1 Recycled reverse osmosis membranes for forward osmosis technology Jorge Contreras-Martíneza, Carmen García-Payoa, Paula Arribasb, Laura Rodríguez-Sáezc,d, Amaia Lejarazu-Larrañagac,d, Eloy García-Calvoc,d, Mohamed Khayeta,c,* a Department of Structure of Matter, Thermal Physics and Electronics, Faculty of Physics, University Complutense of Madrid, Avda. Complutense s/n, 28040 Madrid, Spain. b Department of Electrical Engineering, ICAI, Comillas Pontifical University, Alberto Aguilera 25, 28015 Madrid, Spain. c IMDEA Water Institute, Avda. Punto Com nº 2, 28805 Alcalá de Henares, Madrid, Spain. d Chemical Engineering Department, University of Alcalá, Ctra. Madrid-Barcelona Km 33.600, 28871. Alcalá de Henares, Madrid, Spain. * Corresponding author. Phone: +34-91-3945185; E-mail: khayetm@fis.ucm.es (M. Khayet) KEYWORDS: Discarded reverse osmosis modules; Recycled membranes; Forward osmosis; Interfacial polymerization; Wastewater. 2 ABSTRACT: An alternative use of end-of-life reverse osmosis (RO) membranes is proposed for forward osmosis (FO) application as recycled FO (RFO) membranes and transformed recycled FO (TRFO) membranes. Different passive cleaning protocols in pilot plant and laboratory scale were followed using sodium hypochlorite (NaClO) at different concentrations and exposure time. The RFO with the best performance was selected for its transformation by interfacial polymerization (IP) technique to improve further the FO performance. Both the morphological structure and transport properties of the RFO and TRFO membranes were studied by means of different characterization techniques. Although the RFO membranes are suitable for FO, the TRFO membranes are more competitive. The highest FO water permeate fluxes (12.21 kg/m2h and 15.12 kg/m2h) were obtained for the membrane recycled applying the highest NaClO exposure dose applied in pilot plant (106 ppm·h) followed by IP of a thin polyamide layer. These permeate fluxes were better or at least comparable to commercial membranes used under the same FO conditions. The results indicated that it is possible to use discarded RO membranes in FO technology for wastewater treatment after adequate treatment procedures extending their lifetime and contributing to a circular economy and sustainability in membrane science and related materials. 3 HIGHLIGHTS • Use of end-of-life reverse osmosis (RO) membranes in forward osmosis (FO) • Lifetime extension of discarded RO membranes and minimization of their disposal • Passive cleaning protocols with different NaClO concentrations and exposure time • Formation of polyamide and polyester thin layers by interfacial polymerization • Better FO performance of recycled RO membrane than the commercial membrane CTA- HTI 4 1. Introduction 1 Both the demographic increase and the climatic change aggravate the global situation of water 2 scarcity and the search for unconventional resources is demanded [1]. Desalination and wastewater 3 reuse are two viable cost-effective ways of achieving water security in many regions of the world. 4 Membrane-based technologies have been applied significantly for water treatment because of their 5 high sustainability criteria in terms of environmental impacts, land usage, ease of use, flexibility 6 and adaptability, compared to other techniques such as thermal and chemical approaches [2]. 7 The worldwide total desalination capacity of clean water per day is 99.8 Hm3 among which 70% 8 are represented by reverse osmosis (RO) desalination plants in which the spiral wound thin film 9 composite (TFC)-polyamide (PA) membrane modules are the most used [3]. However, RO 10 desalination technology has some environmental drawbacks that need to be addressed and 11 resolved. The major environmental challenge associated to energy consumption of fossil resources 12 because of their repercussion in climate change, together with both the inland and coastal 13 discharges of high concentrated brines and their adversely negative environmental impacts are the 14 principal issues of RO technology. Some strategies have been proposed for the treatment of RO 15 brines taking into consideration their high osmotic pressure [4]. In addition, the short lifespan of 16 the used membrane modules, which is about six year average [5], is nowadays another 17 environmental issue of this technology. In RO desalination plants, it is required a membrane 18 replacement rate of about 5 – 20 % each year even if different strategies are considered to control 19 and mitigate fouling of RO membrane modules [6]. More than 840000 end-of-life RO membrane 20 modules are discharged annually worldwide [7]. The most common management of the discarded 21 membrane modules is their deposition in landfills near desalination plants. The European Union 22 (EU) in the Directive 2008/98/CE clearly established the following hierarchy for wastes serving 23 as an order of priorities in legislation and policy in favor of prevention and waste management 24 5 (prevention, preparation for reuse, recycling, energy recovery and final disposal) [8]. Therefore, it 25 is necessary to find out new ways to properly manage this large amount of solid waste by reusing 26 and recycling discarded RO membrane modules. 27 Different recycling techniques of end-of-life RO membranes modules have been proposed 28 recently [5, 7, 9, 10]. These techniques include chemical conversion using different oxidative 29 agents of the dense polyamide (PA) active layer of the RO membranes by means of active 30 recirculation or passive immersion. The dense PA layer of the discarded RO membranes can be 31 partially or completely degraded depending on the concentration of the oxidative agent used and 32 its exposure time. As consequence, discarded RO membranes can be converted into NF or UF 33 membranes, depending on the degree of exposure of the membrane to free chlorine [11]. 34 Sodium hypochlorite (NaClO) has been proposed as an oxidative agent of PA because its main 35 effect is the degradation and separation of the thin PA layer from PSf layer through hydroxide-36 induced amide link scission [12]. Different concentrations of free chlorine (i.e. hypochlorous acid 37 and hypochlorite ion) and its exposure time were considered depending on the type and observed 38 effects of membrane fouling. By using this agent, the recovery of discarded RO modules for their 39 reuse in nanofiltration (NF), ultrafiltration (UF), electrodialysis (ED) and membrane bioreactors 40 (MBR) were reported [11, 13-15]. Therefore, the use of recycled membranes contributes to the 41 circular economy scheme. Senán-Salinas et al. [16] evaluated recycled RO membranes with Life 42 Cycle Assessment (LCA) and cost effectiveness analysis. LCA results showed that the most 43 environmentally interesting transformations are passive systems with recycling costs of € 25.9 – 44 41.5 per membrane module. It must be pointed out that NaClO has not been used yet to recover 45 discarded modules for their reuse in forward osmosis (FO) technology. 46 FO is a separation process of emerging interests during last decade for wastewater treatments 47 [17-20]. It allows the concentration of feed aqueous solutions by the application of an osmotic 48 6 pressure difference through a semipermeable membrane using a draw solution with very low 49 chemical potential of water [21]. Compared to RO, FO technology requires less specific energy 50 consumption since water flux takes place naturally by means of the osmotic driving force through 51 a semi-permeable membrane from a low osmotic pressure solution to a high osmotic pressure 52 solution. Other advantages of FO are its lower fouling tendency, resulting in easier cleaning, 53 extending therefore membrane lifetime; and its higher selectivity compared to RO [22, 23]. 54 However, the product of FO is not potable water but a diluted draw solution that requires a 55 regenerating step to extract clean water [22]. It is worth quoting that FO, alone or in combination 56 with other processes [17, 19, 24-27], has been proposed as an appropriate technology for 57 wastewater treatment [19, 28, 29] with a high energy efficiency [20, 30]. 58 A suitable membrane for FO process must exhibit among others a high hydrophilic character, a 59 high water permeance, a highly selective thin layer with a support having a low tortuosity factor 60 and a high porosity [31]. Maximizing the permeate water flux while minimizing the reverse solute 61 permeate flux that causes a decline of the transmembrane osmotic pressure gradient is one of the 62 proposed objectives in FO membrane engineering [32]. Various types of membranes were 63 developed to optimize FO separation process [25, 33, 34]. Some new FO membranes exhibit a 64 thin-film composite (TFC) structure similar to that of RO membranes [35]. Tiraferri et al. [36] 65 studied the influence of the TFC membrane support layer structure on FO performance and 66 concluded that both the active layer transport properties and the support layer structural 67 characteristics should be optimized in order to achieve a high FO performance. 68 In the present study, an attempt has been made for the first time to recycle discarded RO 69 membranes for their application in FO process. This avoids their disposal in landfills giving them 70 a new use. For their cleaning, different concentrations of the NaClO oxidative agent and exposure 71 time have been considered. The recycled FO membranes (RFO) were first characterized and tested 72 7 in FO for the treatment of humic acid (HA) solutions. Then, the RFO membrane with the best FO 73 performance was selected as support of the TFC FO membranes. Transformed recycled FO 74 membranes (TRFO) were prepared by interfacial polymerization (IP) technique in order to 75 improve its FO performance. It is worth quoting that various research studies have been already 76 carried out on IP using different types of monomers in both the aqueous solution and organic 77 solution [37, 38]. The obtained TRFO membranes were also characterized and tested in FO for the 78 treatment of HA solutions and finally compared to FO commercial membranes. 79 80 2. Materials and methods 81 2.1. RO modules and recycled forward osmosis membranes (RFO) 82 End-of-life spiral wound RO modules (TM720-400 by Toray Industries, Inc., Japan) discarded 83 by a brackish water desalination plant located in Almeria (Spain) were used in this study. These 84 modules have 8" (equivalent to 0.2032 m) diameter and contain a TFC–PA membrane composed 85 of three layers: a PE support layer, a PSf porous intermediate layer and a selective PA dense top 86 layer. The water permeance and salt rejection factor of the discarded RO membrane modules are 87 2.04 ± 0.06 L/m2·h·bar and 98.31 ± 0.24%, respectively. The RO tests for evaluating the discarded 88 RO membranes were carried out for 5 min with a feed aqueous solution containing NaCl (2,000 89 ppm), MgSO4 (2,000 ppm) and dextrose (250 ppm) at 303 K under a transmembrane pressure of 90 5 bar. After their autopsy, the membranes of the discarded RO modules presented a clay-like 91 appearance (i.e. brown color, odorless with a colloidal fouling texture composed by 84% inorganic 92 and 16% organic compounds) and a water contact angle of the active layer of 46 ± 2° without any 93 superficial damage [39-41]. The discarded RO modules were conserved in sodium bisulphite (500-94 1000 ppm). Prior their characterization, membrane coupons were first extracted from the modules, 95 washed and conserved in Milli-Q water. For the present study, two types of recycled membranes 96 8 were used according to the followed passive cleaning procedure using free chlorine as PA 97 degradant at basic pH (Table 1). 98 99 Table 1. Free chlorine exposure doses to obtain RFO membranes. 100 Membrane code Cleaning in pilot plant Cleaning in lab Total dose level (ppm·h) Exposure Time (h) Free chlorine (ppm) Dose level (ppm·h) Exposure Time (h) Free chlorine (ppm) Dose level (ppm·h) RFO1 65 12.4·103 8·105 - - - 8·105 RFO2 81 12.4·103 106 - - - 106 RFO3 65 12.4·103 8·105 7 7·103 49·103 8·105+49·103 RFO4 81 12.4·103 106 7 7·103 49·103 106+49·103 101 The choice of free chlorine was motivated by the low tolerance of RO membranes to it (< 1000 102 ppm·h using a free chlorine concentration of 1 ppm) [42]. NaClO solutions were prepared by 103 diluting NaClO in water until achieving the free chlorine concentration set for the experiments 104 (12385 ppm for cleaning in pilot plant and 7000 ppm for cleaning in lab) [41]. In the pilot plant, 105 the free chlorine concentration was controlled according to the in-situ determination of pH, 106 conductivity and redox in order to guarantee a constant concentration in each assay. In laboratory, 107 the free chlorine concentration was measured by a Pharo 100 Spectroquant spectrophotometer 108 (Merck). The NaClO solutions presented a high physical and chemical stability together with a 109 high capacity to be reused in successive treatments. The lowest applied exposure dose was 3·105 110 ppm·h because UF-like membranes from discarded RO membranes were previously obtained 111 following the laboratory cleaning procedure [43]. In addition, starting from this exposure dose, the 112 elimination of the PA layer could be guaranteed obtaining a recycled membrane formed by a PSf 113 layer supported on a PE backing material [43]. 114 9 The first type of RFO membranes (RFO1 and RFO2) was subjected to a passive cleaning process 115 in the pilot plant with NaClO solution of 12385 ppm of free chlorine at different exposure times 116 [41]. Under this cleaning protocol it was not necessary to disassemble the discarded RO modules. 117 The second type of RFO membranes (RFO3 and RFO4) were cleaned twice, first in the pilot plant 118 and then in the laboratory. The second cleaning in laboratory was carried out by a passive cleaning 119 using NaClO (6200 ppm of free chlorine) under a total level dose of 49 103 ppm·h [41]. In this 120 case, it was necessary to disassemble the RO membrane module and cut the necessary membrane 121 coupons for their immersion in the oxidative agent´s solution, increasing therefore the recycling 122 cost with respect to the previous cleaning protocol and limiting the reuse of the complete module 123 as well. Under these high total exposure doses, the final degradation of the PA layer of the RO 124 membranes was ensured as demonstrated elsewhere [11]. The complete degradation of the PA 125 layer is necessary since it is supposed to be the fouled layer of the RO membrane [44]. Therefore, 126 the remained supported PSf membrane (as shown in Fig. S1c of the Supporting Information) was 127 proposed for the first set of FO separation tests and characterization. 128 For sake of comparison of the efficiency of the different cleaning protocols, one more type of 129 RFO membrane (RFO5) was prepared skipping the first cleaning in pilot plant and cleaning the 130 discarded RO membrane only under laboratory protocol with a total exposure dose of 3·105 ppm·h. 131 All related details can be found in the Supporting Information (section S6). 132 133 2.2. Interfacial polymerization (IP) of the recycled forward osmosis membrane (TRFO) 134 Interfacial polymerization (IP) is a well-studied membrane modification method exhibiting 135 various advantages as it allows the formation of TFC membranes with excellent performance in 136 RO, NF, MF and FO applications. Different monomers with different concentrations and 137 application time were considered. Compared to CTA membrane, TFC-FO membranes prepared 138 10 by IP display promising characteristics due mainly to the benefits of both the active skin layer and 139 the support layer. 140 TFC membranes were prepared by IP using as support the RFO membrane with the best FO 141 performance (RFO2 membrane as will be discussed later on). In our previous study [38], IP was 142 followed to prepare PA or PE thin layer on commercial polyethersulfone (PES) filtration 143 membrane (HPWP Merk Millipore, 0.45 µm pore size) using different monomers in the aqueous 144 phase for the formation of the PA layer (m-phenylenediamine, MPD; piperazine, PIP; and 145 combinations of trimethylamine, TEA and polyvinyl alcohol, PVA) and bisphenol A (BPA) for 146 the formation of the PE layer while TMC was the monomer used in the organic phase. These TFC 147 membranes showed high selectivity to humic acid feed aqueous solutions. In the present study, the 148 same solutions and procedure were followed using as support the selected RFO membrane in order 149 to improve further its FO performance. The employed monomers, their concentrations and 150 combinations together with the reaction time of the IP and membrane names are summarized in 151 Table 2. 152 153 Table 2. IP parameters used to prepare thin film composite membranes (TRFO) using as support 154 the selected RFO2 membrane. 155 Membrane TFC type Aqueous phase Organic phase Material w/v (%) Time (min) Material w/v (%) Time (min) RFO-MPD PA MPD 2 60 TMC 2.5 15 RFO-MPD/TEA PA MPD-TEA 1-1 60 TMC 2.5 15 RFO-PIP PA PIP 2 60 TMC 2.5 15 RFO-PIP/TEA PA PIP-TEA 1-1 60 TMC 2.5 15 RFO-PIP/PVA PA PIP-PVA 1-1 60 TMC 2.5 15 RFO-BPA PE BPA 2 60 TMC 2.5 15 11 To prepare the thin layer (PA or PE), the PSf layer of the selected RFO membrane was brought 156 into contact first with the aqueous solution for 1 h, then with the organic solution for 15 min. 157 Subsequently, it was air-dried in a dark environment for 24 h at room temperature (293 K) before 158 characterization. More details of the IP technique can be found elsewhere [38]. All the above cited 159 chemicals were purchased from Sigma-Aldrich Chemical Co. St. Louis, Massachusetts, USA. 160 161 2.3. Commercial membranes 162 For sake of comparison, two asymmetric FO commercial PA (TFC-HTI) and cellulose triacetate 163 CTA (CTA-HTI) membranes supplied by Hydration Technology Innovations (HTI™, LLC, 164 Albany, USA) were used. Both membranes are embedded in a PE mesh as shown in Figs. S1a and 165 S1b. The CTA-HTI membrane has been used extensively in various research studies [45-49]. The 166 characteristics of the prepared membranes were also compared with those of the pristine TM720-167 400 membrane supplied by Toray Industries, Inc., Japan. 168 169 2.4. Membrane characterization 170 Water contact angle (θw) of membrane samples was measured at room temperature by a CAM 171 100 Contact Angle Meter (KSV Instruments Ltd., Monroe, Connecticut, USA). The membrane 172 thickness (δ) was measured by a digital micrometer (model 1724-502 series, Helios-Preisser 173 Instruments, Gammertingen, Germany). Scanning electron microscopy (SEM, JEOL Model JSM-174 6335F, Jeol Ltd., Tokyo, Japan) was used to get images of both the surface and cross-section of 175 the membranes. Previously, the membrane samples were cut in liquid nitrogen and then coated 176 with a thin layer of gold using a sputter coater (EMITECH K550 X, Emitech Groupe, Montigny, 177 France) with a current of 25 mA during 1 min. 178 12 To quantify the surface charge of the membranes streaming potential measurements were 179 performed with SURPASS equipment (Anton Paar GmbH, Austria). All measurements were 180 conducted with an adjustable-gap cell where two membrane samples of 20 mm × 10 mm were 181 fixed on sample holders using double-sided adhesive tape. In this study, a flow channel gap of 100 182 µm was set between the sample surfaces. First, the samples were thoroughly rinsed with the testing 183 electrolyte (1 mM KCl aqueous solution) and the pH was adjusted to the required value using 0.1 184 M HCl or 0.1 M NaOH solution. All zeta potential (ζ-potential) measurements were carried out at 185 25 ± 2 ºC and the pH range was varied from 5.5 to 2.5 with a step of 0.3. Three ζ-potential values 186 were obtained for each pH value. 187 Attenuated Total Reflectance Fourier Transform Infrared Spectroscopy (ATR-FTIR) was used 188 to confirm the total degradation of the PA layer of the discarded RO membranes after cleaning and 189 to analyze the chemical composition of the prepared TRFO membranes. The measurements were 190 carried out with a Nicolet iS50 device equipped with the deuterated triglycine sulfate-potassium 191 bromide (DTGS-KBr) detector, a KBr beam splitter and an infrared source (Ever-Glo); and having 192 a maximum resolution of 0.09 cm-1. 193 Solute separation transport tests were performed using polyethylene glycol (PEG) and 194 polyethylene oxide (PEO) of different molecular weights. The selected molecular weights were 195 0.4, 1, 6, 10, 12, 20 and 35 kg/mol for PEGs and 100 and 200 kg/mol for PEOs [50]. The set-up 196 used was described in our previous paper [51]. The detailed procedure together with the obtained 197 results of water permeance (WP) and solute rejection factor (α) are included in Section S3. All 198 products were purchased from Sigma-Aldrich Chemical Co. St. Louis, Massachusetts, USA. The 199 feed temperature, pressure and solute concentration were 296 K, 1.04 bar and 200 ppm, 200 respectively. 201 13 As it was described in [52], the mean pore size and the geometric standard deviation were 202 determined from the plot of the solute rejection factor as a function of the solute diameter and their 203 correlation according to the log-normal probability function. The mean pore size (𝜇𝑝) and the 204 molecular weight cut-off (MWCO) correspond to the Einstein-Stokes diameters of 50% and 90% 205 solute rejection factors, respectively. The geometric standard deviation is the ratio between the 206 Einstein-Stokes diameters of 84.13% and 50% rejection factors. The pore density (N) and porosity 207 () were determined from the permeance data of PEG and PEO aqueous solution as it is explained 208 in the Supporting Information (Eqs. (S4) and (S5)). 209 To estimate the tortuosity factor (τ), Bruggeman correlation, which has been proved to be a good 210 approximation relating τ with  , was employed [53]: 211 𝜏 = 𝜀−0.5 (1) 212 The structural parameter (S) is commonly considered as an important characteristic of FO 213 membranes because of its influence on the internal concentration polarization (ICP) phenomenon 214 and membrane´s structural resistance to solute mass transport. S, often called the “intrinsic” 215 structural parameter, 𝑆𝑖𝑛𝑡, was described as the average diffusive path length through the 216 membrane support layer as follows [54, 55]: 217 𝑆𝑖𝑛𝑡 = 𝜏𝛿 𝜀 (2) 218 219 2.5. Forward osmosis (FO) experiments 220 All membranes were tested in the FO set-up (Lewis cell) shown in Fig. 1. This is composed of 221 two stainless steel cylindrical chambers having an internal volume of 250 mL. The membrane, 222 with an effective area of 1.59·10-3 m2, is placed between the two chambers with its selective layer 223 14 facing the feed side (FO mode). Both chambers are double walled so that a thermostatic liquid can 224 circulate through them and keep constant the temperature of both the feed and permeate at 296 K. 225 226 Figure 1. FO experimental set-up: 1. magnets; 2. stirrers; 3. temperature sensors (pt100 probes); 227 4. inlets; 5. permeate; 6. reservoir; 7. thermostat (TE-8D, Techne Inc., New Jersey, USA); 8. digital 228 temperature indicator (Temp. Meßgerät pt100, PHYWE Systeme GmbH und Co. KG, Göttingen, 229 Germany); 9. electric motor with speed control (K50640, Kelvin S.A., Madrid, Spain); 10. 230 Membrane support; 11. Balance (Sl-2002, Denver Instrument Company, Arvada, USA). 231 232 2.5.1. Evaluation of the effective structural parameter, Seff, from FO experiments 233 As stated earlier, the structural parameter, S (Eq. (2)), is widely used as an assessment of FO 234 membrane structural characteristics. However, measuring porosity and tortuosity is a challenging 235 task. Most researchers have adopted a fitted parameter mass transfer model to determine Seff from 236 RO test obtaining the membrane water permeance (parameter A) and the membrane salt permeance 237 (parameter B). This method is questionable since A and B are different in RO process subjected to 238 15 much higher pressure than in FO process. Furthermore, the solute concentration at the membrane 239 selective layer is much higher in RO test than in FO because of the higher applied hydrostatic 240 pressure in RO [54]. In addition, FO membranes are not generally designed to bear high hydrostatic 241 pressures like those applied in RO and therefore the permselective properties of the FO membrane 242 could be altered during RO test (i.e. the A and B values could be different) [49]. To estimate an 243 “effective” or fitted structural parameter, semi-empirical models, based on fitted parameter mass 244 transfer equations from FO experimental flux measurements, have been used [54]: 245 𝑆𝑒𝑓𝑓 = 𝐷 𝐽𝑤 ln ( 𝐵 + 𝐴𝜋𝐷,𝑏 𝐵 + 𝐽𝑤 + 𝐴𝜋𝐹,𝑚 ) (3) 246 where, D is the solute coefficient diffusion, 𝐽𝑤 is the average water flux, A is the water permeance 247 coefficient of the membrane, B is the solute permeance coefficient of the membrane, 𝜋𝐷,𝑏 is the 248 osmotic pressure of the bulk draw solution, and 𝜋𝐹,𝑚 is the osmotic pressure of the feed solution 249 at the membrane side. It is worth noting that both the water and solute permeance coefficients (A 250 and B, respectively) are typically obtained from additional RO tests using different applied 251 pressures. 252 In FO experiments, water together with a little quantity of feed solute permeate from the feed to 253 the draw solution, which becomes diluted within the membrane and draw solution/membrane 254 boundary layer causing both internal concentration polarization (ICP) and external concentration 255 polarization (ECP). In fact, ECP occurs externally at both sides of the membrane surface. At the 256 same time, draw solute diffuses from the draw solution through the membrane to the feed solution 257 boundary layer [56]. Therefore, the effects of both the selective layer side and the draw side of the 258 membrane must be taken into account in theoretical modelling. Tiraferri et al. [57] derived a 259 transport model for FO taking into account both the concentrative ECP on the feed side, the dilutive 260 ICP and the reverse solute flux as: 261 16 𝐽𝑤 = 𝐴 { 𝜋𝐷 exp (− 𝐽𝑤𝑆𝑒𝑓𝑓 𝐷 ) − 𝜋𝐹 exp ( 𝐽𝑤 𝑘 ) 1 + 𝐵 𝐽𝑤 [exp ( 𝐽𝑤 𝑘 ) − exp (− 𝐽𝑤𝑆𝑒𝑓𝑓 𝐷 )] } (4) 262 𝐽𝑠 = 𝐵 { 𝐶𝐷 exp (− 𝐽𝑤𝑆𝑒𝑓𝑓 𝐷 ) − 𝐶𝐹 exp ( 𝐽𝑤 𝑘 ) 1 + 𝐵 𝐽𝑤 [exp ( 𝐽𝑤 𝑘 ) − exp (− 𝐽𝑤𝑆𝑒𝑓𝑓 𝐷 )] } (5) 263 where 𝑘 is the feed mass transfer coefficient and D is the bulk diffusion coefficient of the draw 264 solution. Note that the terms exp ( 𝐽𝑤 𝑘 ) and exp (− 𝐽𝑤𝑆𝑒𝑓𝑓 𝐷 ) are related to the concentrative external 265 concentration polarization (ECP) and dilutive internal concentration polarization (ICP) 266 phenomena, respectively. Tiraferri et al. [57] designed an alternative and simple methodology to 267 estimate the parameters A, B and Seff solely from FO experiments using different concentrations of 268 a given draw solution and distilled water as feed. In this study, the Seff structural parameter of RFO, 269 TRFO and FO commercial membranes was determined using the FO experimental results (J and 270 Js as will be explained in Eqs. (7) and (8)) obtained at different NaCl concentrations of the NaCl 271 draw solution (15, 40, 65 and 90 g/L) while Milli-Q water was employed as feed. By using the salt 272 osmotic pressure at different NaCl concentrations and the Excel spreadsheet provided by Tiraferri 273 et al. [57] the parameters A, B and 𝑆𝑒𝑓𝑓 were calculated. 274 By using van’t Hoff approximation, the osmotic pressure of the draw solution was assumed to 275 be linearly proportional to the concentration: 276 𝜋 = 𝑖𝑐𝑅𝑇 (6) 277 where c is the molarity (in mM), T is the absolute temperature (in K), R is the ideal gas constant 278 and i is the van´t Hoff factor presenting the number of ionic species dissociated in the aqueous 279 solution, feed or draw solution. For NaCl aqueous solution, i = 2. 280 17 Since 𝑆𝑖𝑛𝑡 depends only on the geometrical factors of the membrane (see Eq. (2)), it can be 281 assumed independent on the solute transport rates [55]. It is to be noted that although in general 282 𝑆𝑒𝑓𝑓 and 𝑆𝑖𝑛𝑡 are assumed to be identical, some researchers claimed their discrepancy [21, 54, 55, 283 58, 59]. 284 285 2.5.2. Wastewater treatment by FO separation process 286 Sodium chloride (NaCl, Scharlau Chemicals Co., Barcelona, Spain) aqueous solutions of 65 g/L 287 and 200 g/L were used as draw solutions and HA (Sigma-Aldrich Chemical Co. St. Louis, 288 Massachuset, USA) aqueous solutions with 10 mg/L and 100 mg/L concentrations were used as 289 feed wastewater model solutions. Dilute HA solutions were prepared from a standard HA 290 concentrated solution of 1 g/L and the pH was adjusted to 11 (Metrohm pH/Ion meter 692, Herisau, 291 Suiza) by adding 2 M hydrochloric acid (HCl) aqueous solution as needed [38]. Taking into 292 consideration that the ζ-potential of both PSf layer of the RFO membranes, and the PA and PE 293 layers of the TRFO membranes formed by IP are very electronegative at high pH [38, 60] a higher 294 feed pH was selected to minimize the adsorption phenomenon of HA. In other words, at high pH 295 values both the membrane and the HA feed solution have negative electrical charges that generate 296 electrical repulsion between them [61]. Two types of FO tests were carried out using low 297 concentrations of NaCl (65 g/L) and HA (10 mg/L) and high concentrations of NaCl (200 g/L) and 298 HA (100 mg/L) solutions. 299 The feed and permeate temperatures were measured inside each chamber by pt100 sensors 300 connected to a digital meter. Both chambers contain magnetic stirrers set in this study at 750 rpm 301 for all tests. To determine the permeate flux (𝐽) through the effective membrane area (𝐴𝑚), the 302 collected permeate (∆𝑚) was weighed in a precision balance for a predetermined time (∆𝑡) and 303 then calculated as [48, 62]: 304 18 𝐽 = 𝛥𝑚 𝐴𝑚𝛥𝑡 (7) 305 The permeate flux measurements were made in triplicate and for each membrane two samples 306 were tested for 30 min. A mean permeate flux was finally determined with its corresponding 307 standard deviation. 308 The HA and salt concentrations of the both the feed and draw solutions were measured at the 309 beginning and at the end of each experiment for all samples. From these measurements, the reverse 310 salt permeate flux (Js) was calculated using the following equation [48, 62]: 311 𝐽𝑆 = 𝐶𝑠,F,𝑡𝑉𝐹,𝑡 − 𝐶𝑠,F,0𝑉𝐹,0 𝐴𝑚𝛥𝑡 (8) 312 where 𝐶𝑠,𝐹,0 an 𝐶𝑠,𝐹,𝑡 are the initial and final feed salt concentrations, respectively; and 𝑉𝐹,0 and 313 𝑉𝐹,𝑡 are the initial and final feed volumes, respectively. During the FO experiments, the feed 314 volume was maintained constant (250 mL) by the feed reservoir (Fig. 1). 315 The HA rejection factor (RHA) was evaluated using the following equation [63]: 316 𝑅𝐻𝐴 (%) = (1 − 𝐶𝐻𝐴,𝐷,𝑡 𝐶𝐻𝐴,𝐹,0 ) · 100 (9) 317 where 𝐶𝐻𝐴,𝐷,𝑡 is the final HA concentration of the draw solution and 𝐶𝐻𝐴,𝐹,0 the initial HA 318 concentration of the feed solution. 319 Another parameter considered to evaluate the membrane performance is the HA concentration 320 factor (𝐶𝐹𝐻𝐴) defined as [64, 65]: 321 𝐶𝐹𝐻𝐴 = ( 𝐶𝐻𝐴,𝐹,𝑡 𝐶𝐻𝐴,𝐹,0 ) (10) 322 where 𝐶𝐻𝐴,𝐹,𝑡 and 𝐶𝐻𝐴,𝐹,0 are the HA concentrations of the feed solution at the initial and final of 323 the FO experiment, respectively. For sake of comparison, the total experimental time for each 324 membrane was maintained the same (30 min). Since two different HA initial concentrations were 325 19 considered in FO, 𝐶𝐹𝐻𝐴 values could reflect the effectiveness of FO membranes during the 326 concentration processes depending on the feed solutions. Furthermore, this parameter can be taken 327 into consideration to figure out whether feed contaminants could be adsorbed at the membrane or 328 not [64]. 329 To determine the concentration of NaCl and HA, both the electrical conductivity and absorbance 330 of the feed and draw solutions were measured using a conductivity meter (Metrohm 712Ω, 331 Herisau, Suiza) and a spectrophotometer (Genesis 10S UV-Vis, Thermo Scientific Inc., 332 Massachusetts, USA), respectively. It is worth quoting that the HA absorption spectrum has no 333 characteristic peak so in this study the absorbance measurements of HA were taken at 254 nm as 334 reported elsewhere [51]. The HA and NaCl concentrations were determined by comparing the 335 obtained absorbance and electrical conductivity data with appropriate calibration curves presenting 336 the absorbance or electrical conductivity as a function of the HA or NaCl concentration, 337 respectively. The effects of HA on the electrical conductivity of NaCl solutions was verified by 338 measuring the electrical conductivity of the studied saline solutions at different HA concentrations 339 (Fig. S2). Moreover, for all studied HA solutions the absorbance was measured at different NaCl 340 concentrations (Fig. S3). 341 342 3. Results and discussions 343 3.1. Characteristics of recycled forward osmosis (RFO) membranes 344 The water contact angles (w) of both the active top layer and support layer of the RFO 345 membranes together with the total thickness () and the isoelectric point (IEP) are summarized in 346 Table 3. All RFO membranes exhibit similar results regardless of the followed cleaning procedure 347 and the total dose level applied. 348 349 20 Table 3. Water contact angle (w) of the active layer (AL) and support layer (SL), thickness (δ) 350 and isoelectric point (IEP) of the RFO membranes and the commercial membranes TM720-400, 351 CTA-HTI and TFC-HTI. 352 Membrane (w)AL (º) (w)SL (º) δ (μm) IEP (pH) TM720-400 54 ± 3 64 ± 2 110 ± 3 -- RFO1 78 ± 3 65 ± 2 100 ± 6 3.55 RFO2 79 ± 2 65 ± 5 97 ± 4 3.63 RFO3 76 ± 2 69 ± 2 96 ± 5 3.96 RFO4 78 ± 2 65 ± 5 96 ± 5 3.88 CTA-HTI 60 ± 1 69 ± 6 72 ± 5 4.12 TFC-HTI 26 ± 4 N/A 83 ± 4 3.21 353 The water contact angle of the support layer of the RFO membranes is similar to that of the 354 pristine TM720-400 membrane (i.e. 64  2º). However, an increase of the water contact angle of 355 the active layer from 54º to 78º was detected for the discarded RO membranes after NaClO 356 cleaning. This value corresponds to the water contact angle of PSf layer confirming therefore the 357 removal of the PA top layer from the surface of the RFO membranes [66, 67]. The obtained results 358 indicate the low hydrophilic character of the RFO membranes compared to the pristine TM720-359 400 membrane and both CTA-HTI and TFC-HTI membranes. The measured water contact angle 360 of the commercial membranes (CTA-HTI and TFC-HTI) agree well with those reported by [68] 361 (i.e. the active layer water contact angle was 63 ± 2° and 27° ± 1 for CTA-HTI and TFC-HTI 362 membranes, respectively). The thickness of all RFO membranes is almost similar taking into 363 consideration their error intervals. Compared to the thickness of the TM720-400 pristine 364 membrane (110 ± 2 μm), the RFO membranes are thinner (96 – 100 μm) indicating the 10 to 14 365 μm PA layer thickness. 366 21 From the SEM images of the RFO membranes shown in Fig. 2, it can be seen that the surface 367 cleaning improved with the increase of the exposure time (from RFO1 to RFO2) and by adding 368 the laboratory cleaning step (RFO3 and RFO4) although the effect of this last step is not significant 369 for cleaning improvement compared to the increase of the exposure dose in the pilot plant. 370 However, this second cleaning step requires disassembling the membrane module and results in 371 an increase of the cleaning cost. In fact, the SEM images of the RFO2 and RFO4 membranes are 372 similar although the last membrane was subjected to laboratory cleaning. 373 374 Figure 2. SEM images of the PSf layer and PE support of the RFO membranes. 375 The ζ-potential and IEP of the RFO membranes and commercial membranes, CTA-HTI and 376 TFC-HTI, are presented in Fig. 3 and Table 3, respectively. The obtained values of the commercial 377 22 membranes (CTA-HTI and TFC-HTI) agreed well with those presented in previous studies [47, 378 49, 69]. The TFC-HTI membrane was more negatively charged than the CTA-HTI membrane as 379 it was observed elsewhere [70]. In general, the ζ-potential and IEP of the RFO membranes are 380 quite similar and all IEP values of the RFO membranes were between those of TFC-HTI and CTA-381 HTI. Small differences of the ζ-potential and IEP were detected between the RFO membranes. 382 The membranes previously cleaned in the laboratory (RFO3 and RFO4) exhibited slightly higher 383 values than those cleaned in the pilot plant (RFO1 and RFO2). This could be due to the existence 384 of small persistent PA residues on the membrane surface of the RFO membranes cleaned in the 385 pilot plant, which seems to be eliminated after laboratory cleaning step. 386 387 Figure 3. ζ-potential of the RFO membranes together with that of CTA-HTI and TFC-HTI 388 commercial membranes at different pH values. 389 In order to confirm if the total degradation of the PA top layer occurs, the RFO membranes were 390 characterized by ATR-FTIR and their spectra were compared to the pristine TM720-400 391 1 2 3 4 5 6 7 -30 -20 -10 0 10 20 30 Z e ta P o te n ti a l (m V ) pH RFO1 RFO2 RFO3 RFO4 CTA-HTI TFC-HTI 23 membrane spectrum. From Fig. 4, it can be seen that the peaks corresponding to the amide I and 392 amide II bands and the C = C stretching vibrations of the aromatic amide bonds (1664, 1542 and 393 1610 cm-1, respectively [41]) were clearly detected in the spectrum of the pristine TM720-400 394 membrane but not in the spectra of the RFO membranes. This indicates the degradation of the PA 395 layer. It is to be noted that no significant difference was detected between the spectra of the RFO3 396 and RFO4 membranes, while in the RFO1 and RFO2 membranes some absorbance was detected 397 in amide bands. It must be mentioned that several ATR-FTIR spectra were carried out in different 398 parts of the RFO membranes. Depending on the analysed section of the RFO1 and RFO2 399 membrane samples, the absorbance at the amide bands appeared or not. This is probably attributed 400 to PA residual left deep suggesting that the pilot plant cleaning did not completely remove the PA 401 layer of the discarded RO membranes. 402 403 Figure 4. ATR-FTIR spectra of the RFO membranes and the pristine TM720-400 commercial 404 membrane. 405 1750 1700 1650 1600 1550 1500 1450 A b s o rb a n c e Wavenumber (cm-1) TM720-400 pristine RFO1 RFO2 RFO3 RFP4 Amide I (1664 cm-1) Amide II (1542 cm-1) Aromatic amide bond (1610 cm-1) 24 Prior to solute transport experiments with PEG and PEO, the water permeance (WP) was 406 determined for each RFO membrane and the results are summarized in Table S1. By increasing 407 the cleaning exposure time (RFO1 and RFO2 membranes) or adding the laboratory cleaning step 408 (RFO1 and RFO3 membranes) the WP value was increased by 17% and 11%, respectively. 409 However, for the RFO2 membrane no enhanced WP was detected after the addition of the 410 laboratory cleaning step (i.e. both RFO2 and RFO4 membranes exhibit the same WP). It must be 411 stated that the obtained WP of the RFO2 membrane is in good agreement, within 3% deviation, 412 with that reported elsewhere for a similar membrane and for the same cleaning procedure followed 413 to prepare RFO2 membrane [41]. Similar WP values were reported for UF recycled end-of-life RO 414 membranes [13, 43]. 415 The solute separation results are also presented in Table S1. With the application of the 416 laboratory cleaning step, a slight reduction of the solute rejection factor (α) was detected. The 417 RFO1 and RFO2 membranes reached 50% rejection for PEG6, while for the same PEG the 418 rejection factor of the other membranes, RFO3 and RFO4, was lower than 50%. All RFO 419 membranes presented a 100% rejection factor when using PEO200 and between 96-99% for 420 PEO100. For all recycled membranes, an average rejection factor of 35% was obtained when using 421 PEG0.4. This high degree of separation for such low molecular weight compound indicates that 422 RFO membranes preserved NF characteristics after recycling procedure. The use of NF-like FO 423 membranes has been already reported elsewhere [71-73]. 424 In Fig. S4 the rejection factor was plotted as a function of the solute diameter according to log-425 normal probability function. The determined mean pore size (μp), its geometric standard deviation 426 (σp), pore density (N), porosity (ε) and molecular weight cut-off (MWCO) from the solute transport 427 tests, as well as the tortuosity (τ, obtained from Eq. (1)) and the “intrinsic” structural parameter 428 25 (Sint, from Eq. (2)) are summarized in Table 4. The related cumulative pore size distribution and 429 the probability density function curves are presented in Fig. S5. 430 431 Table 4. Mean pore size (μp), geometric standard deviation (σp), pore density (N), porosity (ε), 432 molecular weight cut-off (MWCO), tortuosity (τ) and “intrinsic” structural parameter factor (Sint) 433 of the RFO membranes obtained from the solute transport test. 434 Membrane μp (nm) σp (-) N (pores/μm2) ε (%) MWCO (kDa) τ (-) Sint (µm) RFO1 3.6 2.6 5037 26 33.9 2.15 824 RFO2 3.4 2.1 4910 30 29.5 1.97 631 RFO3 4.1 2.6 4446 28 39.8 2.08 725 RFO4 3.7 2.6 5467 30 38.0 1.98 629 435 Since the PA layer was removed, the obtained mean pore size of the RFO membranes were 436 larger than the typical pore size of RO membranes (< 1nm) and similar to those of NF-like 437 membranes [74]. Almost similar pore sizes were obtained when increasing the cleaning exposure 438 time in the pilot plant (i.e. reduction of less than 6% of the pore size of the membrane RFO2 439 compared to the membrane RFO1) but the MWCO was decreased 12.8%. This is attributed to the 440 reduction of the tail of the probability density function curve of the RFO2 membrane. However, 441 when comparing the RFO1 membrane with the RFO3 membrane, and the RFO2 membrane with 442 the RFO4 membrane, slight enhancements of both the pore size and the MWCO were detected 443 when the laboratory cleaning step was added after the pilot plant cleaning protocol (i.e. 13.9% and 444 8.8% increase of the pore size and 17.5% and 28.9% increase of the MWCO for the RFO3 and 445 RFO4 membranes, respectively). Note that the CTA-HTI membrane has an average pore size of 446 0.3 nm [75]. Therefore, the pore size of the RFO membranes is an order of magnitude greater than 447 26 that of the CTA-HTI membrane. A broad range of pore size (0.2 – 7 nm) for FO membranes is 448 reported in the literature [74-79]. In addition, the porosity (ε) of the RFO membranes was found 449 to be within the values expected for a typical FO membrane. A porosity of 40  3 % was reported 450 for CTA-HTI membrane [45] while porosity values between 16% and 21% were reported for the 451 PSf support of the TFC-PA RO membranes prepared with different PSf concentrations [80]. 452 The pore density decreased by 12% for the RFO3 compared to the RFO1 membrane whereas 453 that of the RFO4 membrane increased by 11% compared to the RFO2 membrane. This indicates 454 that the pore density does not show any relationship with the followed laboratory cleaning 455 protocol. The tortuosity was estimated from Eq. (1) in order to determine the “intrinsic” or 456 geometrical structural parameter, Sint. A slightly lower tortuosity was obtained for the recycled 457 RFO membranes (RFO2 and RFO4) applying the lower exposure dose in pilot plant cleaning. A 458 tortuosity factor of 4.69 was reported in [45] for the commercial membrane CTA-HTI using the 459 structural parameter (620 m) estimated by RO tests and Eq. (2) . If Eq. (1) were used to estimate 460 the tortuosity factor from the value of the porosity ( 40%) also reported in [45], an estimated lower 461 value of 1.69 would be obtained. This indicates that the applied model gives tortuosity values 462 lower than the estimated from RO test and, as a consequence, lower Sint values are obtained. 463 Membranes with low Sint values are preferred for FO membranes in order to reduce the ICP effect. 464 Therefore, the ratio τ/ must be as low as possible. In general, to decrease the structural parameter 465 of the TFC-FO membranes, a thin support layer with a high porosity should be used. The porosity 466 of the support layer of a given FO membrane exerts a significant effect on both J and Js (i.e. both 467 permeate fluxes increase with the increase of the porosity). For active layers with low porosity 468 values, J and Js are more sensitive to the support layer porosity [81]. It must be pointed out that 469 increasing porosity or decreasing tortuosity is not always desirable since increasing J also increases 470 Js. 471 27 3.2. FO experiments of recycled forward osmosis (RFO) membranes 472 The RFO membranes and the two commercial membranes (CTA-HTI and TFC-HTI) were tested 473 in FO using HA aqueous solutions as feed and NaCl aqueous solutions as draw solutions (i.e. low 474 concentration: 10 mg/L HA aqueous solution and 65 g/L NaCl aqueous solution; and high 475 concentration: 100 mg/L HA aqueous solution and 200 g/L NaCl aqueous solution). The results 476 represented in Figs. 5 and 6. 477 The permeate flux (J) of each RFO membrane is quite similar for both the high (3.3-7.2 kg/m2·h) 478 and low (3.2-6.3 kg/m2·h) concentrations tests, although slightly greater values could be detected 479 for the high concentration test because of the enhanced osmotic pressure of the draw solution with 480 the increase of the NaCl concentration (i.e. FO driving force) [82]. Longer exposure time during 481 cleaning in the pilot plant (RFO1 and RFO2 membranes) improves the permeate flux. However, 482 the addition of the laboratory cleaning step (RFO1 and RFO3 membranes, RFO2 and RFO4 483 membranes) does not provide any significant enhancement of the permeate flux, but it increases 484 slightly the specific reverse salt flux Js/J. This could be due to the existence of PA residues in the 485 PSf pores of RFO1 and RFO2 membranes (Fig.4) reducing the reverse draw solute transport (Js) 486 while increasing slightly the water transport (J) and rendering the membrane semi-permeable. 487 488 28 489 490 Figure 5. a) FO permeate flux (J) and b) specific reverse salt flux (Js/J) of the RFO membranes 491 and commercial membranes (CTA-HTI and TFC-HTI) for low concentration (10 mg/L HA feed 492 aqueous solution and 65 g/L NaCl draw aqueous solution) and high concentration (100 mg/L HA 493 feed aqueous solution and 200 g/L NaCl draw aqueous solution). 494 The specific reverse salt flux (Js/J) is a quantitative factor that indicates the bi-directional 495 diffusion in the FO process. Higher (Js/J) reflects a decrease of the selectivity and efficiency of the 496 RFO1 RFO2 RFO3 RFO4 CTA-HTI TFC-HTI 0 5 10 15 20 J ( k g /m 2 h ) Low concentration High concentration a) RFO1 RFO2 RFO3 RFO4 CTA-HTI TFC-HTI 0.00 0.01 0.02 0.03 0.11 0.12 0.13 0.14 Low concentration High concentration J S /J b) 29 membrane. As can be seen in Fig. 5b, Js/J depends on the NaCl concentration of the draw solution. 497 For a higher NaCl concentration, a greater Js/J value was obtained because the osmotic pressure 498 (i.e. the FO driving force) is higher. Similar results were reported in [83]. In order to compare the 499 draw solute loss due to the reverse salt flux for both high and low concentrations, a normalized 500 reverse salt transfer (RTs) was defined as the ratio of the NaCl concentration transported through 501 the FO membrane to the feed side and the initial draw concentration (Section S4, Fig. S6a). For 502 the low concentration test, the RFO membranes cleaned only in the pilot plant (RFO1 and RFO2 503 membranes) have significantly lower RTs values than the other RFO membranes cleaned also 504 following the laboratory scale protocol (RFO3 and RFO4 membranes), confirming therefore that 505 small PA residues in the PSf pores of RFO1 and RFO2 membranes could prevent reverse draw 506 solute transport reducing both the RTs and Js values. 507 The best FO performance (i.e. high J, low Js/J and low RTs) for the low concentration test was 508 obtained with the RFO2 membrane. However, this membrane presents 35% and 65% lower 509 permeate flux than those of the commercial membranes CTA-HTI and TFC-HTI, respectively. 510 The Js/J of the RFO2 membrane is smaller than that of the commercial membrane CTA-HTI and 511 comparable with the best result obtained for the commercial membrane TFC-HTI. For the high 512 concentration test, the RFO2 membrane also presents the highest J flux among all RFO 513 membranes. In general, the above results show that the RFO2 membrane is optimal for both high 514 and low concentrations tests. 515 The observed lower J values of the RFO membranes compared to the commercial ones (CTA-516 HTI and TFC-HTI) may be due to their lower surface porosity and pore density as well as to their 517 lower hydrophilic character as discussed earlier. In addition, the highest permeate flux of the TFC-518 HTI membrane compared to the CTA-HTI membrane may be due to its lower electronegativity at 519 pH 11 that results in greater electrostatic repulsive forces to HA favouring therefore water 520 30 permeation through the membrane. The obtained J and Js/J data of the membranes TFC-HTI and 521 CTA-HTI for low concentration test were found to be within the expected values reported in the 522 literature (i.e. less than 10% deviation) although different experimental systems and FO operating 523 conditions were considered [46, 48, 54, 70]. 524 The humic acid rejection factor (RHA) of both low and high concentration tests are plotted in Fig. 525 6a. For high concentration test, the RHA of all RFO membranes is at least 99.3 %, which is higher 526 than that of the commercial membranes (TFC-HTI, 98.9 % and CTA-HTI 97.7 %). The RFO2 527 membrane has the highest RHA value for the low concentration test, which is 0.6% and 1.9% higher 528 than that of the commercial membranes TFC-HTI and CTA-HTI, respectively. It can be seen from 529 the RHA data of the RFO1 and RFO3 membranes, as well as those of the RFO2 and RFO4 530 membranes, that the laboratory cleaning step does not improve RHA. This also supports that the 531 existence of the PA residues in the PSf pores retards the diffusion of the HA from the feed to the 532 draw solution [22]. 533 The calculated HA concentration factor (𝐶𝐹𝐻𝐴) of all membranes by Eq. (10) at 30 min running 534 time is shown in Fig. 6b for both low and high concentration tests. The 𝐶𝐹𝐻𝐴 factor takes into 535 account both J and RHA results (i.e. sufficiently high J and RHA values results in a high 𝐶𝐹𝐻𝐴 factor). 536 For all membranes, it was observed lower 𝐶𝐹𝐻𝐴 factor for the high concentration test than for the 537 low concentration test. This is because 𝐶𝐹𝐻𝐴 is inversely proportional to the initial HA 538 concentration of the feed solution although both J and RHA are greater for the high concentration 539 value than for the low concentration test. 540 Among all tested membranes, the commercial membrane TFC-HTI exhibits the best 𝐶𝐹𝐻𝐴 541 factors while the CTA-HTI membrane shows the worst factors. For the high concentration test, the 542 RFO membranes also show comparable values to those of the membrane TFC-HTI. Again, the 543 RFO membranes cleaned only in the pilot plant (RFO1 and RFO2 membranes) have slightly better 544 31 𝐶𝐹𝐻𝐴 factors than the other RFO membranes cleaned also following laboratory protocol (RFO3 545 and RFO4 membranes). 546 547 548 Figure 6. HA rejection factor (RHA) (a) and HA concentration factor (CFHA) at 30 min (b) of the 549 RFO membranes and commercial CTA-HTI and TFC-HTI membranes for low concentration (10 550 mg/L HA feed aqueous solution and 65 g/L NaCl draw aqueous solution) and high concentration 551 (100 mg/L HA feed aqueous solution and 200 g/L NaCl draw aqueous solution). 552 RFO1 RFO2 RFO3 RFO4 CTA-HTI TFC-HTI 96.5 97.0 97.5 98.0 98.5 99.0 99.5 100.0 R H A ( % ) Low concentration High concentration a) RFO1 RFO2 RFO3 RFO4 CTA-HTI TFC-HTI 0.8 0.9 1.0 1.1 1.2 1.3 C F H A Low concentration High concentration b) 32 In general, the RFO2 membrane cleaned only in pilot plant with an exposure dose level of 106 553 ppm·h presents among all RFO membranes the best FO results (Table S2). This membrane exhibits 554 reasonably high water permeate fluxes compared to the FO commercial membranes and much 555 lower permeate fluxes ratio (Js/J) than those of the CTA-HTI membrane. Note that in pilot plant 556 cleaning procedure, it is not necessary to disassemble the membrane module allowing therefore 557 the reuse of the complete RO discarded membrane in FO after an appropriate modification. 558 Therefore, as explained previously in section 2.2, this membrane was selected as support for the 559 preparation of TFC-RFO membranes by interfacial polymerization (IP) in order to improve further 560 the FO performance. 561 562 3.3. Characteristics of TRFO membranes 563 The water contact angle of the active and support layers of the prepared TRFO membranes are 564 summarized in Table 5 together with their total thickness. The water contact angle of the support 565 layer of all TRFO membranes is the same as that of the RFO2 membrane (i. e. 65  5º). However, 566 a decrease of the water contact angle of the active layer from 79º to 55-29º was detected after the 567 surface modification of the RFO2 membrane by IP showing their improved hydrophilic character. 568 Taking into consideration the standard deviation of the measured water contact angles, all TRFO 569 membranes with the new PA active layer presented similar water contact angle values (i.e. a mean 570 value of 32 ± 2 º) clearly lower than that of the new PE active layer (55 ± 5 º). The lower 571 hydrophilic character of PA compared to PE was already observed in other studies [38]. The total 572 thickness of all TRFO membranes can also be considered the same taking into account the 573 associated errors, even it was maintained the same as that of the RFO2 membrane (97 ± 4 m). 574 Therefore, a very thin active (PA or PE) layer can be assumed. 575 33 Table 5. Water contact angle (w) of both the active layer (AL) and support layer (SL) and 576 thickness of the TRFO membranes. 577 Membrane (w)AL (º) (w)SL (º) δ (μm) RFO-MPD 31 ± 5 65 ± 4 98 ± 2 RFO-MPD/TEA 33± 5 65 ± 4 99 ± 4 RFO-PIP 35 ± 2 66 ± 4 97 ± 3 RFO-PIP/TEA 29 ± 6 65 ± 5 98 ± 4 RFO-PIP/PVA 29 ± 3 63 ± 3 96 ± 5 RFO-BPA 55 ± 5 64 ± 3 97 ± 5 578 Fig. 7 shows the SEM images of the top surface (PA or PE active layer) of the TRFO membranes. 579 A PA formed layer on the top surface of the RFO2 substrate was observed and as a consequence, 580 the resultant TRFO membranes exhibited a dense top surface. Noticeable morphological 581 differences of the active layers (PA with MPD, PA with PIP or PE) of the TRFO membranes were 582 detected. The formed PA layer using MPD presented the typical characteristic of an interfacial 583 polymerized PA membrane consisting of ‘‘ridge-and-valley’’ morphological structure [77, 79, 84, 584 85]. It must be noted that TFC-FO membranes are commonly prepared with MPD as aqueous 585 phase in the IP process. The surface morphology of the transformed membranes by PIP (RFO-PIP, 586 RFO-PIP/TEA and RFO-PIP/PVA) were also rough with granular-like-structure morphology. 587 Using TEA or PVA in the aqueous phase together PIP induced a smooth granular top surface. 588 When PE layer was formed on the top surface of the RFO2 substrate, a less dense layer was 589 observed. This could be attributed to the fact that PA residual in the pores of PSf substrate of the 590 RFO2 membrane facilitated the penetration of the aqueous phase (MPD or PIP) into the substrate 591 but hindered BPA that was used to form PE layer. Chi et al. [85] prepared a TFC FO membrane 592 supported by polyimide (PI) microporous nanofiber membrane using two IP steps. In the first IP 593 34 step, only few amounts of PA could be deposited on the nanofibers without forming a PA layer. 594 These PA deposits facilitated the formation of a dense and thin PA layer in the second IP step. 595 596 Figure 7. Top-surface SEM images (active IP layer) of the TRFO membranes. 597 35 The obtained ATR-FTIR spectra of the TRFO membranes and that of the RFO2 membrane are 598 presented in Fig. 8. All PA modified TRFO membranes showed an absorption peak corresponding 599 to the C = C stretching vibrations of the aromatic amide bonds at 1610 cm-1 (Fig. 8a and 8b). The 600 amide I and amide II absorbance peaks at 1664 and 1542 cm-1, respectively, were clearly detected 601 for the RFO-MPD and RFO-MPD/TEA membranes as presented in Fig. 8a. However, these peaks 602 were not detected for the transformed membranes by PIP (RFO-PIP, RFO-PIP/PVA and RFO-603 PIP/TEA) in Fig. 8b since the amine salts were consumed as catalyst and was not integrate into 604 the poly(piperazine-amide) layer [86]. The broad peak centred at 3313 cm–1 for the membranes 605 RFO-MPD and RFO-MPD/TEA or at 3440 cm-1 for the membranes RFO-PIP, RFP-PIP/TEA and 606 RFO-PIP/PVA was due to O–H stretching that could arise from the partial hydrolysis of the acyl 607 chloride unit of TMC [87, 88]. 608 The spectrum of the membrane RFO-BPA (Fig. 8c) shows an absorption peak at 1720 cm-1, 609 which correspond to C=O stretching vibrations of the ester groups. This absorption peak was not 610 detected in the spectra of the RFO2 membrane indicating the correct formation of the PE thin layer 611 [89]. Similarly, a broad adsorption peak appeared at 3393 cm-1 due to the asymmetry stretching 612 vibration of the hydroxyl groups (-OH) [90] that could arise from the unreacted hydroxyl groups 613 of BPA in the membrane or from the partial hydrolysis of the acyl chloride unit of TMC. In general, 614 it can be confirmed the correct formation of both the PA and PE thin layers on the top surface of 615 the RFO2 membrane and the successful preparation of TRFO membranes. 616 36 617 618 619 620 Figure 8. ATR-FTIR spectra of the TRFO membranes: a) RFO-MPD and RFO-MPD/TEA, b) 621 RFO-PIP, RFO-PIP/TEA and RFO-PIP/PVA and c) RFO-BPA, together the spectra of the RFO2 622 membrane. 623 3500 3000 1800 1750 1700 1650 1600 1550 1500 A b s o rb a n c e Wavenumber (cm-1) RFO-MPD/TEA RFO-MPD RFO2 a) 1664 cm-1 1610 cm-1 1542 cm-1 3313 cm-1 3500 3000 1800 1750 1700 1650 1600 1550 1500 A b s o rb a n c e Wavenumber (cm-1) RFO-PIP/PVA RFO-PIP/TEA RFO-PIP RFO2 b) 1610 cm-1 3440 cm-1 3500 3000 1800 1750 1700 1650 1600 1550 1500 A b s o rb a n c e Wavenumber (cm-1) RFO-BPA RFO2 c) 1720 cm-1 3393 cm-1 37 Table 6 summarizes the transport parameters A, B and the “effective” structural parameter, Seff, 624 calculated from the model developed in [57] as stated earlier in section 2.5.1. Lower Seff values 625 were obtained for the commercial CTA-HTI and TFC-HTI membranes compared with those 626 reported in the literature using RO test to determine A and B parameters for the same membranes. 627 The calculated average Seff factor of the CTA-HTI membrane obtained by means of RO test in 628 other studies is 520  163 m [45, 46, 48, 49, 54, 91-93]. For the TFC-HTI membrane, the reported 629 Seff value was 533 µm [70]. It was also observed that Seff of the CTA-HTI membrane obtained by 630 means of FO test varied with the type, concentration and temperature of the draw and feed 631 solutions in the range between 200 and 500 m [92]. Moreover, the determined Seff for the CTA-632 HTI membrane based on FO test was found to be 498 ± 37 µm [57]. The Seff value, 402 µm, 633 obtained in our study for this membrane applying FO test is close to these values confirming 634 therefore the validity of the followed procedure. 635 As it was expected, the RFO2 membrane exhibited higher Seff value than that of the TRFO 636 membranes. In addition, all TRFO membranes exhibited lower Seff values than that of the CTA-637 HTI commercial membrane. The lowest Seff values were obtained for the membranes RFO-MPD 638 and RFO-MPD/TEA (i.e. 269 m). The A value of all TRFO membranes was slightly greater than 639 that of the membrane RFO2. As mentioned previously, for TFC-FO membranes both the selective 640 IP layer as well as the support are key factors determining their FO performance. Typically, the 641 active layer accounts for the solute rejection, while the support dominates the permeate fluxes 642 (both J and Js) because of the ICP effects [22]. Among all tested membranes, the RFO-MPD/TEA 643 membrane was found to have the highest A value (0.42 L/m2 h bar) whereas the lowest B value 644 was obtained for the membrane RFO-MPD, showing both desirable characteristics for FO 645 performance. The results of the solute permeance (B) of the TRFO membranes indicated that the 646 IP increased B value except for the two membranes RFO-MPD and RFO-BPA. 647 38 648 Table 6. Water permeance (A), solute permeance (B) coefficients and “effective” structural 649 parameter (Seff) of the TRFO membranes together with the RFO2 membrane and commercial 650 membranes (CTA-HTI and TFC-HTI). 651 Membrane A (L/m2h bar) B (L/m2h) Seff (µm) RFO-MPD 0.33 0.15 269 RFO-MPD/TEA 0.42 0.26 269 RFO-PIP 0.27 0.56 362 RFO-PIP/TEA 0.31 0.79 322 RFO-PIP/PVA 0.30 0.88 374 RFO-BPA 0.33 0.17 297 RFO2 0.23 0.17 551 CTA-HTI 0.32 0.92 402 TFC-HTI 0.62 0.057 245 652 653 3.4. FO experiments of TRFO membranes 654 Similar to RFO membranes, the TRFO membranes were also tested in FO to treat HA aqueous 655 solutions following the same procedure. The results are plotted in Fig. 9 together with those of the 656 membrane RFO2 and the commercial membranes CTA-HTI and TFC-HTI. 657 658 39 659 660 Figure 9. a) FO permeate flux (J) and b) the specific reverse salt flux (Js/J) of the TRFO 661 membranes together with those of the membrane RFO2 and the commercial membranes (CTA-662 HTI and TFC-HTI) for low concentration (10 mg/L HA feed aqueous solution and 65 g/L NaCl 663 draw aqueous solution) and high concentration (100 mg/L HA feed aqueous solution and 200 g/L 664 NaCl draw aqueous solution). 665 R F O -M P D R F O -M P D /T E A R F O -P IP R F O -P IP /T E A R F O -P IP /P V A R F O -B P A R F O 2 C T A -H T I T F C -H T I 0 5 10 15 20 Low concentration High concentration J ( k g /m 2 h ) a) R F O -M P D R F O -M P D /T E A R F O -P IP R F O -P IP /T E A R F O -P IP /P V A R F O -B P A R F O 2 C T A -H T I T F C -H T I 0.00 0.01 0.02 0.03 0.11 0.12 0.13 0.14 J S /J Low concentration High concentration b) 40 The permeate flux (J) of all TRFO membranes, for both low and high concentration tests, were 666 found to be higher than that of the membrane RFO2. This is attributed to the effects of the prepared 667 IP layer of PA or PE, which reduces the water contact angle increasing its hydrophilic character, 668 and as a consequence, decreasing the structural parameter, Seff. This increase in J with respect to 669 the non-transformed RFO membrane confirms the successful IP coating layer since the membrane 670 becomes more semi-permeable. 671 All TRFO membranes exhibited a higher J value for high concentration test (10.8-15.1 kg/m2·h) 672 than for low concentration test (8.3-12.2 kg/m2·h). This was also observed for the RFO membranes 673 and the commercial membranes. As stated earlier, this is due to the enhancement of the osmotic 674 pressure with the increase of the salt concentration of the draw solution. The permeate flux of the 675 PA modified TRFO membranes using MPD alone or mixed with TEA was better than that of the 676 PA modified membranes using PIP or its mixture with TEA or PVA. This is because the use of 677 the MPD monomer in IP process provides very good transport properties as well as antifouling 678 characteristics making it one of the most used monomers in FO membrane synthesis [37, 94]. In 679 addition, using either MPD or the mixture MPD/TEA results in a significant reduction of Js/J 680 compared to PIP or its mixture with TEA or PVA. 681 The PE modified membrane (RFO-BPA) showed J values lower than those of the RFO-MPD or 682 RFO-MPD/TEA membranes, but higher than those of the of the TRFO membranes modified using 683 PIP or its mixed with TEA or PVA. Likewise, the Js/J values of the RFO-BPA membrane was 684 similar to that of the TRFO-MPD membrane and slightly higher than that of the TRFO-MPD/TEA 685 membrane, but much lower than that of the TRFO membranes modified using PIP. 686 In order to compare the draw solute loss (i.e. reduction of the driving force) due to the reverse 687 salt flux for both high and low concentrations, a normalized reverse salt transfer (RTs) was 688 determined (Fig. S6b). No significant difference in the obtained RTs values was detected for all 689 41 TRFO membranes although these values were lower than those of the RFO2 membrane, and even 690 lower than those of TFC-HTI membrane. 691 Among all TRFO membranes, the best FO performance, for both low and high concentration 692 tests, was achieved by the membrane RFO-MPDA/TEA (i.e. (high J and low Js/J). Compared to 693 the commercial membrane CTA-HTI, this membrane showed a competitive behaviour since its 694 permeate flux for low concentration test was 30% higher, and the Js/J value was 97% lower for 695 both low and high concentration tests. Phuntsho et al. [95] compared both J and Js/J of FO and RO 696 membranes in the FO process of several feed and draw solutions. The J values for CTA-HTI 697 membrane was regularly more than 20 times higher than that of the RO membrane, while the Js/J 698 values of CTA-HTI were also significantly higher by several orders of magnitude than that of the 699 RO membrane. Therefore, the TRFO membranes considerably improved the FO performance of 700 the RO membranes and could depict a good alternative to reuse end-of-life RO membranes. 701 The obtained RHA values of the TRFO membranes for both low and high concentration tests are 702 plotted in Fig. 10a. For the low concentration test, all performed surface modifications of the RFO2 703 membrane, did not improve the RHA value. In this case, the high RHA values obtained for the 704 membranes RFO-MPD and RFO-MPD/TEA were lower than those of the membranes RFO2 and 705 TFC-HTI but higher than that of the CTA-HTI membrane. For high concentrations, these two 706 TRFO membranes together with the membrane RFO-PIP/PVA presented RHA results similar to 707 those of the membrane RFO2 and superior to those of the commercial membranes CTA-HTI and 708 TFC-HTI. These results showed greater affinity of the IP thin layers to HA regardless of the 709 monomers used than the PSf layer of the RFO membranes (i.e. RFO2 membrane in this case). This 710 effect is more noticeable for the low concentration test than for the high concentration test. 711 712 42 713 714 Figure 10. HA rejection factor (RHA) (a) and HA concentration factor (CFHA) at 30 min (b) of the 715 TRFO membranes together with the those of the membrane RFO2 and the commercial membranes 716 (CTA-HTI and TFC-HTI) for low concentration (10 mg/L HA feed aqueous solution and 65 g/L 717 NaCl draw aqueous solution) and high concentration (100 mg/L HA feed aqueous solution and 718 200 g/L NaCl draw aqueous solution). 719 R F O -M P D R F O -M P D /T E A R F O -P IP R F O -P IP T E A R F O -P IP /P V A R F O -B P A R F O 2 C T A -H T I T F C -H T I 94 96 98 100 R H A ( % ) Low concentration High concentration a) R F O -M P D R F O -M P D /T E A R F O -P IP R F O -P IP T E A R F O -P IP /P V A R F O -B P A R F O 2 C T A -H T I T F C -H T I 0.8 0.9 1.0 1.1 1.2 1.3 C F H A Low concentration High concentration b) 43 The calculated 𝐶𝐹𝐻𝐴 values at 30 min FO tests of the TRFO membranes are shown in Fig. 10b 720 together with those of the membrane RFO2 and the commercial membranes CTA-HTI and TFC-721 HTI. As observed for the other RFO membranes (Fig. 6b), the 𝐶𝐹𝐻𝐴 value of the TRFO membranes 722 decreases with the increase of both HA and salt concentrations. As stated previously, this is due 723 partly to the fact that 𝐶𝐹𝐻𝐴 is inversely proportional to the initial HA concentration of the feed 724 solution and both J and RHA are not high enough to enhance 𝐶𝐹𝐻𝐴 over that of the low concentration 725 test. The highest 𝐶𝐹𝐻𝐴 was obtained using the membrane TFC-HTI for the low concentration test 726 and the membrane RFO2 for the high concentration test. All TRFO membranes show lower 𝐶𝐹𝐻𝐴 727 values than that of the membrane RFO2 being that of the membrane RFO-MPD/TEA the highest 728 among all TRFO membranes. This decrease in 𝐶𝐹𝐻𝐴 value is due to the increase of HA transport 729 through TRFO membranes, obtaining lower RHA values. 730 For sake of comparison, Table 7 summarizes both J and Js fluxes reported in other studies for 731 different FO membranes used in wastewater treatment with different concentrations of salt as draw 732 solutions. It can be seen that the RFO2 membrane exhibits reasonably good FO performance and 733 the PA modified membrane, RFO-MPD/TEA, exhibits outstanding FO performance compared to 734 other FO membranes. Compared to the previously reported data for different FO membranes used 735 for wastewater treatment, the TRFO membranes proved to be competitive rendering the reuse of 736 discarded RO membrane modules feasible for FO technology extending as consequence their life 737 cycle. 738 739 44 Table 7. FO fluxes, J and Js, reported for different membranes used in wastewater treatment with 740 different concentrations of salt as draw solutions and different FO configurations. 741 Feed Solution Draw solution Membrane material J (kg/m2·h) Js (kg/m2·h) Ref. Wastewater model (10 mg/L HA) 65 g/L NaCl RFO2 6.2 0.009 This study Wastewater model (100 mg/L HA) 200 g/L NaCl RFO2 7.2 0.042 This study Wastewater model (10 mg/L HA) 65 g/L NaCl RFO-MPD/TEA 12.21 0.006 This study Wastewater model (100 mg/L HA) 200 g/L NaCl RFO-MPD/TEA 15.12 0.014 This study Municipal wastewater synthetic seawater CTA 7.95 0.004 [96] Textile wastewater 35 g/L NaCl CTA 6.5 -- [20] Wastewater model (50 mg/L HA) 30 g/L NaCl CA 5.1 -- [69] Dairy wastewater pretreated 60 g/L NaCl TFC-PA 8.4 -- [97] Municipal wastewater synthetic seawater TFC-PA 15.1 -- [98] 742 743 4. Conclusions 744 According to the main waste management principles, membrane recycling should be considered 745 as an environmental action to enhance the sustainability of membrane technology and minimize 746 the ecological impact. The main aim of this study was to recycle and transform end-of-life RO 747 membranes in FO membranes at lab scale for wastewater treatment. The obtained morphological 748 and structural characteristics of the recycled FO membranes (RFO) and transformed RFO 749 membranes (TRFO) by interfacial polymerization indicated their validity and suitability for FO 750 process. Their intrinsic structural parameters were found to be quite similar to those of commercial 751 FO membranes. 752 45 The highest FO water permeate fluxes for RFO membranes were obtained for the RFO2 753 membrane recycled with the highest NaClO exposure dose applied in pilot plant (106 ppm·h) 754 indicating that it is not necessary a second laboratory cleaning step. The permeate fluxes of the 755 RFO2 membrane were comparable to those of the commercial CTA-HTI membrane operated 756 under the same FO conditions. 757 Among all prepared TRFO membranes, the PA surface modified membranes using MPD or its 758 mixture with TEA showed better transport properties than those prepared using PIP or its mixture 759 with TEA or PVA, and PE surface modified membrane using BPA. The best FO performance was 760 achieved by the membrane RFO-MPD/TEA (J = 15.12 kg/m2h, Js = 0.014 kg/m2h, and RHA greater 761 than 99%). These values were found to be superior to those of the commercial membrane CTA-762 HTI membrane and comparable to those the commercial membrane TFC-HTI. 763 It is demonstrated that it is possible to use discarded RO membrane modules in FO separation 764 technology after an adequate treatment procedure. This extends their lifetime contributing 765 therefore to a circular economy and sustainability in membrane science and related materials. 766 However, this research area is relatively new and further studies are necessary such as its 767 environmental impact using Life Cycle Assessment (LCA) and cost effectiveness analysis as well 768 as the possibility to apply IP without disassembling the RO membrane modules. 769 770 Declaration of competing interests 771 The authors declare that they have no known competing financial interests or personal 772 relationships that could have appeared to influence the work reported in this paper. 773 774 775 776 46 Appendix A. Supporting Information. 777 The Supporting Information is available free of charge. S1. Cross-section SEM images of 778 commercial FO and RFO2 membranes; S2. Effects of HA on the electrical conductivity of NaCl 779 solutions and of NaCl on the absorbance of HA solutions; S3. Solute transport tests using 780 polyethylene glycol (PEG) and polyethylene oxide (PEO) of different molecular weights; S4. 781 Reverse salt transfer factor (RTs) of the RFO and TRFO membranes; S5. FO results of the best 782 RFO membrane (RFO2) subjected to a passive NaClO cleaning process in the pilot plant under a 783 level dose of 106 ppm·h. S6. Study of passive cleaning under laboratory protocol with a total 784 NaClO exposure dose of 3·105 ppm·h. (PDF) 785 786 Acknowledgments 787 The authors gratefully acknowledge the financial support of the Ministry of Economy and 788 Competitiveness of Spain (CTM2015-65348-C2-2-R) and the Ministry of Science, Innovation and 789 Universities of Spain (RTI2018-096042-B-C22). 790 791 REFERENCES 792 [1] UNWWAP, United Nations World Water Assessment Programme. The United Nations World Water 793 Development Report 2015: Water for a Sustainable World, Paris, UNESCO, (2015). 794 [2] A. Yusuf, A. Sodiq, A. Giwa, J. Eke, O. Pikuda, G. De Luca, J.L. Di Salvo, S. Chakraborty, A review of 795 emerging trends in membrane science and technology for sustainable water treatment, J. Clean. Prod., 796 266 (2020) 121867. 797 [3] G.W.I. IDA, IDA Desalination Yearbook, in: Ed. Media Analytics Ltd, UK, 2017. 798 [4] J.A. Sanmartino, M. Khayet, M.C. García-Payo, H. El-Bakouri, A. Riaza, Treatment of reverse osmosis 799 brine by direct contact membrane distillation: Chemical pretreatment approach, Desalination, 420 (2017) 800 79-90. 801 [5] W. Lawler, Z. Bradford-Hartke, M.J. Cran, M. Duke, G. Leslie, B.P. Ladewig, P. Le-Clech, Towards new 802 opportunities for reuse, recycling and disposal of used reverse osmosis membranes, Desalination, 299 803 (2012) 103-112. 804 [6] L.F. Greenlee, D.F. Lawler, B.D. Freeman, B. Marrot, P. Moulin, Reverse osmosis desalination: Water 805 sources, technology, and today's challenges, Water Res., 43 (2009) 2317-2348. 806 [7] J. Landaburu-Aguirre, R. García-Pacheco, S. Molina, L. Rodríguez-Sáez, J. Rabadán, E. García-Calvo, 807 Fouling prevention, preparing for re-use and membrane recycling. Towards circular economy in RO 808 desalination, Desalination, 393 (2016) 16-30. 809 47 [8] EPC, European Commisssion. Directive 2008/98/EC of the European Parliament and of the Council of 810 19 November 2008 on waste and repealing certain directives. doi:2008/98/EC.; 32008L0098., in, Off. J. 811 Eur. Union L13 2008, pp. 3-30. 812 [9] E. Coutinho de Paula, M.C. Santos Amaral, Environmental and economic evaluation of end-of-life 813 reverse osmosis membranes recycling by means of chemical conversion, J. Clean. Prod., 194 (2018) 85-814 93. 815 [10] M. Pontié, S. Awad, M. Tazerout, O. Chaouachi, B. Chaouachi, Recycling and energy recovery solutions 816 of end-of-life reverse osmosis (RO) membrane materials: A sustainable approach, Desalination, 423 (2017) 817 30-40. 818 [11] R. García-Pacheco, J. Landaburu-Aguirre, S. Molina, L. Rodríguez-Sáez, S.B. Teli, E. García-Calvo, 819 Transformation of end-of-life RO membranes into NF and UF membranes: Evaluation of membrane 820 performance, J. Membr. Sci., 495 (2015) 305-315. 821 [12] J.M. Gohil, A.K. Suresh, Chlorine attack on reverse osmosis membranes: Mechanisms and mitigation 822 strategies, J. Membr. Sci., 541 (2017) 108-126. 823 [13] W. Lawler, A. Antony, M. Cran, M. Duke, G. Leslie, P. Le-Clech, Production and characterization of UF 824 membranes by chemical conversion of used RO membranes, J. Membr. Sci., 447 (2013) 203-211. 825 [14] J. Morón-López, L. Nieto-Reyes, S. Aguado, R. El-Shehawy, S. Molina, Recycling of end-of-life reverse 826 osmosis membranes for membrane biofilms reactors (MBfRs). Effect of chlorination on the membrane 827 surface and gas permeability, Chemosphere, 231 (2019) 103-112. 828 [15] A. Lejarazu-Larrañaga, S. Molina, J.M. Ortiz, R. Navarro, E. García-Calvo, Circular economy in 829 membrane technology: Using end-of-life reverse osmosis modules for preparation of recycled anion 830 exchange membranes and validation in electrodialysis, J. Membr. Sci., 593 (2020) 117423. 831 [16] J. Senán-Salinas, R. García-Pacheco, J. Landaburu-Aguirre, E. García-Calvo, Recycling of end-of-life 832 reverse osmosis membranes: Comparative LCA and cost-effectiveness analysis at pilot scale, Resour. 833 Conserv. Recycl., 150 (2019) 104423. 834 [17] T.-S. Chung, L. Luo, C.F. Wan, Y. Cui, G. Amy, What is next for forward osmosis (FO) and pressure 835 retarded osmosis (PRO), Sep. Purif. Technol., 156, Part 2 (2015) 856-860. 836 [18] N. Akther, A. Sodiq, A. Giwa, S. Daer, H.A. Arafat, S.W. Hasan, Recent advancements in forward 837 osmosis desalination: A review, Chem. Eng. J., 281 (2015) 502-522. 838 [19] K. Lutchmiah, A.R.D. Verliefde, K. Roest, L.C. Rietveld, E.R. Cornelissen, Forward osmosis for 839 application in wastewater treatment: A review, Water Res., 58 (2014) 179-197. 840 [20] J. Korenak, C. Hélix-Nielsen, H. Bukšek, I. Petrinić, Efficiency and economic feasibility of forward 841 osmosis in textile wastewater treatment, J. Clean. Prod., 210 (2019) 1483-1495. 842 [21] T.Y. Cath, A.E. Childress, M. Elimelech, Forward osmosis: Principles, applications, and recent 843 developments, J. Membr. Sci., 281 (2006) 70-87. 844 [22] M. Mohammadifakhr, J. de Grooth, H.D.W. Roesink, A.J.B. Kemperman, Forward Osmosis: A Critical 845 Review, Processes, 8 (2020) 404. 846 [23] M. Xie, L.D. Nghiem, W.E. Price, M. Elimelech, Comparison of the removal of hydrophobic trace 847 organic contaminants by forward osmosis and reverse osmosis, Water Res., 46 (2012) 2683-2692. 848 [24] R. Valladares Linares, Z. Li, S. Sarp, S.S. Bucs, G. Amy, J.S. Vrouwenvelder, Forward osmosis niches in 849 seawater desalination and wastewater reuse, Water Res., 66 (2014) 122-139. 850 [25] D.L. Shaffer, J.R. Werber, H. Jaramillo, S. Lin, M. Elimelech, Forward osmosis: Where are we now?, 851 Desalination, 356 (2015) 271-284. 852 [26] Y. Zhou, M. Huang, Q. Deng, T. Cai, Combination and performance of forward osmosis and membrane 853 distillation (FO-MD) for treatment of high salinity landfill leachate, Desalination, 420 (2017) 99-105. 854 [27] S. Lee, Y. Kim, S. Hong, Treatment of industrial wastewater produced by desulfurization process in a 855 coal-fired power plant via FO-MD hybrid process, Chemosphere, 210 (2018) 44-51. 856 48 [28] P.S. Goh, A.F. Ismail, A review on inorganic membranes for desalination and wastewater treatment, 857 Desalination, 434 (2018) 60-80. 858 [29] A.J. Ansari, F.I. Hai, W.E. Price, J.E. Drewes, L.D. Nghiem, Forward osmosis as a platform for resource 859 recovery from municipal wastewater - A critical assessment of the literature, J. Membr. Sci., 529 (2017) 860 195-206. 861 [30] S.M. Iskander, S. Zou, B. Brazil, J.T. Novak, Z. He, Energy consumption by forward osmosis treatment 862 of landfill leachate for water recovery, Waste Manage., 63 (2017) 284-291. 863 [31] D. Li, Y. Yan, H. Wang, Recent advances in polymer and polymer composite membranes for reverse 864 and forward osmosis processes, Prog. Polym. Sci., 61 (2016) 104-155. 865 [32] S. Zou, M. Qin, Z. He, Tackle reverse solute flux in forward osmosis towards sustainable water 866 recovery: reduction and perspectives, Water Res., 149 (2019) 362-374. 867 [33] W. Xu, Q. Chen, Q. Ge, Recent advances in forward osmosis (FO) membrane: Chemical modifications 868 on membranes for FO processes, Desalination, 419 (2017) 101-116. 869 [34] Q. Yang, J. Lei, D.D. Sun, D. Chen, Forward Osmosis Membranes for Water Reclamation, Sep. Purif. 870 Rev., 45 (2016) 93-107. 871 [35] N.Y. Yip, A. Tiraferri, W.A. Phillip, J.D. Schiffman, M. Elimelech, High performance thin-film composite 872 forward osmosis membrane, Environ. Sci. Technol., 44 (2010) 3812-3818. 873 [36] A. Tiraferri, N.Y. Yip, W.A. Phillip, J.D. Schiffman, M. Elimelech, Relating performance of thin-film 874 composite forward osmosis membranes to support layer formation and structure, J. Membr. Sci., 367 875 (2011) 340-352. 876 [37] M.D. Firouzjaei, S.F. Seyedpour, S.A. Aktij, M. Giagnorio, N. Bazrafshan, A. Mollahosseini, F. Samadi, 877 S. Ahmadalipour, F.D. Firouzjaei, M.R. Esfahani, A. Tiraferri, M. Elliott, M. Sangermano, A. Abdelrasoul, J.R. 878 McCutcheon, M. Sadrzadeh, A.R. Esfahani, A. Rahimpour, Recent advances in functionalized polymer 879 membranes for biofouling control and mitigation in forward osmosis, J. Membr. Sci., 596 (2020) 117604. 880 [38] P. Arribas, M.C. García-Payo, M. Khayet, L. Gil, Improved antifouling performance of polyester thin 881 film nanofiber composite membranes prepared by interfacial polymerization, J. Membr. Sci., 598 (2020) 882 117774. 883 [39] R. García-Pacheco, F.J. Rabadán, P. Terrero, S.M. Martínez, D. Martínez, E. Campos, e. al., Life+13 884 transfomem: a recycling example within the desalination world, in: XI AEDYR Int. Congr., Valencia, Spain, 885 2016. 886 [40] S. Molina, R. Garcia-Pacheco, L. Rodriguez-Sáez, E. Garcia-Calvo, E. Campos, D. Zarzo, e. al., 887 Transformation of end-of-life RO membranes into recycled NF and UF membranes, surface 888 characterization, in: International Desalination Asociation Word Congress, IDAWC15, San Diego, 889 California, 2015. 890 [41] R. García-Pacheco, J. Landaburu-Aguirre, P. Terrero-Rodríguez, E. Campos, F. Molina-Serrano, J. 891 Rabadán, D. Zarzo, E. García-Calvo, Validation of recycled membranes for treating brackish water at pilot 892 scale, Desalination, 433 (2018) 199-208. 893 [42] R. García-Pacheco, J. Landaburu-Aguirre, A. Lejarazu-Larrañaga, L. Rodríguez-Sáez, S. Molina, T. 894 Ransome, E. García-Calvo, Free chlorine exposure dose (ppm·h) and its impact on RO membranes ageing 895 and recycling potential, Desalination, 457 (2019) 133-143. 896 [43] S. Molina, J. Landaburu-Aguirre, L. Rodríguez-Sáez, R. García-Pacheco, J.G. de la Campa, E. García-897 Calvo, Effect of sodium hypochlorite exposure on polysulfone recycled UF membranes and their surface 898 characterization, Polym. Degrad. Stab., 150 (2018) 46-56. 899 [44] P.A. Araújo, J.C. Kruithof, M.C.M. Van Loosdrecht, J.S. Vrouwenvelder, The potential of standard and 900 modified feed spacers for biofouling control, J. Membr. Sci., 403-404 (2012) 58-70. 901 [45] X. Song, Z. Liu, D. Sun, Nano Gives the Answer: Breaking the Bottleneck of Internal Concentration 902 Polarization with a Nanofiber Composite Forward Osmosis Membrane for a High Water Production Rate, 903 Adv. Mater., 23 (2011) 3256-3260. 904 49 [46] W.A. Phillip, J.S. Yong, M. Elimelech, Reverse draw solute permeation in forward osmosis: modeling 905 and experiments, Environ. Sci. Technol., 44 (2010) 5170-5176. 906 [47] G. Blandin, H. Vervoort, P. Le-Clech, A.R.D. Verliefde, Fouling and cleaning of high permeability 907 forward osmosis membranes, J. Water Process Eng., 9 (2016) 161-169. 908 [48] S. Lim, M.J. Park, S. Phuntsho, L. Tijing, G. Nisola, W.-G. Shim, W.-J. Chung, H.K. Shon, Dual-layered 909 nanocomposite membrane based on polysulfone/graphene oxide for mitigating internal concentration 910 polarization in forward osmosis, Polymer, 110 (2017) 36-48. 911 [49] G. Blandin, H. Vervoort, A. D’Haese, K. Schoutteten, J.V. Bussche, L. Vanhaecke, D.T. Myat, P. Le-Clech, 912 A.R.D. Verliefde, Impact of hydraulic pressure on membrane deformation and trace organic contaminants 913 rejection in pressure assisted osmosis (PAO), Process Saf. Environ. Prot., 102 (2016) 316-327. 914 [50] T. Liu, S. Xu, D. Zhang, S. Sourirajan, T. Matsuura, Pore size and pore size distribution on the surface 915 of polyethersulfone hollow fiber membranes, Desalination, 85 (1991) 1-12. 916 [51] P. Arribas, M. Khayet, M.C. García-Payo, L. Gil, Self-sustained electro-spun polysulfone nano-fibrous 917 membranes and their surface modification by interfacial polymerization for micro- and ultra-filtration, 918 Sep. Purif. Technol., 138 (2014) 118-129. 919 [52] S. Singh, K.C. Khulbe, T. Matsuura, P. Ramamurthy, Membrane characterization by solute transport 920 and atomic force microscopy, J. Membr. Sci., 142 (1998) 111-127. 921 [53] L. Pisani, Simple Expression for the Tortuosity of Porous Media, Transp. Porous Med., 88 (2011) 193-922 203. 923 [54] S.S. Manickam, J.R. McCutcheon, Model thin film composite membranes for forward osmosis: 924 Demonstrating the inaccuracy of existing structural parameter models, J. Membr. Sci., 483 (2015) 70-74. 925 [55] P.K. Kang, W. Lee, S. Lee, A.S. Kim, Origin of structural parameter inconsistency in forward osmosis 926 models: A pore-scale CFD study, Desalination, 421 (2017) 47-60. 927 [56] J.R. McCutcheon, M. Elimelech, Influence of concentrative and dilutive internal concentration 928 polarization on flux behavior in forward osmosis, J. Membr. Sci., 284 (2006) 237-247. 929 [57] A. Tiraferri, N.Y. Yip, A.P. Straub, S. Romero-Vargas Castrillon, M. Elimelech, A method for the 930 simultaneous determination of transport and structural parameters of forward osmosis membranes, J. 931 Membr. Sci., 444 (2013) 523-538. 932 [58] S.S. Manickam, J.R. McCutcheon, Understanding mass transfer through asymmetric membranes 933 during forward osmosis: A historical perspective and critical review on measuring structural parameter 934 with semi-empirical models and characterization approaches, Desalination, 421 (2017) 110-126. 935 [59] N.N. Bui, J.T. Arena, J.R. McCutcheon, Proper accounting of mass transfer resistances in forward 936 osmosis: Improving the accuracy of model predictions of structural parameter, J. Membr. Sci., 492 (2015) 937 289-302. 938 [60] M.J. Ariza, J. Benavente, Streaming potential along the surface of polysulfone membranes: a 939 comparative study between two different experimental systems and determination of electrokinetic and 940 adsorption parameters, J. Membr. Sci., 190 (2001) 119-132. 941 [61] I. Christl, R. Kretzschmar, Relating ion binding by fulvic and humic acids to chemical composition and 942 molecular size. 1. Proton binding, Environ. Sci. Technol., 35 (2001) 2505-2511. 943 [62] N.Z.S. Yahaya, M.Z.M. Pauzi, N. Mu’ammar Mahpoz, M.A. Rahman, K.H. Abas, A.F. Ismail, M.H.D. 944 Othman, J. Jaafar, Chapter 10 - Forward Osmosis for Desalination Application, in: A.F. Ismail, M.A. Rahman, 945 M.H.D. Othman, T. Matsuura (Eds.) Membrane Separation Principles and Applications, Elsevier, 2019, pp. 946 315-337. 947 [63] S. Engelhardt, J. Vogel, S.E. Duirk, F.B. Moore, H.A. Barton, Urea and ammonium rejection by an 948 aquaporin-based hollow fiber membrane, J. Water Process Eng., 32 (2019). 949 [64] G. Blandin, F. Ferrari, G. Lesage, P. Le-Clech, M. Heran, X. Martinez-Llado, Forward Osmosis as 950 Concentration Process: Review of Opportunities and Challenges, Membranes (Basel), 10 (2020). 951 50 [65] E.M. Garcia-Castello, J.R. McCutcheon, Dewatering press liquor derived from orange production by 952 forward osmosis, J. Membr. Sci., 372 (2011) 97-101. 953 [66] M.S. Fahmey, A.-H.M. El-Aassar, M. M.Abo-Elfadel, A.S. Orabi, R. Das, Comparative performance 954 evaluations of nanomaterials mixed polysulfone: A scale-up approach through vacuum enhanced direct 955 contact membrane distillation for water desalination, Desalination, 451 (2019) 111-116. 956 [67] D. Song, J. Xu, Y. Fu, L. Xu, B. Shan, Polysulfone/sulfonated polysulfone alloy membranes with an 957 improved performance in processing mariculture wastewater, Chem. Eng. J., 304 (2016) 882-889. 958 [68] N.M. Mazlan, P. Marchetti, H.A. Maples, B. Gu, S. Karan, A. Bismarck, A.G. Livingston, Organic fouling 959 behaviour of structurally and chemically different forward osmosis membranes – A study of cellulose 960 triacetate and thin film composite membranes, J. Membr. Sci., 520 (2016) 247-261. 961 [69] M. Xie, L.D. Nghiem, W.E. Price, M. Elimelech, Impact of humic acid fouling on membrane 962 performance and transport of pharmaceutically active compounds in forward osmosis, Water Res., 47 963 (2013) 4567-4575. 964 [70] J. Ren, J.R. McCutcheon, A new commercial thin film composite membrane for forward osmosis, 965 Desalination, 343 (2014) 187-193. 966 [71] W.N.A.S. Abdullah, S. Tiandee, W. Lau, F. Aziz, A.F. Ismail, Potential use of nanofiltration like-forward 967 osmosis membranes for copper ion removal, Chin. J. Chem. Eng., 28 (2020) 420-428. 968 [72] P.S. Goh, A.F. Ismail, B.C. Ng, M.S. Abdullah, Recent Progresses of Forward Osmosis Membranes 969 Formulation and Design for Wastewater Treatment, Water, 11 (2019) 2043. 970 [73] Y.-N. Wang, K. Goh, X. Li, L. Setiawan, R. Wang, Membranes and processes for forward osmosis-based 971 desalination: Recent advances and future prospects, Desalination, 434 (2018) 81-99. 972 [74] D.M. Warsinger, S. Chakraborty, E.W. Tow, M.H. Plumlee, C. Bellona, S. Loutatidou, L. Karimi, A.M. 973 Mikelonis, A. Achilli, A. Ghassemi, L.P. Padhye, S.A. Snyder, S. Curcio, C. Vecitis, H.A. Arafat, J.H.t. Lienhard, 974 A review of polymeric membranes and processes for potable water reuse, Prog. Polym. Sci., 81 (2016) 975 209-237. 976 [75] Y. Fang, L. Bian, Q. Bi, Q. Li, X. Wang, Evaluation of the pore size distribution of a forward osmosis 977 membrane in three different ways, J. Membr. Sci., 454 (2014) 390-397. 978 [76] S. Xiong, S. Xu, S. Zhang, A. Phommachanh, Y. Wang, Highly permeable and antifouling TFC FO 979 membrane prepared with CD-EDA monomer for protein enrichment, J. Membr. Sci., 572 (2019) 281-290. 980 [77] W. Xu, Q. Ge, Novel functionalized forward osmosis (FO) membranes for FO desalination: Improved 981 process performance and fouling resistance, J. Membr. Sci., 555 (2018) 507-516. 982 [78] R.C. Ong, T.-S. Chung, J.S. de Wit, B.J. Helmer, Novel cellulose ester substrates for high performance 983 flat-sheet thin-film composite (TFC) forward osmosis (FO) membranes, J. Membr. Sci., 473 (2015) 63-71. 984 [79] Z. Alihemati, S.A. Hashemifard, T. Matsuura, A.F. Ismail, N. Hilal, Current status and challenges of 985 fabricating thin film composite forward osmosis membrane: A comprehensive roadmap, Desalination, 491 986 (2020). 987 [80] M.E. Yakavalangi, S. Rimaz, V. Vatanpour, Effect of surface properties of polysulfone support on the 988 performance of thin film composite polyamide reverse osmosis membranes, J. Appl. Polym. Sci., 134 989 (2017) 44444. 990 [81] M. Kahrizi, J. Lin, G. Ji, L. Kong, C. Song, L.F. Dumée, S. Sahebi, S. Zhao, Relating forward water and 991 reverse salt fluxes to membrane porosity and tortuosity in forward osmosis: CFD modelling, Sep. Purif. 992 Technol., 241 (2020) 116727. 993 [82] W.R. Smith, F. Moučka, I. Nezbeda, Osmotic pressure of aqueous electrolyte solutions via molecular 994 simulations of chemical potentials: Application to NaCl, Fluid Phase Equilib., 407 (2016) 76-83. 995 [83] J.T. Martin, G. Kolliopoulos, V.G. Papangelakis, An improved model for membrane characterization in 996 forward osmosis, J. Membr. Sci., 598 (2020) 117668. 997 51 [84] X. Liu, H.Y. Ng, Fabrication of layered silica–polysulfone mixed matrix substrate membrane for 998 enhancing performance of thin-film composite forward osmosis membrane, J. Membr. Sci., 481 (2015) 999 148-163. 1000 [85] X.-Y. Chi, P.-Y. Zhang, X.-J. Guo, Z.-L. Xu, A novel TFC forward osmosis (FO) membrane supported by 1001 polyimide (PI) microporous nanofiber membrane, Appl. Surf. Sci., 427 (2018) 1-9. 1002 [86] J. Xiang, Z. Xie, M. Hoang, K. Zhang, Effect of amine salt surfactants on the performance of thin film 1003 composite poly(piperazine-amide) nanofiltration membranes, Desalination, 315 (2013) 156-163. 1004 [87] N. Misdan, W.J. Lau, A.F. Ismail, Physicochemical characteristics of poly(piperazine-amide) TFC 1005 nanofiltration membrane prepared at various reaction times and its relation to the performance, J. Polym. 1006 Eng., 35 (2015) 71-78. 1007 [88] P.S. Singh, S.V. Joshi, J.J. Trivedi, C.V. Devmurari, A.P. Rao, P.K. Ghosh, Probing the structural 1008 variations of thin film composite RO membranes obtained by coating polyamide over polysulfone 1009 membranes of different pore dimensions, J. Membr. Sci., 278 (2006) 19-25. 1010 [89] B. Tang, Z. Huo, P. Wu, Study on a novel polyester composite nanofiltration membrane by interfacial 1011 polymerization of triethanolamine (TEOA) and trimesoyl chloride (TMC): I. Preparation, characterization 1012 and nanofiltration properties test of membrane, J. Membr. Sci., 320 (2008) 198-205. 1013 [90] X. Wei, X. Kong, J. Yang, G. Zhang, J. Chen, J. Wang, Structure influence of hyperbranched polyester 1014 on structure and properties of synthesized nanofiltration membranes, J. Membr. Sci., 440 (2013) 67-76. 1015 [91] T.Y. Cath, M. Elimelech, J.R. McCutcheon, R.L. McGinnis, A. Achilli, D. Anastasio, A.R. Brady, A.E. 1016 Childress, I.V. Farr, N.T. Hancock, J. Lampi, L.D. Nghiem, M. Xie, N.Y. Yip, Standard Methodology for 1017 Evaluating Membrane Performance in Osmotically Driven Membrane Processes, Desalination, 312 (2013) 1018 31-38. 1019 [92] M.C.Y. Wong, K. Martinez, G.Z. Ramon, E.M.V. Hoek, Impacts of operating conditions and solution 1020 chemistry on osmotic membrane structure and performance, Desalination, 287 (2012) 340-349. 1021 [93] M. Xie, L.D. Nghiem, W.E. Price, M. Elimelech, A Forward Osmosis–Membrane Distillation Hybrid 1022 Process for Direct Sewer Mining: System Performance and Limitations, Environ. Sci. Technol., 47 (2013) 1023 13486-13493. 1024 [94] H. Jain, M.C. Garg, Fabrication of polymeric nanocomposite forward osmosis membranes for water 1025 desalination—A review, Environ. Technol. Innov., 23 (2021) 101561. 1026 [95] S. Phuntsho, S. Sahebi, T. Majeed, F. Lotfi, J.E. Kim, H.K. Shon, Assessing the major factors affecting 1027 the performances of forward osmosis and its implications on the desalination process, Chem. Eng. J., 231 1028 (2013) 484-496. 1029 [96] Y. Sun, J. Tian, Z. Zhao, W. Shi, D. Liu, F. Cui, Membrane fouling of forward osmosis (FO) membrane 1030 for municipal wastewater treatment: A comparison between direct FO and OMBR, Water Res., 104 (2016) 1031 330-339. 1032 [97] B.K. Pramanik, F.I. Hai, F.A. Roddick, Ultraviolet/persulfate pre-treatment for organic fouling 1033 mitigation of forward osmosis membrane: Possible application in nutrient mining from dairy wastewater, 1034 Sep. Purif. Technol., 217 (2019) 215-220. 1035 [98] M. Zhan, G. Gwak, B.G. Choi, S. Hong, Indexing fouling reversibility in forward osmosis and its 1036 implications for sustainable operation of wastewater reclamation, J. Membr. Sci., 574 (2019) 262-269. 1037 1038 Supporting Information 1 2 3 Recycled reverse osmosis membranes for forward 4 osmosis technology 5 Jorge Contreras-Martíneza, Carmen García-Payoa, Paula Arribasb, Laura Rodríguez-Sáezc,d, 6 Amaia Lejarazu-Larrañagac,d, Eloy García-Calvoc,d, Mohamed Khayeta,c,* 7 8 a Department of Structure of Matter, Thermal Physics and Electronics, Faculty of Physics, 9 University Complutense of Madrid, Avda. Complutense s/n, 28040 Madrid, Spain. 10 b Department of Electrical Engineering, ICAI, Comillas Pontifical University, Alberto Aguilera 11 25, 28015 Madrid, Spain. 12 c IMDEA Water Institute, Avda. Punto Com nº 2, 28805 Alcalá de Henares, Madrid, Spain. 13 d Chemical Engineering Department, University of Alcalá, Ctra. Madrid-Barcelona Km 33.600, 14 28871. Alcalá de Henares, Madrid, Spain. 15 16 * Corresponding author. 17 Phone: +34-91-3945185; E-mail: khayetm@fis.ucm.es (M. Khayet) 18 19 20 21 22 S1. Cross-section SEM images of the asymetric FO commercial membranes and RFO2 23 membrane. 24 (a) (b) (c) 25 Figure S1. Cross-section SEM images of the asymetric FO commercial (a) PA (TFC-HTI) and (b) 26 CTA (CTA-HTI) membranes embedded in a polyster mesh supplied by Hydration Technology 27 Innovations (HTI™, LLC, Albany, USA) and (c) the RFO2 membrane. 28 S2. Effects of HA on the electrical conductivity of NaCl solutions and NaCl on the absorbance 29 of HA solutions. 30 31 Figure S2. Measured electrical conductivity of two NaCl concentrations (200 g/L and 65 g/L) at 32 different concentrations of HA in water solutions. 33 34 Figure S3. Measured absorbance of two HA concentrations (100 mg/L and 10 mg/L) at different 35 concentrations of NaCl in water. 36 As can be seen in Figs. S2 and S3, no interaction of HA on the electrical conductivity of NaCl 37 measurements was observed. Moreover, the absorbance of HA was kept almost constant for 38 different NaCl concentrations (from 0 to 200 g/L of NaCl in water). 39 0 200 400 600 800 1000 50 100 150 200 250 [NaCl] 65 g/L [NaCl] 200 g/L C o n d u c ti v it y ( m S /c m 2 ) [HA] (mg/L) 0 50 100 150 200 0,25 0,30 0,35 0,40 0,45 0,50 0,55 0,60 0,65 0,70 0,75 0,80 0,85 [HA] 10 mg/L [HA] 100 mg/L [NaCl] (g/L) A b s o rb a n c e [ H A ] 1 0 m g /L 3,6 3,7 3,8 3,9 4,0 4,1 4,2 4,3 4,4 4,5 4,6 4,7 4,8 4,9 5,0 5,1 5,2 5,3 5,4 A b s o rb a n c e [ H A ] 1 0 0 m g /L S3. Solute transport tests using polyethylene glycol (PEG) and polyethylene oxide (PEO) of 40 different molecular weights. 41 Solute separation transport tests were performed using polyethylene glycol (PEG) and 42 polyethylene oxide (PEO) of different molecular weights. The selected molecular weights were 43 0.4, 1, 6, 10, 12, 20 and 35 kg/mol for PEGs and 100 and 200 kg/mol for PEOs [1]. The set-up 44 used was described in our previous paper [2]. All chemicals were purchased from Sigma-Aldrich 45 Chemical Co. St. Louis, Massachusetts, USA. The feed temperature, pressure and solute 46 concentration were 296 K, 1.04 bar and 200 ppm, respectively. 47 First, pure water permeance (WP) of each membrane was determined using distilled water as 48 feed. Then, PEG and PEO aqueous solutions were filtered through the membrane following an 49 ascending order of their molecular weights for one hour each and the product permeation rate was 50 determined. The PEG and PEO concentrations of both the feed and permeate were evaluated using 51 the total organic carbon (TOC) analyzer (model CA16 for Skalar Analytical B.V., Breda, 52 Netherlands). The solute rejection factor (α) was calculated as [3]: 53 𝛼 = (1 − 𝐶𝑝 𝐶𝑓 ) · 100 (S1) 54 where 𝐶𝑝 and 𝐶𝑓 are the solute (PEG or PEO) permeate and feed concentrations, respectively. 55 To determinate the size of PEG and PEO, the following equations that relate the Einstein-Stokes 56 radius to the molecular weight of PEG (Eq. (S2)) and PEO (Eq. (S3)) were considered [4]: 57 𝑎𝑟,𝑃𝐸𝐺 = 16.73 · 10−10𝑀0.557 (S2) 58 𝑎𝑟,𝑃𝐸𝑂 = 10.44 · 10−10𝑀0.587 (S3) 59 where M is the molecular weight in g/mol and 𝑎𝑟 is the Einstein-Stokes radius in cm. 60 The obtained water permeance (WP) and solute rejection factor (α) of the RFO membranes are 61 shown in Table S1. 62 Table S1. Water permeance (WP) and solute rejection factor (α) of the RFO membranes obtained by solute transport experiments of 63 aqueous solutions containing PEG and PEO of different molecular weights (Mw). 64 Membrane WP (L/m2·h·bar) α (%) PEG 0.4a PEG 1b PEG 6c PEG 10d PEG 12e PEG 20f PEG 35g PEO 100h PEO 200i RFO1 37 ± 1 34 ± 1 44 ± 4 54 ± 4 64 ± 2 68 ± 3 80 ± 4 97 ± 2 98 ± 1 100 RFO2 43 ± 2 33 ± 1 37 ± 3 55 ± 5 68 ± 3 80 ± 3 93 ± 6 98 ± 2 99 ± 1 100 RFO3 41 ± 1 35 ± 1 36 ± 1 41 ± 4 51 ± 3 63 ± 4 74 ± 7 89 ± 2 99 ± 1 100 RFO4 43 ± 1 36 ± 1 37 ± 2 47 ± 6 58 ± 5 67 ± 3 79 ± 2 91 ± 3 96 ± 1 100 a PEG 0.4: Mw = 0.4 kg/mol, Einstein-Stoke radius (ar,PEG) = 0.94 nm; b PEG 1: Mw = 1 kg/mol, radius (ar,PEG) = 1.57 nm; c PEG 6: Mw = 6 kg/mol, radius (ar,PEG) 65 = 4.26 nm; d PEG 10: Mw = 10 kg/mol, radius (ar,PEG) = 5.66 nm; e PEG 12: Mw = 12 kg/mol, radius (ar,PEG) = 6.26 nm; f PEG 20: Mw = 20 kg/mol, radius (ar,PEG) 66 = 8.32 nm; g PEG 35: Mw = 35 kg/mol, radius (ar,PEG) = 11.36 nm; h PEO 100: Mw = 100 kg/mol, radius (ar,PEO) = 17.98 nm; i PEO 200: Mw = 200 kg/mol, radius 67 (ar,PEO) = 27.01 nm.68 S6 The mean pore size and the geometric standard deviation were determined from the plot of the rejection factor as a function of the solute diameter and their correlation according to the log-normal probability function (Fig. S4) [4]. Figure S4. Log-normal probability plot of the rejection factor of the RFO membranes as a function of solute diameter. The mean pore size (𝜇𝑝) and the molecular weight cut-off (MWCO) correspond to the Einstein-Stokes diameters of 50% and 90% rejection factors, respectively. The geometric standard deviation is the ratio between the Einstein-Stokes diameters corresponding to 84.13% and 50% rejection factors [4]. The pore density (N) and porosity () were determined from the obtained permeance data using the following equations [4]: 𝑁 = 128𝜂𝛿𝐽𝑡 𝜋∆𝑃 ∑ 𝑓𝑖𝑑𝑖 4𝑑𝑚𝑎𝑥 𝑑𝑚𝑖𝑛 (𝑆4)  = ( 𝑁𝜋 4 ∑ 𝑓𝑖𝑑𝑖 2𝑑𝑚𝑎𝑥 𝑑𝑚𝑖𝑛 ) 100 (S5) where 𝜂 is the solvent viscosity, 𝛿 is the pore length considered equivalent to the membrane thickness, ∆𝑃 is the pressure difference across the membrane, 𝐽𝑡 is the total permeate flux and 𝑓𝑖 the fraction of pores with diameter 𝑑𝑖. 1 10 40 70 95 99.5 99.999 RFO1 RFO2 RFO3 RFO4 R e je c ti o n f a c to r (% ) Solute diameter (nm) S7 From the mean pore size (𝜇𝑝) and the geometric standard deviation, the cumulative pore size distribution and the probability density function curves are presented in Fig. S5. Figure S5. Cumulative pore size distribution (a) and probability density function (b) curves of the RFO membranes. 0 5 10 15 20 25 0.00 0.05 0.10 0.15 0.20 P ro b a b il it y d e n s it y f u n c ti o n d f( d i) /d (d i) , (n m -1 ) Pore size (nm) 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 P o re f ra c ti o n w it h s iz e l e s s th a n s ta te d v a lu e RFO1 RFO2 RFO3 RFO4 S8 S4. Reverse salt transfer factor (RTs) of the RFO and TRFO membranes For FO membranes, NaCl is one of the most used draw solutes due to its small ion sizes and its high diffusivity in the FO membrane support. In this study, FO tests were carried out for low concentration (10 mg/L HA feed aqueous solution and 65 g/L NaCl draw aqueous solution) and high concentration (100 mg/L HA feed aqueous solution and 200 g/L NaCl draw aqueous solution). The specific reverse salt flux (Js/J) is a quantitative factor indicating the bi-directional diffusion in the FO separation process. A reduced (Js/J) value indicates an increase of the selectivity and efficiency of the membrane. However, as shown in Figs. 5b and 8b, Js/J values depend on the NaCl concentration of the draw solution. A higher NaCl concentration of the draw solution induced a greater Js/J value. Similar results were reported elsewhere [5]. The higher reverse draw solute flux detected for a higher NaCl concentration reduces further the driving force for water transport (i.e. decrease of the permeate flux). In this study two different conditions of both the feed and draw solutions have been tested to evaluate the FO performance of RFO and TRFO membranes. In order to compare the draw solute loss due to the reverse salt flux for both high and low concentrations, a normalized reverse salt transfer (𝑅𝑇𝑠 ) parameter was defined as [6]: 𝑅𝑇𝑠 (%) = ( 𝐶𝑠,𝐹,𝑡 𝐶𝑠,𝐷,0 ) · 100 (S6) where 𝐶𝑠,𝐹,𝑡 is the final salt (NaCl) concentration of the feed solution and 𝐶𝑠,𝐷,0 is the initial salt concentration of the draw solution. RTs reveals the amount of NaCl transported across the FO membrane to the feed side in relation to the initial draw concentration. The obtained normalized reverse salt transfer (RTs) of all membranes is shown in Fig. S6. In general, as it was expected, RTs increases with the increase of the salt concentration in the draw solution except for the CTA-HTI membrane. Compared to all membranes, the commercial membrane TFC-HTI exhibits the lowest RTs values while the CTA-HTI membrane showed the worst results (Fig. S6a). The RTs values were found to be higher for the RFO membranes than for the TRFO membranes. For the low concentration test, the RFO membranes cleaned only in the pilot plant (RFO1 and RFO2 membranes) show significantly lower RTs results than the other RFO membranes cleaned also following the laboratory scale protocol (RFO3 and RFO4 membranes). This can be due to the existence of the PA residues in the pores of the PSf support that retards the diffusion of the salt from the draw solution to S9 the feed solution [7]. However, for the TRFO membranes no significant difference was observed (Fig. S6b). Figure S6: Normalized reverse salt transfer (RTs) of the RFO (a) and TRFO (b) membranes together with that of the commercial CTA-HTI and TFC-HTI membranes for low concentration (10 mg/L HA feed aqueous solution and 65 g/L NaCl draw aqueous solution) and high concentration (100 mg/L HA feed aqueous solution and 200 g/L NaCl draw aqueous solution). S10 S5. FO results of the best RFO membrane (RFO2) subjected to a passive NaClO cleaning process in the pilot plant under a level dose of 106 ppm·h. Table S2. FO results of the RFO2 membrane obtained for low (Lc) and high (Hc) concentration tests: permeate flux (J), reverse salt permeate flux (Js), specific reverse salt flux (Js/J), reverse salt transfer factor (RTs), HA rejection factor (RHA) and HA concentration factor (CFHA). J (kg/m2·h) Js (10-3 kg/m2·h) Js/J (10-3) RTs (%) RHA (%) CFHA Lc 6.2 ± 0.3 4 ± 2 0.7 ± 0.4 1.48 ± 0.09 99.52 ± 0.24 1.12 ± 0.02 Hc 7.2 ± 0.3 42 ± 1 6 ± 1 2.36 ± 0.09 99.82 ± 0.06 1.11 ± 0.01 S11 S6. Study of passive cleaning under laboratory protocol with a total NaClO exposure dose of 3·105 ppm·h, To identify the efficacy of the followed cleaning protocols carried out either in the pilot plant, in the laboratory or both, cleaning was carried out only under the laboratory protocol. After disassembling the discarded modules, the membranes were exposed to an aqueous solution of NaClO at 6200 ppm·h for 48h, reaching a total exposure dose of 3·105 ppm.h. The membrane is named hereafter RFO5 membrane. The membrane thickness and water contact angle of this membrane are summarized in Table S3. The obtained water contact angle of both sides of the membrane RFO5 as well as its thickness can be considered similar to the values obtained for the other RFO membranes (See Table 3). The IEP value obtained for the RFO5 membrane is slightly higher than that of the membranes RFO1 and RFO2 that were recycled following the pilot plant cleaning protocol, but comparable to that of the membranes RFO3 and RFO4 that were subjected to Laboratory cleaning after the pilot plant cleaning. Table S3. Water contact angle (w) of the active layer (AL) and support layer (SL), thickness (δ) and isoelectric point (IEP) of the RFO5 membrane. Membrane (w)AL (º) (w)SL (º) δ (μm) IEP (pH) RFO5 77 ± 3 65 ± 7 98 ± 3 4.01 From the SEM image (Figure S7) it can be seen a similar surface to that of the RFO2 membrane (Figure 2) recycled following only the pilot plant cleaning protocol under an exposure dose of 106 ppm·h. Figure S7. SEM images of the support and active layer of the RFO5 membrane. S12 The water permeance (WP) and the solute transport experiments with PEGs and PEOs were also determined for the membrane RFO5 and the results are summarized in Table S4. The WP values of the membranes RFO5 and RFO1 (Table S1) are similar indicating that the cleaning procedure in the laboratory is more efficient since less exposure time was applied compared to that in the pilot plant. The solute rejection factor values obtained in solute transport tests for the RFO5 membrane were similar to the values obtained for the RFO3 membrane (Table S1) and slightly lower than the values obtained for the RFO1 membrane (Table S1). Table S4. Water permeance (WP) and solute rejection factor (α) of the RFO5 membrane obtained by the solute transport experiments of aqueous solutions containing PEG and PEO of different molecular weights (Mw). WP (L/m2·h·bar) α (%) PEG 0.4a PEG 1b PEG 6c PEG 10d PEG 12e PEG 20f PEG 35g PEO 100h PEO 200i 38 ± 1 38 ± 2 45 ± 2 47 ± 3 54 ± 2 64 ± 2 71 ± 2 80 ± 4 99 ± 1 100 a PEG 0.4: Mw = 0.4 kg/mol, Einstein-Stoke radius (ar,PEG) = 0.94 nm; b PEG 1: Mw = 1 kg/mol, radius (ar,PEG) = 1.57 nm; c PEG 6: Mw = 6 kg/mol, radius (ar,PEG) = 4.26 nm; d PEG 10: Mw = 10 kg/mol, radius (ar,PEG) = 5.66 nm; e PEG 12: Mw = 12 kg/mol, radius (ar,PEG) = 6.26 nm; f PEG 20: Mw = 20 kg/mol, radius (ar,PEG) = 8.32 nm; g PEG 35: Mw = 35 kg/mol, radius (ar,PEG) = 11.36 nm; h PEO 100: Mw = 100 kg/mol, radius (ar,PEO) = 17.98 nm; i PEO 200: Mw = 200 kg/mol, radius (ar,PEO) = 27.01 nm. The membrane characteristics obtained from the solute transport experiments (mean pore size (𝜇𝑝), its geometric standard deviation (σp), density (N), porosity (ε) and molecular weight cut-off (MWCO), together with the tortuosity (τ, Eq. (1)) and the “intrinsic” structural parameter (Sint, from Eq. (2)) are presented in Table S5. The membrane RFO5 recycled with a lower total dose level following the laboratory cleaning protocol exhibits similar WP to those of the RFO1 membrane recycled with more than 2 times total dose level following the pilot plant cleaning (See Table 4). This is attributed mainly to the greater pore size of the membrane RFO5, which is 16% larger than that of the membrane RFO1. However, the pore density and porosity of the membrane RFO5 are lower than those of the other RFO S13 membranes (i.e. 4446-5467 pores/µm2 for the pore density and 26-30% for the porosity). In addition, among all recycled membranes studied, the MWCO, tortuosity and Sint values of the RFO5 membrane are the highest (See Table S5). Table S5. Mean pore size (μp), geometric standard deviation (σp), pore density (N), porosity (ε), molecular weight cut-off (MWCO), tortuosity (τ) and “intrinsic” structural parameter factor (Sint) of the RFO5 membrane obtained from solute transport experiments. Membrane μp (nm) σp (-) N (pores/μm2) ε (%) MWCO (kDa) τ (-) Sint (µm) RFO5 4.3 2.6 3191 23 46.5 2.33 1003 Table S6 summarized the results of the FO tests of the membrane RFO5. As can be seen from the FO results listed in Table S2, the membranes RFO2 and RFO5 exhibit similar J and Js/J values for the low concentration FO test. This is due to the cleaning step in the laboratory that requires three times less total dose level than in the pilot plant to prepare recycled membranes with similar morphological characteristics. The observed increase of J (i.e. J = 0.1 kg/m2h) of the RFO5 membrane for the high concentration test from that of the low concentration test is very low. This increase is 90% less than that observed for the membrane RFO2 (i.e. J = 1 kg/m2h). The Js/J value corresponding to the high concentration test of the RFO5 membrane is greater than that of the other RFO membranes. It could be caused by a high Sint structural parameter (low porosity and high tortuosity) of the RFO5 membrane inducing high ICP effects. The membranes RFO5 and RFO2 exhibit similar RHA values for both low and high concentrations tests. Table S6. FO results of the RFO5 membrane obtained for low (Lc) and high (Hc) concentration tests: permeate flux (J), reverse salt permeate flux (Js), specific reverse salt flux (Js/J), reverse salt transfer factor (RTs), HA rejection factor (RHA) and HA concentration factor (CFHA). J (kg/m2·h) Js (10-3 kg/m2·h) Js/J (10-3) RTs (%) RHA (%) CFHA Lc 6.2 ± 0.3 9 ± 3 1.4. ± 0.4 2.30 ± 0.07 98.57 ± 0.42 1.09 ± 0.05 Hc 6.3 ± 0.3 135 ± 7 21 ± 1 2.60 ± 0.05 99.89 ± 0.06 1.06 ± 0.02 S14 In general, cleaning the discarded RO membrane following the laboratory protocol with an exposure dose three times lower than the pilot plant cleaning protocol resulted in a recycled membrane with very similar intrinsic characteristics, but different transport properties appreciated mainly in high concentration FO tests in which the Js/J values obtained were higher. REFERENCES [1] T. Liu, S. Xu, D. Zhang, S. Sourirajan, T. Matsuura, Pore size and pore size distribution on the surface of polyethersulfone hollow fiber membranes, Desalination, 85 (1991) 1-12. [2] P. Arribas, M. Khayet, M.C. García-Payo, L. Gil, Self-sustained electro-spun polysulfone nano-fibrous membranes and their surface modification by interfacial polymerization for micro- and ultra-filtration, Sep. Purif. Technol., 138 (2014) 118-129. [3] M. Khayet, C.Y. Feng, K.C. Khulbe, T. Matsuura, Preparation and characterization of polyvinylidene fluoride hollow fiber membranes for ultrafiltration, Polymer, 43 (2002) 3879- 3890. [4] S. Singh, K.C. Khulbe, T. Matsuura, P. Ramamurthy, Membrane characterization by solute transport and atomic force microscopy, J. Membr. Sci., 142 (1998) 111-127. [5] J.T. Martin, G. Kolliopoulos, V.G. Papangelakis, An improved model for membrane characterization in forward osmosis, J. Membr. Sci., 598 (2020) 117668. [6] B.R. Babu, N.K. Rastogi, K.S.M.S. Raghavarao, Effect of process parameters on transmembrane flux during direct osmosis, J. Membr. Sci., 280 (2006) 185-194. [7] M. Mohammadifakhr, J. de Grooth, H.D.W. Roesink, A.J.B. Kemperman, Forward Osmosis: A Critical Review, Processes, 8 (2020) 404. DES-D-21-00938_R1 DES-D-21-00938_Supporting information