Using Reverse Osmosis Membrane at High Temperature for Water Recovery and Regeneration from Thermo-Responsive Ionic Liquid-Based Draw Solution for Efficient Forward Osmosis

Forward osmosis (FO) membrane process is expected to realize energy-saving seawater desalination. To this end, energy-saving water recovery from a draw solution (DS) and effective DS regeneration are essential. Recently, thermo-responsive DSs have been developed to realize energy-saving water recovery and DS regeneration. We previously reported that high-temperature reverse osmosis (RO) treatment was effective in recovering water from a thermo-responsive ionic liquid (IL)-based DS. In this study, to confirm the advantages of the high-temperature RO operation, thermo-sensitive IL-based DS was treated by an RO membrane at temperatures higher than the lower critical solution temperature (LCST) of the DS. Tetrabutylammonium 2,4,6-trimethylbenznenesulfonate ([N4444][TMBS]) with an LCST of 58 °C was used as the DS. The high-temperature RO treatment was conducted at 60 °C above the LCST using the [N4444][TMBS]-based DS-lean phase after phase separation. Because the [N4444][TMBS]-based DS has a significantly temperature-dependent osmotic pressure, the DS-lean phase can be concentrated to an osmotic pressure higher than that of seawater at room temperature (20 °C). In addition, water can be effectively recovered from the DS-lean phase until the DS concentration increased to 40 wt%, and the final DS concentration reached 70 wt%. From the results, the advantages of RO treatment of the thermo-responsive DS at temperatures higher than the LCST were confirmed.


Introduction
Recently, water treatment using forward osmosis (FO) membranes has attracted significant attention. Conventional membrane separation processes, such as reverse osmosis (RO), nanofiltration (NF), and ultrafiltration (UF), are pressure-driven separation methods that permeate water using mechanical pressure as the driving force for water permeation. The FO membrane process is osmotic pressure-driven [1]; herein, a semipermeable FO membrane is arranged between the draw solution (DS) with higher osmotic pressure and feed solution (FS) with lower osmotic pressure than that of DS; subsequently, water permeates through the FO membrane from FS to DS by the osmotic pressure difference generated between the two solutions. Therefore, mechanical pressure is not required to draw water from the FS through the FO membrane, and the pump power of the FO process is smaller Recently, we found that the osmotic pressure of thermo-responsive IL-DS is temperature-dependent and smaller at elevated temperatures [32]. In addition, we confirmed that a high-water flux could be realized by treating the DS-lean phase with an RO membrane at high temperatures and proposed the effectiveness of the high-temperature DS regeneration process.
In our previous work, the DS-lean phase was treated using an RO membrane at high temperature. However, these temperatures were lower than the lower critical solution temperature (LCST) of the DS. In this work, the RO membrane treatment of the DS-lean phase was investigated at temperatures higher than the LCST (Figure 1b), and the superiority of the high-temperature RO treatment of dilute DS was demonstrated. The hightemperature RO membrane treatment of the thermo-responsive DS, in which the osmotic pressure decreases at high temperatures, will not only increase the water flux but also raise the upper limit of the DS concentration by the RO treatment as shown in Figure 1b. In addition, in this study, we propose a further advantage of the high-temperature RO treatment of the DS-lean phase at temperatures higher than the LCST of the DS. Under the condition that the DS separates into DS-rich and -lean phases, the osmotic pressures of the two liquid phases were found to be equal. During the concentration of DS by RO treatment in the two-phase region of the liquid-liquid phase separation, the volume fractions of the DS-rich phase and the DS-lean phase vary, but their osmotic pressure does not change. Therefore, in the two-phase region, the driving force of water permeation, which is the pressure difference between the mechanical pressure applied to the FS and the osmotic pressure of the DS, for the RO membrane treatment does not increase but is constant. Therefore, at temperatures higher than the LCST, it is considered that a high-water Temperature DS concentration RO   Recently, we found that the osmotic pressure of thermo-responsive IL-DS is temperaturedependent and smaller at elevated temperatures [32]. In addition, we confirmed that a high-water flux could be realized by treating the DS-lean phase with an RO membrane at high temperatures and proposed the effectiveness of the high-temperature DS regeneration process.
In our previous work, the DS-lean phase was treated using an RO membrane at high temperature. However, these temperatures were lower than the lower critical solution temperature (LCST) of the DS. In this work, the RO membrane treatment of the DSlean phase was investigated at temperatures higher than the LCST (Figure 1b), and the superiority of the high-temperature RO treatment of dilute DS was demonstrated. The hightemperature RO membrane treatment of the thermo-responsive DS, in which the osmotic pressure decreases at high temperatures, will not only increase the water flux but also raise the upper limit of the DS concentration by the RO treatment as shown in Figure 1b. In addition, in this study, we propose a further advantage of the high-temperature RO treatment of the DS-lean phase at temperatures higher than the LCST of the DS. Under the condition that the DS separates into DS-rich and -lean phases, the osmotic pressures of the two liquid phases were found to be equal. During the concentration of DS by RO treatment in the two-phase region of the liquid-liquid phase separation, the volume fractions of the DS-rich phase and the DS-lean phase vary, but their osmotic pressure does not change. Therefore, in the two-phase region, the driving force of water permeation, which is the pressure difference between the mechanical pressure applied to the FS and the osmotic pressure of the DS, for the RO membrane treatment does not increase but is constant. Therefore, at temperatures higher than the LCST, it is considered that a high-water flux for recovering water from the DS-lean phase can be maintained over a wide DS concentration range. In this study, we experimentally confirmed the advantages of the high-temperature RO membrane treatment of the DS-lean phase after liquid-liquid phase separation at temperatures higher than the LCST.

Materials and Reagents
The heat-tolerant RO membrane (NTR-759HG) composed of polyamide active layer and polysulfone support was purchased from Nitto Denko Co. (Osaka, Japan). The maximum allowable temperature and pressure for continuous use are 60 • C and 49 bar, respectively. At 60 • C, the pure water flux was about 70 LMH, and the NaCl rejection for 0.11 mol/L NaCl aqueous solution was 94%, respectively. Tetrabutylammonium hydroxide (40% solution in water) ([N 4444 ][OH]), sodium dimethylbenzenesulfonate monohydrate, sodium mesitylenesulfonate, and trifluoroacetic acid were purchased from Tokyo Chemical Industry Co., Ltd. (Tokyo, Japan). Tetrabutylphosphonium hydroxide (40% in water) was purchased from Sigma-Aldrich Co. (Tokyo, Japan). Pure water was generated using a Millipore Milli-Q system (Millipore, Burlington, MA, USA). All materials and reagents were used as received without further treatment or purification.  3 COO]) were used to prepare thermo-responsive IL-DSs. They were synthesized in our laboratory using the same procedure reported in our previous study [30].

Evaluation of Osmotic Pressure of the LCST-Type Solution
The osmotic pressure of the aqueous solution of [N 4444 ][TMBS] at 25 • C was measured using a water activity meter (AquaLab Seris4TDL, AINEX Co., Ltd., Japan). Aqueous solution samples (7 mL) were injected into the sample vial, maintained at 25 • C for more than 15 min, and the activities of water in each sample (a w ) were measured. The measured a w was used to calculate the osmotic pressure, π, using the following equation: The RO filtration setup is illustrated in Figure 2. A piece of heat-tolerant RO membrane was placed inside the thermostatted RO cell, which was immersed in a constanttemperature water bath for precise temperature control. The diaphragm pump was adjusted to provide the FS at a flow rate of 10 mL min −1 , and the pressure was adjusted using a valve. The cell was constantly stirred with a magnetic stirrer to prevent concentration polarization at the membrane surface. The filtrate was automatically collected and weighed, and the stabilized mass data were used to calculate the water flux.
Membranes 2021, 11, x FOR PEER REVIEW 5 of 13 polarization at the membrane surface. The filtrate was automatically collected and weighed, and the stabilized mass data were used to calculate the water flux. Because the polymeric RO membrane may deform at high temperatures, which could result in a potential impact on water permeance, we pre-treated the membranes by filtrating pure water at 60 °C and 15 bar for 12 h before each experiment. After the pretreatment, the water flux became almost constant at about 80% of the initial value. Thus, the effects of membrane deformation and consolidation can be eliminated. This is because the temperature-responsive IL forms aggregates at high temperatures and the concentration of free IL molecules in the DS decreases [32]. Focusing on the concentration dependence on osmotic pressure, the osmotic pressure at a temperature lower than the LCST increased monotonically as the IL concentration increased (for example, Figure  3c). In contrast, the osmotic pressure barely increases in the concentration range where the phase separation occurs (two-phase region). In the two-phase region, the DS separates into DS rich phase and DS lean phase. When the mole of IL in the DS-rich and -lean phases are respectively n1 and n2, and the volumes of the DS-lean and -rich phases are respectively V1 and V2, the IL concentration of the DS rich phase, C1, and that of the DS lean phase, C2, are C1 = n1/V1 and C2 = n2/V2, respectively. It should be noted that, in the two-phase region, the IL concentration shown in the x-axis of Figure 3 is an apparent IL concentration calculated from the total mole of IL in the total volume of the DS. That is, in the abovementioned case, the apparent IL concentration is calculated as (n1 + n2)/(V1 + V2). If the apparent IL concentration changed in the two-phase region, n1, n2, V1, and V2 are also changed, but the IL concentrations of the DS rich and lean phases, C1 and C2, are not changed. Thus, in the two-phase region, the osmotic pressures of the DS-rich and -lean phases are constant regardless of the apparent IL concentration. In addition, the DS-rich and -lean phases should have the same osmotic pressure. This is because the chemical potentials of H2O and IL in the DS-lean and -rich phases formed by phase separation are equal. In fact, in the case of [N4444][DMBS] aqueous solution, which has a two-phase region at 40 °C and 50 °C, the osmotic pressures in the two-phase regions at these temperatures were almost constant, as shown in Figure 3a. In other words, in the two-phase region at temperatures higher than the LCST, it is predicted that the driving force of water permeation from the DS-lean phase through RO or NF membranes (applied pressure-osmotic pressure) will be constant. Moreover, the osmotic pressure of the thermo-responsive IL aqueous solution decreases at high temperatures. Therefore, it is expected that the driving force of pure water permeation during high-temperature operation could be much larger than that during the Because the polymeric RO membrane may deform at high temperatures, which could result in a potential impact on water permeance, we pre-treated the membranes by filtrating pure water at 60 • C and 15 bar for 12 h before each experiment. After the pretreatment, the water flux became almost constant at about 80% of the initial value. Thus, the effects of membrane deformation and consolidation can be eliminated. The osmotic pressure of each IL-DS decreased as the temperature increased. This is because the temperature-responsive IL forms aggregates at high temperatures and the concentration of free IL molecules in the DS decreases [32]. Focusing on the concentration dependence on osmotic pressure, the osmotic pressure at a temperature lower than the LCST increased monotonically as the IL concentration increased (for example, Figure 3c). In contrast, the osmotic pressure barely increases in the concentration range where the phase separation occurs (two-phase region). In the two-phase region, the DS separates into DS rich phase and DS lean phase. When the mole of IL in the DS-rich and -lean phases are respectively n 1 and n 2 , and the volumes of the DS-lean and -rich phases are respectively V 1 and V 2 , the IL concentration of the DS rich phase, C 1 , and that of the DS lean phase, C 2 , are C 1 = n 1 /V 1 and C 2 = n 2 /V 2 , respectively. It should be noted that, in the two-phase region, the IL concentration shown in the x-axis of Figure 3 is an apparent IL concentration calculated from the total mole of IL in the total volume of the DS. That is, in the abovementioned case, the apparent IL concentration is calculated as (n 1 + n 2 )/ (V 1 + V 2 ). If the apparent IL concentration changed in the two-phase region, n 1 , n 2 , V 1 , and V 2 are also changed, but the IL concentrations of the DS rich and lean phases, C 1 and C 2 , are not changed. Thus, in the two-phase region, the osmotic pressures of the DS-rich and -lean phases are constant regardless of the apparent IL concentration. In addition, the DS-rich and -lean phases should have the same osmotic pressure. This is because the chemical potentials of H 2 O and IL in the DS-lean and -rich phases formed by phase separation are equal. In fact, in the case of [N 4444 ][DMBS] aqueous solution, which has a two-phase region at 40 • C and 50 • C, the osmotic pressures in the two-phase regions at these temperatures were almost constant, as shown in Figure 3a. In other words, in the two-phase region at temperatures higher than the LCST, it is predicted that the driving force of water permeation from the DS-lean phase through RO or NF membranes (applied pressure-osmotic pressure) will be constant. Moreover, the osmotic pressure of the thermo-responsive IL aqueous solution decreases at high temperatures. Therefore, it is expected that the driving force of pure water permeation during high-temperature operation could be much larger than that during the low-temperature operation. In fact, in a previous study, we confirmed that the water flux from the DS was higher when the RO operation was performed at a high temperature [32]. Therefore, from the viewpoint of obtaining a high-water flux, it is suggested that high-temperature RO (or NF) membrane operation is effective for treating the DS-lean phase.

Temperature Dependence of Osmotic Pressure of IL-Based DSs
Membranes 2021, 11, x FOR PEER REVIEW 6 of 13 low-temperature operation. In fact, in a previous study, we confirmed that the water flux from the DS was higher when the RO operation was performed at a high temperature [32]. Therefore, from the viewpoint of obtaining a high-water flux, it is suggested that hightemperature RO (or NF) membrane operation is effective for treating the DS-lean phase. [TMBS] aqueous solution measured at 25 °C were those previously reported in Ref. [32]. Reproduced with permission from Ref. [32]. Elsevier, 2021.

Effect of High-Temperature RO Treatment for DS Regeneration
The treatment of the DS-lean phase enables the regeneration of the DS-lean phase in addition to water recovery. For the regeneration of the DS-lean phase, the IL concentration should be increased as much as possible by RO (or NF) operation. Therefore, we evaluated the efficiency of high-temperature operation for DS-lean phase treatment by RO or NF membranes from the viewpoint of the IL concentration limit of the DS-lean phase.
For seawater desalination by FO, the osmotic pressure of the regenerated DS must be higher than the osmotic pressure of seawater (approximately 30 bar). In other words, the DS-lean phase should be concentrated such that its osmotic pressure is higher than that of seawater. For example, if seawater desalination is performed at 20 °C (room temperature),   [32]. Reproduced with permission from Ref. [32]. Elsevier, 2021.

Effect of High-Temperature RO Treatment for DS Regeneration
The treatment of the DS-lean phase enables the regeneration of the DS-lean phase in addition to water recovery. For the regeneration of the DS-lean phase, the IL concentration should be increased as much as possible by RO (or NF) operation. Therefore, we evaluated the efficiency of high-temperature operation for DS-lean phase treatment by RO or NF membranes from the viewpoint of the IL concentration limit of the DS-lean phase.
For seawater desalination by FO, the osmotic pressure of the regenerated DS must be higher than the osmotic pressure of seawater (approximately 30 bar). In other words, the DS-lean phase should be concentrated such that its osmotic pressure is higher than that of seawater. For example, if seawater desalination is performed at 20 • C (room temperature), the minimum DS concentration required for the FO desalination operation can be estimated from the osmotic pressure at 20 • C, shown in Figure 3. The minimum IL concentrations required were approximately 69 wt%, 58 wt%, and 32 wt% for [ Figure S1). Furthermore, the upper concentration limit of the concentrated DS-lean phase treated by RO and NF membrane operation at a certain mechanical pressure and temperature can be determined from the osmotic pressure with respect to the operating temperature ( Figure 3 the minimum DS concentration required for the FO desalination operation can be estimated from the osmotic pressure at 20 °C, shown in Figure 3.  Figure S1). Furthermore, the upper concentration limit of the concentrated DS-lean phase treated by RO and NF membrane operation at a certain mechanical pressure and temperature can be determined from the osmotic pressure with respect to the operating temperature ( Figure 3)   The efficiency of the seawater desalination process when using each IL-DS is discussed herein. As shown in Figure 4a  Furthermore, it is worth noting that the water flux at 60 °C was also substantially high. As shown in Figure 5, the initial water flux increases with increasing operation temperature. This is because of the low osmotic pressure of the DS and the low viscosity of the FS and DS at high temperatures [32]. In addition, the high-water flux at 60 °C was  Furthermore, it is worth noting that the water flux at 60 • C was also substantially high. As shown in Figure 5, the initial water flux increases with increasing operation temperature. This is because of the low osmotic pressure of the DS and the low viscosity of the FS and DS at high temperatures [32]. In addition, the high-water flux at 60  Figure 3b shows that the osmotic pressure of the rich and lean DS phases at 60 • C is less than 5 bar, which is the osmotic pressure of a 12 wt% [N 4444 ][TMBS] aqueous solution at 50 • C. Therefore, when the mechanical pressure for the RO operation was 15 bar, in the concentration range of 12-58 wt%, the pressure difference between FS and DS, which is the driving force for water permeation by the RO membrane, is maintained at more than 10 bar. Owing to the high and constant driving force, a high-water flux was maintained in the [N 4444 ][TMBS] concentration range of 12 wt% to 40 wt%. In other words, from this result, it can be said that the concept of high-speed RO treatment under phase separation conditions was verified.
Another interesting phenomenon was observed during the RO at 60 When the volume of the DS-rich phase became larger than that of the DS-lean phase, it is considered that the transformation of the continuous phase and disperse phase could have occurred. Thus, the DS-rich phase could become the continuous phase and cover the active layer of the RO membrane (Figure 3b). We speculated that this phase transition would strongly affect the drastic water flux change during RO operation at 60 • C. As mentioned above, the concentrations of the rich and lean DS phases were 58 wt% and 12 wt% at 60 • C, respectively. Thus, when the apparent [N 4444 ][TMBS] concentration was in the range of 12 wt% to approximately 35 wt% (half of the sum of 12 wt% and 58 wt%), the lean phase is the continuous phase, and the rich phase is the disperse phase, according to the phase separation theory [42,43]. In the apparent [N 4444 ][TMBS] concentration range of approximately 35 wt% to 58 wt%, the opposite configuration occurs; that is, the rich phase becomes the continuous phase. The boundary concentration between the higher and lower water fluxes (40 wt% of the apparent [N 4444 ][TMBS] concentration) observed in Figure 5 roughly corresponds to the concentration at which the phase configuration is transformed (approximately 35 wt%). From 12 wt% to 40 wt%, the lean phase mainly contacted the active layer of the RO membrane because the lean phase is the continuous phase in this apparent [N 4444 ][TMBS] concentration range ( Figure S3a). In contrast, the rich phase contacted the active layer of the RO membrane from 40 wt% to 58 wt% because the rich phase is the continuous phase in this apparent [N 4444 ][TMBS] concentration range ( Figure S3b). The sudden decrease in the water flux at 40 wt% was mainly due to the decrease in the diffusion rate of H 2 O in the DS near the surface of the RO membrane. The viscosity of the DS-lean phase was 0.92 mPa s, whereas that of the DS-rich phase was 5.26 mPa s. Thus, the diffusion coefficient of H 2 O in the lean phase is larger than that in the rich phase. Furthermore, when the lean phase is in contact with the membrane surface, the boundary layer thickness is smaller than when the rich phase is in contact with the membrane surface. In addition, from the phase diagram of the [N 4444 ][TMBS] aqueous solution, it is estimated that the H 2 O concentrations of the lean and rich DS phases are 48.8 mol/L and 23.6 mol/L, respectively. Therefore, when the lean phase covers the membrane surface, the driving force for H 2 O diffusion in the boundary layer near the membrane surface would be large. Conversely, when the rich phase covers the membrane surface, the driving force would be small. Thus, it can be considered that the diffusion rate of H 2 O via the boundary layer of the membrane surface is higher when the lean phase is in contact with the membrane surface. Consequently, the water flux is drastically decreased, by approximately 40 wt%, upon changing the phase configuration.
According to the phase separation theory [42,43], at DS concentrations of 12 to 58 wt%, the volume ratio of the rich phase to the lean phase increases with increasing DS concentration. However, the concentrations of the lean and rich phases were maintained at 12 wt% and 58 wt%, respectively. Thus, the water fluxes were almost constant, at high levels, for DS concentrations of 12 wt% to 40 wt%. In addition, they were also constant at low levels for DS concentrations of 40 wt% to 58 wt%, as shown in Figure 5.
The change in the [N 4444 ][TMBS] aqueous solution near the RO membrane surface from the lean phase to the rich phase is the reason for the rapid decrease in water flux. In this context, it is considered that the RO efficiency could be improved by surface modification of the RO membranes. That is, if an RO membrane with a highly hydrophilic surface can be developed, the DS-lean phase will preferentially cover the hydrophilic surface, even in the case of the isolated configuration of the DS-lean phase, and a high-water flux can be maintained over a wide concentration range. Improvement of the water flux by surface modification of RO membranes is a topic for future research.

Conclusions
The effect of high-temperature RO on water recovery and DS regeneration was investigated using thermo-responsive IL-DSs. Specifically