Tubular PAN/CNC thin film nanocomposite (TFN) pressure retarded osmosis (PRO) membrane: fabrication and preliminary evaluation in desalination process

The pressure retarded osmosis (PRO) process requires high performance, high flux, high rejection, and resistant membranes under harsh conditions. Since conventional phase-inversion membranes are insufficient to permit the required water flux, alternative membrane fabrication methods need to be developed. Many studies have recently been carried out to fabricate strong enough nanofiber PRO membranes resistant to higher pressure while providing high flux and high rejection rates. This work aims to fabricate tubular nanofiber PRO membranes by the electrospinning technique. In the study, cellulose nanocrystals (CNCs) were added to polyacrylonitrile (PAN) polymer solution to fabricate nanocomposite nanofiber PRO membranes. According to the scanning electron microscopy (SEM), FT-IR, dynamic mechanical analysis, porometer, and contact angle analysis results, it is concluded that PAN and CNCs provided a complete mixture, and the addition of CNCs increased the mechanical strength in the PAN membranes, which is the crucial phenomena in PRO applications. In this study, the newly fabricated membrane achieves a higher PRO water flux of 405.38 LMH using 1 M NaCl and a DI as feed water. The corresponding salt flux is found as 2.10 gMH, which is higher than our previous study. The selectivity of the reversed flux represented by the ratio of the water flux to the reversed salt flux (Jw/Js) was able to be kept as high as 193.03 L/g for PRO operation. As far as we know, the performance of the work-developed membrane in this study has shown better performance than all PRO membranes reported in the literature previously.

Abstract The pressure retarded osmosis (PRO) process requires high performance, high flux, high rejection, and resistant membranes under harsh conditions. Since conventional phase-inversion membranes are insufficient to permit the required water flux, alternative membrane fabrication methods need to be developed. Many studies have recently been carried out to fabricate strong enough nanofiber PRO membranes resistant to higher pressure while providing high flux and high rejection rates. This work aims to fabricate tubular nanofiber PRO membranes by the electrospinning technique. In the study, cellulose nanocrystals (CNCs) were added to polyacrylonitrile (PAN) polymer solution to fabricate nanocomposite nanofiber PRO membranes. According to the scanning electron microscopy (SEM), FT-IR, dynamic mechanical analysis, porometer, and contact angle analysis results, it is concluded that PAN and CNCs provided a complete mixture, and the addition of CNCs increased the mechanical strength in the PAN membranes, which is the crucial phenomena in PRO applications. In this study, the newly fabricated membrane achieves a higher PRO water flux of 405.38 LMH using 1 M NaCl and a DI as feed water. The corresponding salt flux is found as 2.10 gMH, which is higher than our previous study. The selectivity of the reversed flux represented by the ratio of the water flux to the reversed salt flux (Jw/Js) was able to be kept as high as 193.03 L/g for PRO operation. As far as we know, the performance of the work-developed membrane in this study has shown better performance than all PRO membranes reported in the literature previously.

Introduction
Fossil fuels have a variety of harmful environmental effects by the emission of various toxins such as sulfur oxides (SOx) and nitrogen oxides (NOx), VOCs and greenhouse gases (Zhu et al. 2010;Alami et al. 2020). Researchers are investigating the potential of generating energy from wind, sun, water, biomass, and thermal heat to minimize the use of fossil fuels (Hussain et al. 2017;Tawalbeh et al. 2020a, b;Tawalbeh et al. 2020a, b). Electricity has recently been produced from osmotic pressure using an emerging power generation technology called pressure retarded osmosis (PRO). The mixing of two aqueous solutions of different salinities releases the Gibbs free energy of the mixing process, transformed into hydraulic pressure through a PRO process. The PRO process will then produce the electrical power through hydroturbines (Tawalbeh et al. 2020a, b;Jia et al., 2014;Skilhagen 2008).
PRO has attracted significant interest as a possible technology capable of extracting sustainable osmosis energy from salinity gradients (Cui et al. 2014;Bui and McCutheon 2014). PRO uses an osmotic pressure difference between feed (FS) and draw solution (DS) pressurized by hydraulic pressure (Loeb 1976;Yip et al. 2011;Kwon et al. 2021). The energy recovery method relies on placing on the draw stream a higher absolute pressure than that added to the feed stream. As a result, osmosis in a rise in the flow rate of the high-pressure stream-as long as the osmotic pressure differential (Dp = pD -pF) exceeds the difference in the applied absolute pressure (DP = PD -PF).
Where p and P are the osmotic and hydraulic pressure of the fluids, and D and F are the draw and feed streams, respectively. This illustrates a classic optimization problem that is essential for the implementation of PRO units. Higher applied pressure results in higher energy recovery per mole of solvent transfer and at a lower rate of solvent transfer at a steady state (Manzoor et al. 2020).
The osmotic gradient will determine the quantity of energy generated, similarly to other forward osmosis (FO) applications. By merging river and seawater, the estimated osmotic energy produced worldwide is between 1750 and 2000 TWh per year, which goes over the limit of one-tenth of the world's energy demand with a working pressure of 13.5 bar, equal to a 135 MWC in a hydroelectric power plant (Sikdar 2014;Chia et al. 2020). The average electricity generation in 2017 is about 481 TWh from biofuels, 4197 TWh from hydraulics, and 1127 TWh from wind energy, compared to other renewable energy technologies (Tawalbeh et al. 2020a, b;Chia et al. 2020;IEA 2010).
Many of the studies published in the literature were carried out in small-sized and, are aimed to improve the viability of using PRO for generating electricity. Figure 1 shows recent PRO research focus directions. To prove its feasibility in extracting energy back in 1976 using hollow fiber membranes, Loeb et al. (1976) experimentally tested the PRO process. Mehta and Loeb (1978) examined the influence of osmotic pressure on high pressures. They modeled the water transport in the PRO method in 1981. In this model, the internal effects of concentration polarization were considered (Lee et al. 1981). In 1990, Loeb and his colleagues performed an analysis evaluating the mechanical efficiency of the continuous operation of the PRO plant. The constant process of the underground PRO plant and the alternating flow configuration of the terrestrial PRO plant (Loeb et al. 1990). In 2009, the Norwegian company (Statkraft) built Loeb's first design of the PRO plant in Oslo, Norway (Statkraft 2009). Because of the expensive pre-treatment of river water needed and the limited electricity capacity due to the slight osmotic differential pressure between the seawater and the river, work at Statkraft terminated in December 2013 (Sarp et al. 2015). The internal concentration polarization model was developed by Xu et al. (2010) to integrate the effect of drawing solution using spiral-wound FO membrane modules. Increasing water permeation across the membrane has been shown to raise the concentration of draw solution; nevertheless, the negative impact of internal polarization of concentration has also been enhanced (Xu et al. 2010). Much of the studies performed up to 2011 centered on investigating and improving membranes in PRO operations. It should be mentioned that a thin film consisting of an active layer of polyamide and a support layer of polysulfone for the PRO process was developed by Yip et al. (2011). To enhance their hydrophilic properties and improve water permeability, polydopamine was applied over the supporting layers of two suitable reverse osmosis (RO) membranes (Arena et al. 2011).
Despite promising developments in PRO technology, owing to a lack of membranes with the suitable high-power density and good durability under highly pressurized PRO operating conditions, the PRO process has yet to be commercialized (Skilhagen et al. 2008;Cai et al. 2016). For instance, because of their poor water flux and low power density, prototype commercial cellulose acetate (CTA) membranes are Fig. 1 The focus directions in the recent PRO researches (Redrawn from Tawalbeh et al. 2020a, b) not commercially viable for use in the PRO process (under the conditions of smaller than 1 W m -2 using river water as feed and seawater as draw solution) (Kwon et al. 2021;Song et al. 2013;Chou et al. 2012;Vos 1966).
To withstand the harsh operating conditions of the PRO process, the effects of the CNC additive on the final membrane to be produced were investigated. Natural needle-rod fragments arising from acid hydrolysis of raw cellulose are CNCs, also known as nanocrystalline cellulose (NCC) (De Souza Lima 2004). The amorphous segments of cellulose, consisting of cellulose fibers, lignin, waxes, etc., degrade and remain remarkably crystalline as biomass is processed in a harsh acid environment (Asempour et al. 2018). Cellulose and cellobiose have the same b-(1-4)linkage as cellobiose. This is an expected situation since the most common method of obtaining cellobiose from cellulose hydrolysis ( Fig. 2) (French, 2017). Concerning the termini of its molecular axis, each cellulose chain has dimensional chemical asymmetry: one end is a chemically reduced group (i.e., hemiacetal unit) and the other end is a pendant hydroxyl unit, the nominal non-reducing end. The number of glucose units or the degree of polymerization (DP) is up to 20 000. Still, shorter cellulose chains may occur and are often located in the primary walls of the cells (Habibi 2010).
Depending on the source of cellulose and the conditions of acid treatment, the size of the CNCs varies, but typically they are a few hundred nanometers long and a few nanometers in diameter (Habibi et al. 2010;Peng et al. 2011). They have a high specific density, a large specific surface area, and a considerable negative zeta potential, as well as Young's modulus. They also have a highly reactive surface that makes them favorable for chemical functionalization due to the single bond -OH groups (Habibi et al. 2010;Peng et al. 2011). CNCs are biodegradable, renewable, have shallow impacts on the ecosystem, and are generally referred to as non-toxic and harmless particles (Habibi et al. 2010;Peng et al. 2011;Canada 2020). For a wide variety of possible uses, these appealing properties of CNCs have attracted considerable interest in the PRO process.
Usually, two categories of membranes have been extensively used for the PRO process. The first one is the thin-film composite (TFC) membrane with selective layers on porous supports and the second one is an engineered shield membrane developed using cellulose acetate (CA) and cellulose triacetate (CTA) (Sun and Chung et al. 2013;Lee et al. 2020). TFC membranes are commonly fabricated with a supporting substrate and a thin polyamide selective layer. It is feasible to fabricate TFC membranes to ensure support, and selective layers have a greater water flux than CA or CTA membranes (Gonzales et al. 2019). PRO membranes using a TFC layer have also been studied in various researches (Lee et al., 2020). In other experiments, Bui and McCutcheon (2014) and Song et al. (2013) developed nanofiber-based PRO membranes with nanofiber support for polyacrylonitrile (PAN) polymer. Nanofibers are known to have a high potential for developing osmotically-driven membranes due to their high porosity and low tortuosity (Bui et al. 2014;Shirazi et al. 2017;Son et al. 2018). The produced membranes achieved a power density of 21.3 W/m 2 @ 15.2 bar and 8.0 W/ m 2 @ 11.5 bar, when 1.06 M NaCl and 0.5 M NaCl were used as drawing solutions, respectively (Lee et al. 2020). The comparison of various studies was given in Table 1.
Mentioned studies have proved the feasibility of nanofibers for use in the fabrication of PRO membranes with high power density. However, owing to their comparatively lower mechanical flexibility at high hydraulic pressures in nanofiber-oriented PRO membranes, the risk of membrane deformation still  (French 2017) exists. To address the poor mechanical stability of nanofiber-based PRO membranes at different ratios, the mechanical strength of polyacrylonitrile (PAN) nanofiber support was enhanced by creating a CNC nanocomposites structure.
Apart from the recent PRO research direction, this study focused on strengthening the membrane material to cope with creeping phenomena. Polymeric membranes are mainly produced from petrochemical materials. One of the most important purposes is to use fewer petro-chemical materials during the production of membranes while increasing the strength in using the membrane in the PRO process. Because of this reason, this study concentrates on the biocompatible approach in this study by adding cellulose nanocrystals (CNCs) to the TFC membranes based on nanofiber.

Materials
Polyacrylonitrile (PAN) homopolymer (Mw = 150,000 Da from Sigma-Aldrich/USA) nanofiber base material and nanocellulose addition arranged as 1, 2, 5, and 10% wt/wt. DMAc is used as solvent from AK-KIM Chemicals/Turkey. A tailor-made tubular electrospinning device was used to fabricate nanofiber membranes over a hollow braided rope, and cellulose nanocrystals (CNCs) supplied from BGB Company/ Canada and hollow braided rope from Kord Technical Ropes/Turkey. m-Phenylenediamine (MPD) brought from Merck, Trimesoyl chloride (TMC) and n-hegzane supplied from Sigma Aldrich company.

Characterization methods of tubular nanofibrous membranes
Scanning electron microscopy (SEM) analysis The morphology of nanofiber membranes and thin-film nanocomposite (TFN) layer were described by the FEI Quanta FEG 250/Czech Republic equipment. Nanofiber membranes coated with Gold and Palladium (Au-Pd) with a thickness of nearly 3-4 nm using Quorum SC7620 equipment.
Pore size distributions analysis Using the Quantachrome 3G Porometer, the pore size distributions were measured. For porometry measurements, the membranes were split and pinched on both ends. Based on the total area, the pore sizes were calculated. Quantachrome Porofil was used as a porometry weighting agent with a low surface impedance of 16 dynes/cm (Quantachrome Ins., Florida, USA).
Dynamic mechanical (DMA) analysis Dynamic mechanical analysis (DMA) of fabricated membranes is performed for different CNC addition ratios using SEIKO Dynamic mechanical spectrometer Exstar 6100/Japan.
Contact angle analysis The hydrophilicity of fabricated CNC/PAN nanocomposite membranes is measured using a goniometer (Attension-KSV-Espoo/ Finland) with 2 lL of DI water droplet. 10 images are obtained from three different points on different sides of the membranes, and the mean average contact angle is recorded. The range is 0°-180°and the magnitude of the difference is ± 0.1°.
Polymer viscosity analysis Polymer viscosity analysis were completed with AND Vibro Viscometer SV-10/Japan equipment. The measurement range of the equipment between 0.3 and 10,000 mPa.s.

Fabrication of nanofibrous tubular mats on hollow braided rope
The configuration of the multi-nozzle bottom-up electrospinning system used in this research is shown in Fig. 3. A voltage adjuster in the ranges of 0-40 kV was used to apply a 30 kV voltage as a power supply. In this study, the distance between the nozzle and the collector was 20 cm, can be manually adjusted to achieve a uniform fiber formation and fiber layers  Interfacial polymerization reaction mechanism to form polyamide (PA) separation layer (Li and Wang 2010) without beads. In order to prepare the final polymer solution, first DMAc solvent was added to a clean polymer preparation bottle, and then a homogeneous mixture was obtained by adding CNC additive. An ultrasonic probe was used to completely dissolve the CNC additive in DMAc solvent. Before the PAN polymer solution was added to the mixture, it was left in the oven for a minimum of 5 h to remove any moisture that might be trapped in the solution. Then, 16% PAN polymer was added to the solution and mixed overnight under 50°C. After the polymer solution was prepared, the solvent in the polymer solution was allowed to cool down to room temperature to prevent rapid evaporation. The polymer solution is filled into a syringe and injected into the electrospinning system previously prepared at a pump flow rate of 1 ml/min. For each condition, the spinning shaft speed was kept constant and taken as 20 cm long. A hollow braided rope mounted on thin cylindrical stainless steel was coated with nanofibers. Each fabrication step is shown in Fig. 3, respectively.

Fabrication of thin-film nanocomposite (TFN) tubular CNC pressure retarded osmosis membranes
Fabricated support nanofiber membranes are primarily immersed in DI water for at least one hour. Trimesochloride (TMC) solution in hexane and aqueous m-phenylenediamine (MPD) solution are then applied to the membrane, respectively. To remove air bubbles in the solution, nitrogen gas was applied to the MPD solution. The interfacial polymerization reaction scheme is shown in Fig. 4. Fabricated final membranes were heated at 70°C for 5 min and stored in distilled water after post-treatment. A schematic view of the fabrication steps is shown in Fig. 5.
The thin-film nanocomposite membrane coating procedure is given in Table 2.

Pressure retarded osmosis setup
For the experiment, water flux and reverse leakage of salt have been measured. Figure 6 demonstrates the sizes of the TFC membranes fabricated from the electrospun-PAN/CNC-based PRO system. A pneumatic polyethylene (PE) hose with an internal diameter of 6 mm was used as housing in cross-flow  experiments. The holes at both ends of the hose are filled with heavy-duty industry-scale silicone. The concentrate was circulated via the shell-side of the membrane, while the feed solution was contacted through the lumen side. The PRO process was used a pump speed of 280 rpm on the feed side and 600 rpm on the draw solution side. The salt concentration changes in both the feed and draw solutions measured by EC meter. Feed side weight changes were recorded with digital balance. As a consequence of water passage from the feed solution side, the initial concentration of the drawing solution decreases. PRO tests were performed using 1 M NaCI as draw and distilled water as the feed solution under standard room temperature. Fabricated membrane modules are shown in Fig. 6. Membrane modules have been developed to facilitate the study of water and salt fluxes in a cross-flow system. A schematic draw of the PRO test setup is shown in Fig. 7.

Polymer viscosity measurement results
Polymer viscosity is one of the most important parameters in the electrospinning process. The solution viscosity increases as the CNC additive was added to the PAN polymer solution. At very low polymer viscosities, the polymer coming out of the nozzle tip does not form a Taylor cone, and when there is a very dense solution, the solvent evaporates and blockages occur as a result of a long waiting time at the nozzle tip. Very dense and very low polymer viscosities are undesirable in electrospinning processes. In Fig. 8, the prepared solutions are among the solution viscosities applicable in the electrospin process.  Surface morphology and fiber thickness of the nanofiber membranes are given for the support layer in Fig. 9, the surface of thin-film composite (TFN) layer in Fig. 10, and the cross-section of the TFN layer in Fig. 11.

Determination of pore size
It is commonly recognized that decreasing the concentration of the solvent or the molecular weight can decrease the fiber diameter due to lower viscosity (Tan et al. 2005). The viscosity of the solution produces an anti-electrostatic repulsion force responsible for the stretching and thinning of the solution jet. Unlike several other parameters, it has repeatedly been shown that higher viscosity results in higher fiber diameters, irrespective of the material used.
In effect, the diameter and thickness relate to the distribution of the pore size and the shape of the pore. The electrospinning polymer does not have adequate entanglements to create stability at a very low viscosity of the polymer solution, which leads to droplet formation (Murthe et al. 2019). Figure 12 reveals that increased volume of CNC addition to the PAN polymer solution, fabricated nanocomposite nanofiber membranes display a diminishing porosity pattern (Pasaoglu and Koyuncu 2021).

Dynamic mechanical analysis
Dynamic mechanical analysis (DMA) of fabricated membranes is performed for different CNC addition ratios using SEIKO Dynamic mechanical spectrometer Exstar 6100/Japan. Before testing the fabricated membranes were taken by stripping from the tubular hollow braided rope, and the measurements were made in the test equipment similar to the measurement of flat-sheet nanofiber membranes. Figure 13 shows that an increasing amount of CNCs are being increased for Fig. 9 Surface SEM images of tubular nanofibrous support youngs' modulus of engineered membranes, which is a critical parameter for resistance to high pressure. The higher viscosity of the polymer solution at higher CNC concentrations is the factor for this efficiency improvement.

Fourier-transform infrared spectroscopy (FT-IR) of the fabricated membranes
Fourier-transform infrared spectroscopy (FT-IR) has been identified for the dry support layer membrane to check the chemical composition of the organic  molecules and the potential structural modifications that occur by CNC in addition to the PAN process. Figure 14 reveals that the FT-IR range of fabricated membranes proves that with a rise in crystal nanocellulose, the C-N group reacted with the cellulose group and disappeared at 1,664.58 cm -1 point.
The peak of CNC was present at 1,620 cm -1 in the spectrum due to C=O interconnectivity, which is one characteristic of lignin and lignin/hemicellulose (Asrofi et al. 2017). The peaks between 3200 and 3600 cm -1 correspond to O-H streching vibrations. Figure 14 also indicates that the PAN spectrum value of 1,452 cm -1 is the normal C-H stretching methylene band characteristic, and the C-H stretching vibration is the typical vibration of 2,923 cm -1 (Jin et al. 2018). Figure 15 reveals that CNC/PAN TFC layer FT-IR results, PA observed at approximately 1663 cm -1 (C=O streching, amid I), 1610 cm -1 (H-bonded C=O streching) and 1542 cm -1 (N-H plane bending, amide II) .

Contact angle analysis of the fabricated membranes
The hydrophilicity of fabricated CNC/PAN nanocomposite membranes is measured using a goniometer (Attension-KSV-Espoo/Finland) with 2 lL of DI water droplet. 10 images are obtained from three different points on different sides of the membrane, and the mean average contact angle is recorded. The range is 0°-180°and the magnitude of the difference is ± 0.1°. According to the findings, the contact angle dramatically decreased with increased CNC to PAN solution ratios tending to increase membrane water flux in the process shown in Fig. 16.
According to literature studies by Voisin et al. (2017) and Cheng et al. (2017) the increasing amount of CNC helps reduce the value of the membrane contact angle.

PRO test results
The membrane is PRO-oriented while the drawing solution is placed against the active layer and the feed solution is positioned against the support layer. Under the same conditions, 2% CNC added PAN nanocomposite nanofiber TFC-PRO membranes perform higher water flux which is shown in Fig. 17. The addition of CNC in the PAN polymer solution allowed the flux to increase in the membranes. Although a very high flux was seen on the membrane with a 5% CNC additive, this flux value decreased rapidly. Most of the fabricated membranes maintain their initial flux after 60 min of operation.
The transport parameters A, B and, salt rejection percentage of fabricated PAN/CNC(2%) membrane were measured and calculated results were given in Table 3.
Reverse salt flux indicates salt leakage from the PRO membrane to the side of the feed solution. After a certain time salt leakage, the draw side is diluted, leading to a reduced osmotic pressure difference. Thus, although ideal for the development of PRO membranes, it is crucial to have the lowest reverse salt flux and higher water flux. Thus, although ideal for the development of PRO membranes, it is crucial to have the lowest reverse salt flux and higher water flux. This research proved that in Fig. 18, where 1% CNC added PAN nanocomposite nanofiber TFC membranes provide the best results.
A further distinction was made between this research and the literature for the application of PRO in Table 4 in terms of water flux and reverse salt flux of membranes. The results show that fabricated membranes in this study have one of the highest J w /J s ratios in the literature.

Conclusions
Thin-film nanocomposite (TFN) CNC added PRO membranes were successfully fabricated with a tailormade electrospinning machine. During operation, fabricated PRO membranes fulfill the needs caused by higher pressures. Increasing the amount of CNCs has an impressive impact on the youngs' modulus that makes the PRO membrane stronger. FT-IR spectra indicate that similar groups reacted with cellulose groups and modified the spectra, and it was observed that increasing the amount of CNCs made the membrane more hydrophilic. Apart from the FT-IR spectrum findings, it can be inferred that CNCs content is