic nanomaterials into PA thin layers using NPs dispersed in iso-par-TMC monomer solution (Jeong et al. 2007; Lind et al. 2009, 2010; Fathizadeh et al. 2011).

Inorganic NPs have been reported as potential fillers to improve the properties of microporous ultrafiltration (UF) membranes in terms of water permeability, fouling propensity, mechanical, and thermal properties in recent years, provided the number of NPs added was not in excess amount. The titanium dioxide (TiO2) nanoparticle is the most extensively employed nanomaterial in the creation of nanocomposite membranes among the NPs examined (Emadzadeh et al. 2014; Ismail et al. 2021).

Incorporating hydrophilic NPs, such as zeolite NPs, has been demonstrated in the previous study to be beneficial (Zhao et al. 2012). Also, silica dioxide (SiO2) can be used (Li et al. 2009; Yasukawa et al. 2015). Carbon nanotube (CNT) is considered one of the most important NPs (Chen et al. 2017). Halloysite nanotube (HNTs) (Tian et al. 2014) and graphene oxide (GO) (Baghbanzadeh et al. 2016; Sirinupong et al. 2018; Song et al. 2021) could enhance not just the hydrophilicity of the back support layer or substrate (Lai et al. 2016), but also its morphology, which is linked to the structural (S) parameter (Yasukawa et al. 2015). TFC-FO membranes with hydrophilic polymers in the support layers increased water flux while reducing internal concentration polarization (ICP) during the FO procedure (Xing et al. 2021; Han et al. 2012; Bui & McCutcheon 2013; Huang et al. 2013; Huang & McCutcheon 2014; Puguan et al. 2014; Tian et al. 2014; Ong et al. 2015; Yasukawa et al. 2015; Obaid et al. 2016; Chen et al. 2017). Incorporating Zeolite nanocomposite into a PSF-based membrane, for example, lowered the water contact angle and enhanced water flux by 55–64%, according to the dose added (Lind et al. 2009).

In all of these studies, nanoparticle agglomerations were inescapable, severely limiting the TFN membrane's ability to attain its full potential. Chemical modification of nanoparticle surfaces has lately attracted interest as a means of circumventing the drawbacks of nanoparticle aggregation. This research tries to achieve this by using a method based on the interaction (or lack there of) between NPs and the resulting PA matrix.

In mixed-matrix membranes, TiO2 is one of the most often used NPs (Han et al. 2012). Due to their tendency to breakdown natural organic materials, generate more hydrophilic structures, chemical stability, minimal toxicity, and commercial availability, they have recently been used in membrane surface and structure changes (Yang et al. 2007; Emadzadeh et al. 2014; Mollahosseini & Rahimpour 2014). The key properties of TiO2 NPs explored for nanocomposite membrane development are their exceptionally small particle size (less than 21 nm) and super-hydrophilic surface (Kim et al. 2003; Emadzadeh et al. 2014). The usage of TFC membrane as a FO membrane for separation is due to its good separation capabilities with a smaller skin layer covering and wide pH resistance. In recent studies, wastewater treatment has received a lot of attention (Jia et al. 2021).

TFN membranes were synthesized in this study by integrating PSF, PVP, and TiO2 with the solvent NMP. Interfacial polymerization (IP) was used to produce the membranes. These membranes had different structures, e.g., integral asymmetric membrane (Wei et al. 2021). The findings of this study are important because they show that, rather than changing the properties of the PA layer of TFC membranes, the substrate properties can also be changed to produce appropriate FO membranes. Membrane porosities were calculated (Kim et al. 2003).

The major goals of the current work are to fabricate a flat sheet TFNC-FO membrane and its performance which was enhanced by adding TiO2 NPs to the substrate membrane and to look at the impacts of TiO2 NPs loading on the properties of the PSF substrate, as well as how changes in substrate properties affect the TFC membrane performance in the FO process.

The research findings are significant in showing that modifying the properties of the PSF substrate rather than changing the characteristics of the PA layer can improve the FO membrane.

Materials

Transparent Polysulfone (PSF) (Pellets) was used as the main polymer which is provided by Sigma–Aldrich, with an average molecular weight of ∼35,000 with a linear formula of [C6H4–4–C(CH3)2C6H4–4–OC6H4–4–SO2C6H4–4–O]n. The solvent of PSF (Pellets) was acquired from Sigma–Aldrich Company; N-methyl pyrrolidone with the synonym N-methyl-2-pyrrolidone, 1-methyl-2-pyrrolidone; NMP, C5H9NO (more than 99.5%), with an average molecular weight of ∼99.13. The substrate membrane was made from polyvinylpyrrolidone (PVP, K30, Sigma–Aldrich). PVP is a pore-forming ingredient that can also enhance the viscosity of the casting solution. It has been claimed that adding PVP to a membrane structure can cause macro voids to form, resulting in increased water flux. 1,3-M-phenylenediamine with the chemical formula 1,3-(NH2) 2C6H4; MPD (more than 99%), n-hexane (more than 99%) and 1,3,5-benzenetricarbonyl trichloride, C6H3(COCl)3, TMC (more than 98%), with an average molecular weight of ∼265.48 were the monomers utilized for the preparation of PA selective layer for TFC membranes, and they were obtained from Sigma–Aldrich Company. Sigma–Aldrich synthesized titanium dioxide NPs (TiO2 NPs) with a particle size of less than 21 nm. TiO2 NPs were utilized as an additive to modify the characteristics of the substrate. Table 1 presents the physicochemical properties of TiO2 NPs. Pure sodium chloride, NaCl (more than 99.5%), was used for producing salt (draw) solution at different concentrations for FO tests.

Table 1

Physicochemical properties of TiO2 NPs

CharacterValue
Appearance (color) White  
Appearance (form) Powder  
Surface area 35–65 m2/g (BET) 
Particle size <21 nm (TEM) 
Density 4.26 g/ml 
CharacterValue
Appearance (color) White  
Appearance (form) Powder  
Surface area 35–65 m2/g (BET) 
Particle size <21 nm (TEM) 
Density 4.26 g/ml 
Table 2

Composition of monomer solution used in FO membrane preparation

Composition of aqueous solution
Composition of organic solution
FO membrane (TFC/TFNC)MPD (wt%)Water (v%)TMC (wt%)Hexane (v%)
 100 ml 0.1 100 ml 
Composition of aqueous solution
Composition of organic solution
FO membrane (TFC/TFNC)MPD (wt%)Water (v%)TMC (wt%)Hexane (v%)
 100 ml 0.1 100 ml 

Methods

The method of TFNC-FO membrane synthesis is described in the following.

Preparation of substrate (TiO2 NPs addition)

The main procedures for the preparation of the TFC-FO membrane are shown in Figure 1. The dope solution compositions used to form PSF substrate with and without NPs incorporation are presented in Figure 2. To prepare the dope solution, an appropriate amount of TiO2 NPs was first added to NMP and introduced to ultra-sonication for 60 min to minimize the TiO2 NPs agglomeration. PVP was then added to the mixture that was then introduced to a vigorous stirring by the hot plate magnetic stirrer for 15 min at 80 °C. It was later taken by including PSF into the mixture under vigorous blending for about 120–180 min until all PSF pellets were dissolved. The prepared homogeneous solution was left at room temperature for about 2 h to expel the air bubbles trapped within the solution. The dope solution was cast on a glass plate. The casting step was carried out by the automatic film applicator to form a wet membrane thickness of 200 μm. This step was immediately followed by immersion into a coagulation water bath at room temperature for phase inversion to take place. After membrane formation, the formed membranes were transferred to a second water bath (distilled water) and further soaked for at least 24 h to remove solvent/PVP residual. After that, the membranes were stored in deionized water until using it.

Figure 1

Preparation of the TFNC-FO membrane.

Figure 1

Preparation of the TFNC-FO membrane.

Close modal
Figure 2

Compositions of the substrate solution (control TFC) and (TFN: 0.5, 1, 1.5 wt% TiO2) used for the FO membrane preparation.

Figure 2

Compositions of the substrate solution (control TFC) and (TFN: 0.5, 1, 1.5 wt% TiO2) used for the FO membrane preparation.

Close modal

PA rejection layer formulation

A rejection (PA) layer sits on top of a very porous support layer in a conventional TFC membrane, the PSF substrate. This active rejection layer of the TFC membrane was formed by IP. IP is an irreversible process, a rapid reaction that has been used to create a thin rejection layer for TFC membranes that can be employed in FO preparation. It is carried out as follows: first, on the top surface of the PSF substrate, 50 ml of a 2% (w/v) aqueous MPD solution were decanted, which stayed horizontally for 2 min to ensure the breakthrough of MPD aqueous solution into the pores of the substrate (Table 2). The remaining MPD was removed using a smooth rubber roller after the excess MPD solution was drained. Then, 50 ml of the organic solution was poured onto the top surface of the substrate by the slope to remove any residue from the MPD, which was applied horizontally for 1 min to ensure the completion of the IP process. Then, the excess of the TMC solution was drained off and dried in the oven for 10 min at 60 °C. Finally, the TFC membrane was saved in deionized water until use. The polymerization that occurs at the interface depends on the reactivity and concentration of the reacting monomers, the number of reactive groups on each monomer, and the stability of the solvent interface. All these influence the properties of the produced polymer.

TFC–TFNC-FO membrane characterizations

Morphology of TFC–TFNC-FO membranes

Scanning electron microscopy (SEM) was used to observe the morphology of the cross-section, top and bottom surface of the TFC and TFNC. The membrane samples were covered with gold to produce electrical conductivity. The top surfaces of membranes were scanned with a JSM-IT200 scanning electron microscope (SEM) to show the pores of membranes.

Functional groups of the TFC-TFNC-FO membranes

Fourier Transform Infrared (FT-IR) analysis was performed on the prepared membrane. It was used to distinguish the functional groups of the membrane. In the spectral region, 3,500–500 cm−1 with a resolution of 5 cm−1, a smart type of beam splitter was equipped with the FT-IR spectrophotometer.

Thermal stability of TFC–TFNC-FO membranes

The knowledge of the thermal properties of materials is very important in determining their thermal stability. Detection of chemical and physical changes, characterization, and application possibilities TGA are thus used to test the polymer thermal stability at high temperatures. The weight of the sample changes with the temperature, and this is used to analyze the thermal stability. The TGA test is carried out with a Thermogravimetric Analyzer (TGA-50, SHIMADZU) from 0 to 600 °C at a rate of 5°/min (minute variations in mass can be observed).

TFC–TFNC-FO membrane elements

EDX (energy-dispersive X-ray) spectroscopy is an analytical technique for determining the compositions of various elements in a sample (Hilal et al. 2017).

Porosity measurements

Membrane porosity (ɛ) was determined by measuring the dry mass (mdry) and wet mass (mwet) of membrane samples according to the following equation (Li et al. 2009).

In this research, to measure the porosity, the weight of a wet sample of the membrane (Ww) with an area of 3 cm × 3 cm was measured first to determine the porosity. The water on the surface was then dried. After that, the sample membrane was dried completely in a desiccator for 24 h. Finally, the membrane's dry weight (Wd) was determined, and the membrane porosity, approximate size of pores, and the number of pores were calculated using the following equation.
formula
(1)
where A is the effective membrane area, L is the membrane thickness, ρ is the water density, Ww and Wd are the weights of wet and dry membranes, individually.

Membrane wettability measurements

The hydrophilic characteristic and moist ability of the membrane surface are determined by measuring the contact angle. A compact video microscope (CVM) was used to calculate the contact angle (Abdallah et al. 2018).

Filtration test of the prepared membrane (TFC-FO)–(TFNC-FO)

The experimental setup for this membrane performance test was carried out as follows: A laboratory-scale FO circulated system was applied to evaluate the overall efficiency of FO membranes (Graphical abstract). The cross-flow FO cell was provided with two pumps for both feed and draw solutions. Experiments were performed at 25 °C by keeping the pressure at 6 psi using the feed and drawing solution concentrate valves (Ezugbe et al. 2021). The TFC membranes were tested in both operating modes for asymmetric membrane (FO) regarding the membrane orientation. The active layer of a pressure-retarded osmosis (PRO) system faces the draw solution, whereas the active layer of an FO system faces the feed solution. To accurately determine the water flux, weight changes of the draw solution were calculated using a digital weight balance that was placed under the tank of the draw solution. In addition, the conductivity of the feed solution was estimated by a bench conductivity meter (AD 310, Adwa). Then, the conductivity was converted into a concentration by employing a calibration curve to determine the reverse salt flux. In the FO experiment, NaCl aqueous solution was used as a feed solution with different concentrations (10–120 mM) and suggested drawing solutions with different concentrations (0.5–2.5 M) NaCl.

Basic performance measurements (FO test)

The most important measurements to evaluate the membrane performance are:

  • [1] The FO membrane flux Jv (l/m2 · h)

  • [2] Reverse salt flux (Js, gm−2 h−1)

FO water flux and reverse solute flux were measured using Equations (2) and (3), and all measurements were done in triplicate with the average user. The FO membrane flux Jv (l/m2 h) was obtained by measuring the weight change in the draw solution tanks at the different periods (each experiment was performed for 30 min) according to the following equation (Emadzadeh et al. 2014; Song et al. 2021)
formula
(2)
where Δv is the change in the volumes of the feed solution (L), Am is the effective membrane area (m2), ρ is the density of the draw solution (kg/m3), Δm is the change in the weights of the draw solution (kg) and Δt is the measuring time interval (h). The following equation was used to calculate the reverse salt flux (Js, gm−2 h−1) (Song et al. 2021)
formula
(3)
where Ct (mg/l) and Vt (m3) are the salt concentration and the volume of the feed measured at the beginning and the end of the time interval, respectively, Am is the effective membrane area (m2) and Δt is the time interval of the experiment (h) (Song et al. 2021).

Calculations of the membrane permeability

The Ap and Bs constants are important parameters for the mathematical models of FO, where the essential selectivity for molecules of water can be evaluated by the permeability of the pure water (Ap). Subsequently, the pure water flux increases in proportion to an increase in Ap regarding the following equation (Kim et al. 2017).
formula
(4)
where Ap is the permeability of the pure water (l/m2 · h bar) and Δπ is the difference between the feed and draw solutions in terms of osmotic pressure. The osmotic pressure was calculated according to (Van't Hoff equation)
formula
(5)
where π is the osmotic pressure in (atm), i is the Van't Hoff constant which is the number of discrete ions in a formula unit that equals 2 for both the feed and draw solution, M is the molarity of the aqueous solution, R is the universal gas constant (0.08206 L atm/mole K), T is the temperature of both feed and draws solution (K).
Also, the solute diffusion through the FO membrane has occurred as a result of the difference in the concentration of solute between the two separated solutions. Therefore, the proportional constant of the solute transport (Equation (6)) can act as an indicator for the solute selectivity of an FO membrane:
formula
(6)
where Js is the solute flux, Bs is the solute permeability and ΔC is the feed solute concentration difference before and after an FO process (Su & Chung 2011).
Analytical and characterization methods for (TFC–TFNC) membranes

The characterization of the synthetic FO membranes TFC and TFNC were carried out using:

  • SEM,

  • FT-IR,

  • TGA,

  • Membrane porosity test,

  • Membrane contact angle.

Membrane support morphology (SEM)

In this work, an SEM was used to capture the top, bottom, and cross-sections of the membranes. Membranes were coated with gold for 2 min using a current (15 mA) sputter coater. Results of SEM for the series of nanocomposite PSF substrates' top and bottom surfaces as well as their cross-sectional form are presented in Figure 3(a). In a comparison of the top and bottom morphologies for the four types of PSF substrates (A–E), it was shown that the pores in the bottom surface are clearer than the top surface. This reflects that more nanomaterials were accumulated on the bottom part of the substrate during the phase inversion stage (Sirinupong et al. 2018).

Figure 3

SEM spectra for substrate surfaces (upper and the bottom) for TFC–TFNC-FO membranes: (a) substrate (control), (b) substrate TiO2 (0.5%), (c) substrate TiO2 (1.0%) and (d) substrate TiO2 (1.5%).

Figure 3

SEM spectra for substrate surfaces (upper and the bottom) for TFC–TFNC-FO membranes: (a) substrate (control), (b) substrate TiO2 (0.5%), (c) substrate TiO2 (1.0%) and (d) substrate TiO2 (1.5%).

Close modal

The addition of TiO2 NPs in the main structure of the membrane leads to an increase in membrane porosity as the number of pores increases in the TFNC as shown in the bottom surface of the membrane in Figure 3(d) compared to the bottom surface of the TFC membrane as shown in Figure 3(a). Also, Figure 3(e) indicates the image of the bottom surface of the TFNC membrane, where the NPs are agglomerated due to increasing TiO2 percentage to 1.5%. This suggests that adding more than 1.0% of TiO2 results in clear aggregation on the surface of the substrate, minimizing the contact area of hydroxyl groups occurred by TiO2 NPs and likely deteriorating the substrate structural integrity (Emadzadeh et al. 2014).

From the cross-section images, it can be observed that the sponge-like holes and voids extended from the top to the bottom of the substrate surface. Figure 4(b) and 4(d) illustrate long finger-like voids compared to the pure PSF substrate Figure 4(a), because of the inclusion of TiO2 NPs in the dope solution, water passage from the water coagulation bath to the cast polymer layer is greatly facilitated, resulting in the formation of long finger-like pores that improves overall porosity and facilitates water flux (Emadzadeh et al. 2014).

Figure 4

SEM cross-section images for TFC–TFNC-FO membranes: (a) substrate (control), (b) substrate TiO2 (0.5%), (c) substrate TiO2 (1.0%) and (d) substrate TiO2 (1.5%).

Figure 4

SEM cross-section images for TFC–TFNC-FO membranes: (a) substrate (control), (b) substrate TiO2 (0.5%), (c) substrate TiO2 (1.0%) and (d) substrate TiO2 (1.5%).

Close modal

Membrane elemental analysis (EDX)

EDX element analysis of the upper and bottom surfaces of the substrate without adding TiO2 NPs and in the presence of different doses of the NPs is shown in Table 3. Due to the chemical structure of TiO2, which contains (O) elements in the organic structure, the surface of the substrate TiO2 displays substantially more oxygen (O) elements than the surface of the substrate control. The substrate TiO2 investigates the increase in (Ti) element on the bottom surface as a result of more nanomaterials accumulating on the bottom surface of the substrate during the phase inversion process, resulting in a decrease in internal concentration polarization of the TFC membrane during the FO process.

Table 3

Comparison between the elements of control and TiO2 substrates (EDX analysis)

PSF substrateElement of the upper surface (wt%)
Element of the bottom surface (wt%)
CO2STiO2CO2STiO2
Substratecontrol element 76.56 20.63 2.81 – 75.26 22.63 2.11 – 
Substrate TiO2 added 72.85 23.47 2.75 0.93 71.59 23.38 3.69 1.34 
PSF substrateElement of the upper surface (wt%)
Element of the bottom surface (wt%)
CO2STiO2CO2STiO2
Substratecontrol element 76.56 20.63 2.81 – 75.26 22.63 2.11 – 
Substrate TiO2 added 72.85 23.47 2.75 0.93 71.59 23.38 3.69 1.34 

Membrane porosity (ε)%

Equation (1) was used to calculate the porosity of the substrate membrane (%). The overall porosity percentage of the prepared TFC/TFNC membranes is shown in Figure 5. The total porosity of the membrane rose from 65% in the TFC membrane to 77% in the TFNC membrane, this means that the use of TiO2 NPs increased the membrane porosity that leads to a decrease in the porosity of the membrane's internal structures during the FO processes that increase the water permeability. The excessively porous support can also make constructing an excellent active selective layer with the required mass transport capabilities more challenging. The sponge-like structure with small pores surrounded by dense walls, on the other hand, may make it easier to construct an integrated thin active layer and may provide superior mechanical stability to the finger-like characteristic; however, it increases mass transfer resistance.

Figure 5

Porosity (ε) percent of TFC–TFNC-FO membranes.

Figure 5

Porosity (ε) percent of TFC–TFNC-FO membranes.

Close modal

TFNC membrane surface wettability characterization

The addition of hydrophilic nanomaterials to the substrate layer of a TFC-FO membrane can result in increased porosity, greater hydrophilicity, and decreased tortuosity, all of which work together to reduce ICP. The water contact angle (WCA: θ) is the common measurement to estimate the membrane surface hydrophilicity, and the findings are shown in Figure 6. The hydrophilic membranes have a WCA of 0° < θ < 90°, while hydrophobic membranes have a WCA of 90° < θ < 180°. In the case of the prepared TFC/TFNC membranes, there was a decrease in contact angle with increasing the TiO2 percentage. The contact angle reduces from 66.4° to 52.2° as the TiO2 percentage increased from 0 to 1.5 wt% due to the increase of membrane hydrophilicity. Increasing hydrophilicity leads to improvement in membrane performance.

Figure 6

Contact angle (θ) of TFC–TFNC membranes.

Figure 6

Contact angle (θ) of TFC–TFNC membranes.

Close modal

FT-IR spectrophotometer

One of the most widely used analytical techniques for materials analysis is infrared spectroscopy. The main concept is summarized as follows: when infrared light passes through a sample, some of the light is absorbed by the sample's chemical bonds, creating molecular vibrations. The absorbed IR radiation's energies are determined by the bond strengths and atoms, and the resulting spectrum shows molecular absorbance. Figure 7 shows the FT-IR spectrum of the prepared flat sheets membranes for substrate (control) and TFC membrane. TFC spectra indicate that the absorption peak of wave number equals 1,154.03 cm−1 which shows the presence of symmetric O = S = O stretching and indicated in the 1,151–1,155 cm−1 range. The absorption peak at 1,294.71 cm−1 indicates the presence of asymmetric O = S = O stretching and is indicated in the 1,294–1,295 cm−1 range. The absorption peak at 1,240.96 cm−1 indicates the presence of asymmetric C − O − C stretching and is indicated in the 1,240–1,258 cm−1 range. The absorption peak at 1,490.47 cm−1 indicates the presence of CH3 − C − CH3 stretching and indicated in the 1,486–1,490 cm−1 range and 1,490 cm−1 (C = C aromatic ring stretching) correspond to the functional groups of substrates made of the PSF polymer. Regarding the PA layer, the characteristic peak at 1,652.00 cm−1 indicates the presence of C = O, where the carbonyl functional group spectra are in the range from 1,600 to 1,800 cm−1 wave number (C = O amide stretching) and 1,583.30 cm−1 (C − N amide stretching). So, it can be concluded that there are PVP, PSF, TMC, and MPD in the produced membrane and this is considered for both TFC and TFNC membranes. As can be seen, the peaks of virgin PSF and PSF substrate with TiO2 NPs have no obvious change. FT-IR spectroscopy is mostly used for a qualitative study of various materials, but it may also be used for quantitative analysis utilizing modern software that calculates the area of various peaks in the IR spectrum (Stuart 2004).

Figure 7

FT-IR of the prepared TFC-FO membrane (blank, TFN0.5, TFN1.0, and TFN1.5).

Figure 7

FT-IR of the prepared TFC-FO membrane (blank, TFN0.5, TFN1.0, and TFN1.5).

Close modal

Thermal gravimetric analysis

TFC-FO and TFNC-FO, as well as their components, have thermal stability. The weight change that occurs while a sample is heated at a constant pace was assessed using TGA, which was employed as an analytical technique to determine the fraction of volatile components. Figure 8 presents TGA curves for the four prepared membranes; the first one for the TFC-FO membrane, (b), (d), and (e) of the figure represent TFNC-FO membranes (0.5, 1, 1.5 TiO2). From the obtained results, all the prepared membrane samples have relatively good thermal stability. The weight losses were evaluated, and no significant values occurred until 100 °C. From 100 to 500 °C is the step of removing residual solvent from the membrane matrix. Solvent removal in the TFC-FO case is more than in the TFNC-FO cases. Figure 8 indicated that the weight loss (degradation) in the TFC occurred in two steps at midpoint 100 °C by 53.52% and at midpoint 584 °C by 24.52%. A total overall weight loss of 38%, in the case of TFNC-0.5% TiO2, at midpoint 100 °C, and 21.15%, at midpoint 521 °C. In general, similar results were obtained at the degradation step (100–500 °C) in TFNC 1, 1.5; there was no significant difference in TFNC with different NPs doses. It is observed that the weight loss of the TFN-0.5% is less than the weight loss in the TFC membrane, which improves the thermal stability in the nanocomposite membrane. This improvement may be attributed to the interaction of coordinate bonds, covalent bonds, Van Der Waals forces, or hydrogen bonds between TiO2 NPs and polymeric chains.

Figure 8

TGA of TFC-FO and TFNC-FO membranes.

Figure 8

TGA of TFC-FO and TFNC-FO membranes.

Close modal

Orientation of FO membrane effect on its performance

The difference in osmotic pressures between the draw and feed solutions is thought to provide the driving factor for the FO process, hence draw solutions are crucial. Because the FO membrane has an asymmetric structure, water flows have been determined at both membrane orientations using 10 mM NaCl as the feed solution and 1.5 M NaCl as the draw solution. To improve the efficiency of the FO process, the optimum draw solutions must have specific qualities. To begin, it must produce a larger osmotic pressure than the feed solution, which can be accomplished by raising the concentration of the feed solution. Second, there must be a low reverse solute flux as a result. Third, following the FO process, the diluted draw solution must be able to collect with minimal energy and expense. Finally, to decrease ICP, the orientation of the membrane has a critical factor. The two main membrane orientations are (1) active layer facing draw solution (AL-DS) and (2) active layer facing feed solution (AL-FS) (Ahmed et al. 2021). When the active layer of the membrane is facing the feed solution, it is in FO mode; when the active layer is facing the draw solution, it is in PRO mode. As presented in Figure 9, the water flux was 1.7–2.0 times higher in the PRO mode as compared to the FO mode due to internal concentration polarization being more severe in the FO mode. But placing the support layer against the feed solution would result in higher fouling and scaling (Razmjou et al. 2013; Yasukawa et al. 2015).

Figure 9

Water flux in FO and PRO modes (test conditions: 25 °C, 6 psi).

Figure 9

Water flux in FO and PRO modes (test conditions: 25 °C, 6 psi).

Close modal

The reverse solute flux of the membranes tested in PRO and FO modes is shown in Figure 10. When comparing the PRO and FO modes, it was discovered that the PRO mode had a greater reverse draw solute flux. Due to the strong reverse draw solute flux in the PRO mode, which minimizes the osmotic driving force, all experiments in this study were conducted in the FO mode. As a result, the membrane efficiency is reduced. Results agreed with Ismail et al. (2021); Ismail et al. found a significant increase in solute flux in TFNC in the AL-DS mode and in the AL-FW mode compared to control TFC membrane in the AL-DS mode and the AL-FW mode.

Figure 10

Reverse solute fluxes in FO and PRO modes (test conditions: 25 °C, 6 psi).

Figure 10

Reverse solute fluxes in FO and PRO modes (test conditions: 25 °C, 6 psi).

Close modal

Effect of TiO2 nanoparticle dose on the performance of TFNC membrane

The effects of TiO2 dose on membrane water flux and reverse solute flux were assessed using a cross-flow FO experimental laboratory setup. The studies were carried out with a feed solution of 10 mM NaCl and various draw solution concentrations. (0.5 M NaCl (29,500–24.74 bar), 1 M NaCl (59,000–49.48 bar), 1.5 M NaCl (88,500–74.24 bar), 2 M NaCl (118,000–98.98 bar), and 2.5 M NaCl (147,500–123.7 bar)). The results indicate that TFN membranes prepared from PSF–TiO2 nanocomposite substrates (0.5–1 wt%) provide much higher water flux regardless of changing the draw solution concentrations compared to the control TFC membrane. This significant improvement is due to the improved structural properties of PSF–TiO2 substrates which minimize the transport resistance against water permeation. Nevertheless, a further increase in TiO2 NPs loading to 1.5 wt% showed lower water flux due to agglomeration of TiO2 on the substrate surface. These results agreed with previous work (Emadzadeh et al. 2014), which proved that the TFN membranes prepared from PSF–TiO2 nanocomposite substrates showed much higher water flux for both orientations regardless of draw solution concentration as compared to the blank (control) TFC membrane. These findings are similar in the case of using zeolite NPs. The water flux of the fabricated FO membranes increased sequentially with the increase in the content of zeolite NPs and increase in the PSF sheet. (Ismail et al. 2021).

Figure 11(a) and 11(b) shows the effect of TiO2 doses on membrane water flux and reverse solute flux, the results indicating that the TFNC-0.5 (0.5 wt%TiO2) provides the higher water flux by 120% from the TFC (control) (from 12.2 to 24 l/m2 h). Likewise, it can be seen that there is an insignificant increase in solute flux in the TFN-0.5 membrane compared to the control TFC membrane (from 2.4 to 4.3 g/m2 h), using draw solution concentrations (1.5 M NaCl) vs. 600 mg/l feed water.

Figure 11

(a) Water flux of the prepared TFNC membrane from the PSF substrate made of different TiO2 doses (test conditions: 25 °C, 6 psi, draw solution 1.5 M NaCl at pH 6.8 (AL-FW)). (b) The reverse solute flux of the prepared TFNC membrane from the PSF substrate made of different TiO2 doses (test conditions: 25 °C, 6 psi, draw solution 1.5 M NaCl at pH 6.8 (AL-FW)).

Figure 11

(a) Water flux of the prepared TFNC membrane from the PSF substrate made of different TiO2 doses (test conditions: 25 °C, 6 psi, draw solution 1.5 M NaCl at pH 6.8 (AL-FW)). (b) The reverse solute flux of the prepared TFNC membrane from the PSF substrate made of different TiO2 doses (test conditions: 25 °C, 6 psi, draw solution 1.5 M NaCl at pH 6.8 (AL-FW)).

Close modal
Figure 12

Effects of draw solution concentration on FO flux (test condition: 25 °C, 6 psi).

Figure 12

Effects of draw solution concentration on FO flux (test condition: 25 °C, 6 psi).

Close modal

Increased TiO2 NPs loading to more than 0.5 wt%, on the other hand, had a negative influence on the separation performance, resulting in a lower degree of PA cross-linking, reduced membrane efficiency, and increased reverse solute flow due to increased membrane porosity (Emadzadeh et al. 2014). Previous work compared the separation properties of similar membranes which were made in this study to commercial cellulose triacetate (CTA) membranes. In terms of water permeability, they discovered that the PSF substrate with TiO2 added had a significant impact on TFC membrane performance. The water permeability of all TFN membranes made from a PSF–TiO2 nanocomposite substrate was significantly higher than that of commercial CTA membranes (Emadzadeh et al. 2014; Ismail et al. 2021).

The FO membrane flux was characterized by different draw solution concentrations. As can be seen from Figure 12, increasing the draw solution concentration from 0.5 to 2.5 M resulted in a flux increase from 7.14 to 17 l/m2 h of TFC (control) and from 16.2 to 31 l/m2 h of TFNC-0.5. As a result, as the concentration of the draw solution increased, the FO membrane flux increased. This was predicted because a higher concentration causes a stronger osmotic driving force in FO, which leads to a higher flow (Yasukawa et al. 2015).

Membrane permeability constants

Table 4 shows the results of both water permeability and solute permeability for the tested TFNC membrane prepared from the PSF substrate made of different TiO2 loadings. The water permeability of our synthesized TFNC-0.5 membrane is A = 0.355 l/m2 h bar; this value is 53.5% higher than the TFC control membranes. This result is due to the high porosity of TFNC-0.5 which is 70% as a result of adding TiO2. On the contrary, TFC (control) has the lower permeate flux due to its low porosity which is 65%.

Table 4

Water permeability and solute permeability results for 1.5 M NaCl draw solution at 10 mM NaCl feed solution

FO membraneWater flux, LMHSolute flux, gMHWater permeabilitySolute permeability
TFC-0.0 12.2 2.4 0.165 0.31 
TFNC-0.5 24.6 4.3 0.355 0.25 
TFNC-1.0 34.3 15.2 0.465 0.67 
TFNC-1.5 21.7 3.8 0.294 0.27 
FO membraneWater flux, LMHSolute flux, gMHWater permeabilitySolute permeability
TFC-0.0 12.2 2.4 0.165 0.31 
TFNC-0.5 24.6 4.3 0.355 0.25 
TFNC-1.0 34.3 15.2 0.465 0.67 
TFNC-1.5 21.7 3.8 0.294 0.27 

Also, Table 4 shows the effect of TiO2 loading on solute permeability, which indicates that the TFNC-0.5 provides a lower solute permeability of 0.25 l/m2 h compared to the TFC (control) which had a higher solute permeability of 0.31 l/m2 h.

In this current work, fabrication of a TFC-FO membrane was investigated and the effect on the performance of the membrane by adding TiO2 NPs into the PSF matrix was studied. Based on the findings, we can conclude that:

  • The support layer of the TFC-FO membrane has a significant impact on its performance. To achieve high salt retention, the active layer of an ideal FO membrane must be exceedingly thin and thick.

  • TiO2 NPs were incorporated into a PSF matrix, and both surface porosity and hydrophilicity of PSF-based substrates were enhanced. We can say that the modified surface hydrophilicity and porous structure formed are the most important factors leading to higher water flux compared with the TFC membrane/control.

  • The increase in the concentration from 0.5 to 1 wt% of TiO2 leads to an increase in both the porosity and hydrophilicity of the nanocomposite substrate and consequently, an increase in the water flux. To obtain high flux and minimum fouling, the hydrophilicity must be high. According to the current research, using innovative nanomaterials, substrates, and layer-by-layer assumptions in the fabrication of FO membranes significantly improve water flux.

  • TFNC-FO membrane has slightly reverse salt rejection compared with the TFC-FO membrane.

  • Under the same conditions, the fabricated TFNC-FO membrane indicated a much higher FO water flux showing a slight increase in the reverse solute flux.

  • Water flux TFNC-FO membrane under long filtration time was also recorded to be much higher, basically due to the enhancing properties of the substrate upon addition of nanoparticle of TiO2.

  • FO member water flux increases with the increase of draw solution concentration. As a result, a higher concentration produces a higher osmotic driving force.

  • The problem of internal concentration polarization is a problem because of its impact to decrease the effective osmotic pressure through the membrane's active layer.

  • The performance of an FO membrane is determined by a complex mix of external test conditions, including the osmotic strength of feed and draw solutions, the orientation of the FO membrane, designing an experiment (run-in times, measuring times, equilibration times, etc.).

A.H.K. conceptualized the study, refined the research idea, and edited and revised the study. H.Z.A. wrote the manuscript, drew the figures, wrote the results, and helped in discussion. S.N. wrote, edited, and revised the manuscript. M.G.E. wrote the manuscript, drew the figures, edited, and revised the manuscript.

The authors declare that they have no conflict of interest.

None.

None.

Not required.

Not applicable.

Not applicable.

All relevant data are available from an online repository or repositories. https://www.doi.org/10.2166/wqrj.2022.034.

Abdallah
H.
,
Jamil
T. S.
,
Shaban
A. M.
,
Mansor
E. S.
&
Souaya
E. R.
2018
Influence of the polyacrylonitrile proportion on the fabricated UF blend membranes’ performance for humic acid removal
.
Journal of Polymer Engineering
38
(
2
),
129
136
.
Ahmed
F. E.
,
Khalil
A.
&
Hilal
N.
2021
Emerging desalination technologies: current status, challenges, and future trends
.
Desalination
517
.
Elsevier B.V
.
https://doi.org/10.1016/j.desal.2021.115183
.
Baghbanzadeh
M.
,
Rana
D.
,
Lan
C. Q.
&
Matsuura
T.
2016
Effects of hydrophilic silica nanoparticles and backing material in improving the structure and performance of VMD PVDF membranes
.
Separation and Purification Technology
157
,
60
71
.
Bui
N.-N.
&
McCutcheon
J. R.
2013
Hydrophilic nanofibers as new supports for thin-film composite membranes for engineered osmosis
.
Environmental Science & Technology
47
(
3
),
1761
1769
.
Cath
T. Y.
,
Childress
A. E.
&
Elimelech
M.
2006
Forward osmosis: principles, applications, and recent developments
.
Journal of Membrane Science
281
(
1–2
),
70
87
.
Charcosset
C.
2009
A review of membrane processes and renewable energies for desalination
.
Desalination
245
(
1–3
),
214
231
.
Cornelissen
E. R.
,
Harmsen
D.
,
de Korte
K. F.
,
Ruiken
C. J.
,
Qin
J.-J.
,
Oo
H.
&
Wessels
L. P.
2008
Membrane fouling and process performance of forward osmosis membranes on activated sludge
.
Journal of Membrane Science
319
(
1–2
),
158
168
.
Emadzadeh
D.
,
Lau
W. J.
,
Matsuura
T.
,
Rahbari-Sisakht
M.
&
Ismail
A. F.
2014
A novel thin-film composite forward osmosis membrane prepared from PSF-TiO2 nanocomposite substrate for water desalination
.
Chemical Engineering Journal
237
,
70
80
.
Ezugbe
E. O.
,
Tetteh
E. K.
,
Rathilal
S.
,
Asante-Sackey
D.
&
Amo-Duodu
G.
2021
Desalination of municipal wastewater using forward osmosis
.
Membranes
11
(
2
),
1
12
.
https://doi.org/10.3390/membranes11020119
.
Han
G.
,
Chung
T.-S.
,
Toriida
M.
&
Tamai
S.
2012
Thin-film composite forward osmosis membranes with novel hydrophilic supports for desalination
.
Journal of Membrane Science
423
,
543
555
.
Hilal
N.
,
Ismail
A. F.
,
Matsuura
T.
&
Oatley-Radcliffe
D.
2017
Membrane Characterization
.
Elsevier
,
Amsterdam
.
Hoek
E. M. v.
,
Yan
Y.
&
Jeong
B.-H.
2020
Nanocomposite Membranes and Methods of Making and Using Same
.
Google Patents
.
Huang
L.
,
Bui
N.-N.
,
Meyering
M. T.
,
Hamlin
T. J.
&
McCutcheon
J. R.
2013
Novel hydrophilic nylon 6, 6 microfiltration membrane supported thin-film composite membranes for engineered osmosis
.
Journal of Membrane Science
437
,
141
149
.
Ismail
A. M.
,
Kotp
Y. H.
,
Eloffy
M. G.
,
Konsowa
A. H.
&
Shawky
H. A.
2021
Synthesis and characterization of novel forward osmosis thin film nanocomposite membrane to enhance the reverse osmosis desalination economics
.
Desalination and Water Treatment
244
,
90
109
.
https://doi.org/10.5004/dwt.2021.27958
.
Jeong
B.-H.
,
Hoek
E. M. v.
,
Yan
Y.
,
Subramani
A.
,
Huang
X.
,
Hurwitz
G.
,
Ghosh
A. K.
&
Jawor
A.
2007
Interfacial polymerization of thin film nanocomposites: a new concept for reverse osmosis membranes
.
Journal of Membrane Science
294
(
1–2
),
1
7
.
Jia
Q.
,
Li
Y.
,
Liu
Y.
&
Duan
D.
2021
Layer-controlled synthesis of a silanol-graphene oxide nanosheet composite forward osmosis membrane by surface self-assembly
.
Langmuir
37
(
27
),
8095
8106
.
https://doi.org/10.1021/acs.langmuir.1c00570
.
Lai
G. S.
,
Lau
W. J.
,
Goh
P. S.
,
Ismail
A. F.
,
Yusof
N.
&
Tan
Y. H.
2016
Graphene oxide incorporated thin-film nanocomposite nanofiltration membrane for enhanced salt removal performance
.
Desalination
387
,
14
24
.
Lee
S.
,
Boo
C.
,
Elimelech
M.
&
Hong
S.
2010
Comparison of fouling behavior in forward osmosis (FO) and reverse osmosis (RO)
.
Journal of Membrane Science
365
(
1–2
),
34
39
.
Li
J.-F.
,
Xu
Z.-L.
,
Yang
H.
,
Yu
L.-Y.
&
Liu
M.
2009
Effect of TiO2 nanoparticles on the surface morphology and performance of microporous PES membrane
.
Applied Surface Science
255
(
9
),
4725
4732
.
Lind
M. L.
,
Jeong
B.-H.
,
Subramani
A.
,
Huang
X.
&
Hoek
E. M. v.
2009
Effect of mobile cation on zeolite-polyamide thin film nanocomposite membranes
.
Journal of Materials Research
24
(
5
),
1624
1631
.
Lind
M. L.
,
Eumine Suk
D.
,
Nguyen
T.-V.
&
Hoek
E. M. v.
2010
Tailoring the structure of thin-film nanocomposite membranes to achieve seawater RO membrane performance
.
Environmental Science & Technology
44
(
21
),
8230
8235
.
Ma
N.
,
Wei
J.
,
Liao
R.
&
Tang
C. Y.
2012
Zeolite-polyamide thin film nanocomposite membranes: towards enhanced performance for forward osmosis
.
Journal of Membrane Science
405
,
149
157
.
Mollahosseini
A.
&
Rahimpour
A.
2014
Interfacially polymerized thin-film nanofiltration membranes on TiO2 coated polysulfone substrate
.
Journal of Industrial and Engineering Chemistry
20
(
4
),
1261
1268
.
Obaid
M.
,
Mohamed
H. O.
,
Yasin
A. S.
,
Fadali
O. A.
,
Khalil
K. A.
,
Kim
T.
&
Barakat
N. A. M.
2016
A novel strategy for enhancing the electrospun PVDF support layer of thin-film composite forward osmosis membranes
.
RSC Advances
6
(
104
),
102762
102772
.
Ong
R. C.
,
Chung
T.-S.
,
de Wit
J. S.
&
Helmer
B. J.
2015
Novel cellulose ester substrates for high-performance flat-sheet thin-film composite (TFC) forward osmosis (FO) membranes
.
Journal of Membrane Science
473
,
63
71
.
Sirinupong
T.
,
Youravong
W.
,
Tirawat
D.
,
Lau
W. J.
,
Lai
G. S.
&
Ismail
A. F.
2018
Synthesis and characterization of thin film composite membranes made of PSF-TiO2/GO nanocomposite substrate for forward osmosis applications
.
Arabian Journal of Chemistry
11
(
7
),
1144
1153
.
Song
X.
,
Zhang
Y.
,
Abdel-Ghafar
H. M.
,
Abdel-Aal
E. S. A.
,
Huang
M.
,
Gul
S.
&
Jiang
H.
2021
Polyamide membrane with an ultrathin GO interlayer on macroporous substrate for minimizing internal concentration polarization in forward osmosis
.
Chemical Engineering Journal
412
.
https://doi.org/10.1016/j.cej.2021.128607.
Stuart
B. H.
2004
Infrared Spectroscopy: Fundamentals and Applications
.
John Wiley & Sons
,
Hoboken, NJ
.
Wei
J.
,
She
Q.
&
Liu
X.
2021
Insights into the influence of membrane permeability and structure on osmotically-driven membrane processes
.
Membranes
11
(
2
),
1
22
.
https://doi.org/10.3390/membranes11020153
.
Xing
Y. L.
,
Xu
G. R.
,
An
Z. H.
,
Liu
Y. H.
,
Xu
K.
,
Liu
Q.
,
Zhao
H. L.
&
Das
R.
2021
Laminated GO membranes for water transport and ions selectivity: mechanism, synthesis, stabilization, and applications
.
Separation and Purification Technology
259
.
Elsevier B.V
.
https://doi.org/10.1016/j.seppur.2020.118192.
Yang
Y.
,
Zhang
H.
,
Wang
P.
,
Zheng
Q.
&
Li
J.
2007
The influence of nano-sized TiO2 fillers on the morphologies and properties of PSF UF membrane
.
Journal of Membrane Science
288
(
1–2
),
231
238
.
Yasukawa
M.
,
Mishima
S.
,
Shibuya
M.
,
Saeki
D.
,
Takahashi
T.
,
Miyoshi
T.
&
Matsuyama
H.
2015
Preparation of a forward osmosis membrane using a highly porous polyketone microfiltration membrane as a novel support
.
Journal of Membrane Science
487
,
51
59
.
Zhao
S.
,
Zou
L.
,
Tang
C. Y.
&
Mulcahy
D.
2012
Recent developments in forward osmosis: opportunities and challenges
.
Journal of Membrane Science
396
,
1
21
.
This is an Open Access article distributed under the terms of the Creative Commons Attribution Licence (CC BY 4.0), which permits copying, adaptation and redistribution, provided the original work is properly cited (http://creativecommons.org/licenses/by/4.0/).