Scaling Reduction in Carbon Nanotube-Immobilized Membrane during Membrane Distillation

: Membrane distillation (MD) is fast evolving as a desalination technology for high-salinity waters where scaling remains a major challenge. This paper reports the scaling reduction in carbon nanotube-immobilized membranes (CNIMs) and by the use of the antiscalant polyacrylic acid. High concentrations of CaSO 4 , CaCO 3 , and BaSO 4 were deliberately used to initiate scaling on the membranes. It was observed that after ten hours of operation in a highly scaling CaSO 4 environment, the CNIM showed 127% higher ﬂux than what was observed on a membrane without the CNTs. The trends were similar with CaCO 3 and BaSO 4 , where the CNIM showed signiﬁcantly improved antiscaling behavior. The normalized ﬂux declination for CNIM was found to be 45%, 30%, and 53% lower compared to the pristine membrane with CaSO 4 , CaCO 3 , and BaSO 4 solutions, respectively. The use of antiscalant in the feed solution was also found to be e ﬀ ective in improving antiscaling behavior, which reduced salt deposition up to 28%, and the water vapor ﬂux was 100% and 18% higher for the pristine polypropylene and CNIM, respectively. Results also showed that the presence of CNTs facilitated the removal of deposited salts by washing, and the CNIM regained 97% of its initial water ﬂux, whereas the polypropylene only regained 85% of the original value. CaCO 2.5 for BaSO 4 salt Antiscalant (50 mg) was added in 1 L of feed solution for the experiment. The data were reported after repeating each experiment at least three conﬁrm reproducibility. of CNIM using TGA. The TGA curves of the two membranes in Figure It can be seen from the TGA curves that both membranes are quite stable within the operating temperature ranges. The initial weight loss of the membrane was started at ~270 ◦ C and it completely decomposed at ~380 ◦ C. The TGA curve for CNIM shifted slightly upward, which exhibited the enhanced thermal stability of the CNIM due to the presence of CNTs [47]. The contact angles of the unmodiﬁed PP and CNIM for pure water and salt solution are shown as above in Figure 4. A droplet size of 4 mm was used to measure contact angles. The presence of CNTs dramatically altered the contact angle. With 100% water in the feed, the contact angle for CNIM was higher (125 ◦ ) than the unmodiﬁed PP membrane (115 ◦ ). In presence of salt, the contact angle was slightly reduced. The liquid entry pressure (LEP) was measured using a previously described method [42]. The LEP of the pure water solution was found to be 207 and 186 kPa, which changed to 179 and 165 kPa with CaSO 4 salt solution (2.95 g / L) for unmodiﬁed PP and CNIM, respectively.


Introduction
Reverse osmosis (RO), multistage flash, and multieffect distillation are the most common desalination techniques and have shown much promise. However, they have limitations such as fouling of RO membranes in the presence of scaling ions, and high energy and capital cost in thermal methods [1][2][3]. As a result, alternative desalination technologies, including solar evaporation and membrane distillation (MD), are being explored [4][5][6][7]. The MD process paired with solar energy or low-grade heat source can be an attractive alternate to the conventional membrane-based desalination [8,9]. MD is also evolving to be an effective desalination technique for treating the high-salinity water that RO is unable to handle due to high osmotic pressure and extensive pretreatment requirements [10][11][12].
A major obstacle in membrane-based desalination techniques is fouling [13][14][15][16] from the deposition of suspended or dissolved substances on the active membrane surface and/or within its pores [17][18][19]. Several types of fouling are common, including inorganic fouling or scaling, particulate and colloidal fouling, organic fouling, and biofouling [20][21][22]. Fouling has also been an important issue in RO, nanofiltration, and ultrafiltration [23][24][25]. Due to the use of porous hydrophobic membranes, the fouling in MD tends to be significantly less than in RO, but is still an important consideration for high-salinity water desalting. Several approaches for fouling reduction, such as the introduction of nanobubble, ultrasound, and microwave irradiation have been reported [26][27][28][29][30]. Of particular interest has been the deposition of inorganic salts such as calcium carbonate (CaCO 3 ), calcium sulfate (CaSO 4 ), and barium sulfate (BaSO 4 ), which are found in hard waters as well as industrial waste from power plants, hydraulic fracturing, and wastewaters from industries such as textiles and pulp and paper [31]. The formation of a scaling layer comprising of salt crystals formed on/within the pores of MD membranes are known to cause progressive wettability of the membrane [32]. The reduction of scaling [33] by adding several antiscalants has been reported for RO and thermal distillation processes. The use of antiscalants could potentially help to lower the scaling in thermally driven MD process without any adverse effect [34][35][36].
We have described the development of a carbon nanotube-immobilized membrane (CNIM), where the carbon nanotubes (CNTs) increased the partitioning of water vapor while rejecting the liquid phase, leading to a dramatic increase in MD flux [37][38][39][40]. Besides this, it is expected that the presence of CNTs may reduce scale formation on membrane surfaces where the CNTs serve as a screen. This may resist membrane pore blocking by salt deposition, and it is conceivable that the salt crystals deposited on the screen-like CNTs can be removed or washed off rather easily (as shown in Figure 1). In this way, the CNT screen maintains pore opening for the permeation of water vapor while repelling liquid water and salt clusters for longer periods of time. The objective of this project was to study the scaling behavior of the CNIM for various highly concentrated feed mixtures and to evaluate the performance of CNIM with addition of antiscalant. industrial waste from power plants, hydraulic fracturing, and wastewaters from industries such as textiles and pulp and paper [31]. The formation of a scaling layer comprising of salt crystals formed on/within the pores of MD membranes are known to cause progressive wettability of the membrane [32]. The reduction of scaling [33] by adding several antiscalants has been reported for RO and thermal distillation processes. The use of antiscalants could potentially help to lower the scaling in thermally driven MD process without any adverse effect [34][35][36].
We have described the development of a carbon nanotube-immobilized membrane (CNIM), where the carbon nanotubes (CNTs) increased the partitioning of water vapor while rejecting the liquid phase, leading to a dramatic increase in MD flux [37][38][39][40]. Besides this, it is expected that the presence of CNTs may reduce scale formation on membrane surfaces where the CNTs serve as a screen. This may resist membrane pore blocking by salt deposition, and it is conceivable that the salt crystals deposited on the screen-like CNTs can be removed or washed off rather easily (as shown in Figure 1). In this way, the CNT screen maintains pore opening for the permeation of water vapor while repelling liquid water and salt clusters for longer periods of time. The objective of this project was to study the scaling behavior of the CNIM for various highly concentrated feed mixtures and to evaluate the performance of CNIM with addition of antiscalant.
Multiwalled CNTs were purchased from Cheap Tubes, Inc. (Brattleboro, VT, USA). The MWCNTs were functionalized with carboxylic acid groups in a Microwave Accelerated Reaction System (Model: CEM Mars, CEM Corporation, Matthews, NC, USA) in our laboratory. The PP membrane was used as the base membrane to fabricate the carbon nanotube-immobilized membrane (CNIM). The CNTs are immobilized on the membrane surface using small amount of polyvinylidene fluoride (PVDF) as binder. Excess PVDF was removed from the surface by washing with acetone after fabrication. The functionalization process and CNIM fabrication methods have been previously
Multiwalled CNTs were purchased from Cheap Tubes, Inc. (Brattleboro, VT, USA). The MWCNTs were functionalized with carboxylic acid groups in a Microwave Accelerated Reaction System (Model: CEM Mars, CEM Corporation, Matthews, NC, USA) in our laboratory. The PP membrane was used as the base membrane to fabricate the carbon nanotube-immobilized membrane (CNIM). The CNTs are immobilized on the membrane surface using small amount of polyvinylidene fluoride (PVDF) as binder. Excess PVDF was removed from the surface by washing with acetone after fabrication. The functionalization process and CNIM fabrication methods have been previously reported [41][42][43]. Our previous studies have already proven the ability to retain the CNT coating on the surface for longer periods of time [44,45]. Figure 2 demonstrates the DCMD experiment setup in the laboratory. The details have been previously described [29]. The hot aqueous feed solutions were circulated on one side of the membrane in the DCMD cell. The initial salt concentrations in the feed were 2.95 g/L for CaSO 4 solution, 3.5 g/L for CaCO 3 , and 2.5 g/L for BaSO 4 salt solution. Antiscalant (50 mg) was added in 1 L of feed solution for the experiment. The data were reported after repeating each experiment at least three times to confirm reproducibility. reported [41][42][43]. Our previous studies have already proven the ability to retain the CNT coating on the surface for longer periods of time [44,45]. Figure 2 demonstrates the DCMD experiment setup in the laboratory. The details have been previously described [29]. The hot aqueous feed solutions were circulated on one side of the membrane in the DCMD cell. The initial salt concentrations in the feed were 2.95 g/L for CaSO4 solution, 3.5 g/L for CaCO3, and 2.5 g/L for BaSO4 salt solution. Antiscalant (50 mg) was added in 1 L of feed solution for the experiment. The data were reported after repeating each experiment at least three times to confirm reproducibility.

Hot Water
Cold Water Scanning electron microscopy (SEM) (JEOL, Model JSM-7900F, JEOL USA Inc.; Peabody, MA, USA) was used to characterize the surface morphology of the PP membrane and the CNIM before and after the experiment. Thermogravimetric analysis (TGA, PerkinElmer Pyris 7 TGA system at a heating rate of 10 °C/min in air) was used to investigate the thermal stability of the membranes.

DCMD Performance
The MD performances of CNIM membrane with and without the antiscalant were studied as a function of time, temperature, and feed flow rate. The water vapor flux, , is measured as where is the mass of permeated water in time t through surface area . The flux can also be denoted as The overall mass transfer coefficient ( ) was computed as where and are the feed and permeate side water vapor pressure, respectively. To compare scaling between the PP membrane and CNIM, the flux was measured over time and the normalized flux decline, , was determined as Scanning electron microscopy (SEM) (JEOL, Model JSM-7900F, JEOL USA Inc.; Peabody, MA, USA) was used to characterize the surface morphology of the PP membrane and the CNIM before and after the experiment. Thermogravimetric analysis (TGA, PerkinElmer Pyris 7 TGA system at a heating rate of 10 • C/min in air) was used to investigate the thermal stability of the membranes.

DCMD Performance
The MD performances of CNIM membrane with and without the antiscalant were studied as a function of time, temperature, and feed flow rate. The water vapor flux, Jw, is measured as where Wp is the mass of permeated water in time t through surface area A. The flux can also be denoted as The overall mass transfer coefficient (k) was computed as where P f and P p are the feed and permeate side water vapor pressure, respectively. To compare scaling between the PP membrane and CNIM, the flux was measured over time and the normalized flux decline, FDn, was determined as where J 0 and J f are the respective initial and final permeate flux over a period of time, t. Figure 3a,b shows the surface SEM images of the original PP membrane and CNIM. Figure 3a illustrates the pores present on PP membrane surface, and the incorporation of the carboxylated CNTs led to a modification in morphology as can be seen from Figure 3b. AFM images from our previous studies have shown that the incorporation of CNTs on membrane surface also increases the surface roughness [45]. The gas permeation test of the membranes demonstrated no significant change in the effective surface porosity over the effective pore length of the membranes as only a small quantity of CNTs were immobilized on the membrane surface [42,46]. where and are the respective initial and final permeate flux over a period of time, t. Figure 3a,b shows the surface SEM images of the original PP membrane and CNIM. Figure 3a illustrates the pores present on PP membrane surface, and the incorporation of the carboxylated CNTs led to a modification in morphology as can be seen from Figure 3b. AFM images from our previous studies have shown that the incorporation of CNTs on membrane surface also increases the surface roughness [45]. The gas permeation test of the membranes demonstrated no significant change in the effective surface porosity over the effective pore length of the membranes as only a small quantity of CNTs were immobilized on the membrane surface [42,46].

Characterization of the Membranes
(c) The thermal stability of the PP membrane and CNIM was studied using TGA. The TGA curves of the two membranes are shown in Figure 3c. It can be seen from the TGA curves that both membranes are quite stable within the operating temperature ranges. The initial weight loss of the membrane was started at ~270 °C and it completely decomposed at ~380 °C. The TGA curve for CNIM   The thermal stability of the PP membrane and CNIM was studied using TGA. The TGA curves of the two membranes are shown in Figure 3c. It can be seen from the TGA curves that both membranes are quite stable within the operating temperature ranges. The initial weight loss of the membrane was started at~270 • C and it completely decomposed at~380 • C. The TGA curve for CNIM shifted slightly upward, which exhibited the enhanced thermal stability of the CNIM due to the presence of CNTs [47].
The contact angles of the unmodified PP and CNIM for pure water and salt solution are shown as above in Figure 4. A droplet size of 4 mm was used to measure contact angles. The presence of CNTs dramatically altered the contact angle. With 100% water in the feed, the contact angle for CNIM was higher (125 • ) than the unmodified PP membrane (115 • ). In presence of salt, the contact angle was slightly reduced. The liquid entry pressure (LEP) was measured using a previously described method [42]. The LEP of the pure water solution was found to be 207 and 186 kPa, which changed to 179 and 165 kPa with CaSO 4 salt solution (2.95 g/L) for unmodified PP and CNIM, respectively.
Water 2019, 11, x FOR PEER REVIEW 5 of 16 shifted slightly upward, which exhibited the enhanced thermal stability of the CNIM due to the presence of CNTs [47]. The contact angles of the unmodified PP and CNIM for pure water and salt solution are shown as above in Figure 4. A droplet size of 4 mm was used to measure contact angles. The presence of CNTs dramatically altered the contact angle. With 100% water in the feed, the contact angle for CNIM was higher (125°) than the unmodified PP membrane (115°). In presence of salt, the contact angle was slightly reduced. The liquid entry pressure (LEP) was measured using a previously described method [42]. The LEP of the pure water solution was found to be 207 and 186 kPa, which changed to 179 and 165 kPa with CaSO4 salt solution (2.95 g/L) for unmodified PP and CNIM, respectively.

Effect of Temperature and Feed Flowrate on the Water Vapor Flux
The influence of feed solution temperature on water permeation rate for PP membrane and CNIM with pure water and aqueous CaSO4 solution (2.95 g/L) at feed and distillate flow rate 200 mL min −1 is shown in Figure 5a. The legend key PP-AS or CNIM-AS in the graph denotes the membrane performance using AS in the feed solution. In all cases, the initial permeate flux followed a direct relationship with temperature as the vapor pressure gradient increased with temperature. Among the two membranes, CNIM exhibited higher flux at any particular temperature, which was in line with our previously reported results [42]. It is important to note that the addition of PAA (antiscalant) did not show any negative effect on water permeation. Similar water vapor flux was observed for both membranes with antiscalant when pure water was used as feed. The mass transfer coefficient (k) was observed to be enhanced for CNIM (5.67 × 10 −4 and 5.01 × 10 −4 kg/m 2 •s•kPa, for CNIM and PP membrane, respectively) compared to the PP membrane. However, the addition of antiscalant in pure water feed did not show any significant change in the mass transfer coefficient for both membranes. At such high concentrations, the CaSO4 was expected to quickly foul the membrane and the temperature dependence graph did not show an exponential increment pattern as observed using pure water as feed, as the scaling rate increased at higher temperatures [1,48]. A slight increase in initial flux was due to the antiscaling effect of antiscalant for short time periods (1 h

Effect of Temperature and Feed Flowrate on the Water Vapor Flux
The influence of feed solution temperature on water permeation rate for PP membrane and CNIM with pure water and aqueous CaSO 4 solution (2.95 g/L) at feed and distillate flow rate 200 mL min −1 is shown in Figure 5a. The legend key PP-AS or CNIM-AS in the graph denotes the membrane performance using AS in the feed solution. In all cases, the initial permeate flux followed a direct relationship with temperature as the vapor pressure gradient increased with temperature. Among the two membranes, CNIM exhibited higher flux at any particular temperature, which was in line with our previously reported results [42]. It is important to note that the addition of PAA (antiscalant) did not show any negative effect on water permeation. Similar water vapor flux was observed for both membranes with antiscalant when pure water was used as feed. The mass transfer coefficient (k) was observed to be enhanced for CNIM (5.67 × 10 −4 and 5.01 × 10 −4 kg/m 2 ·s·kPa, for CNIM and PP membrane, respectively) compared to the PP membrane. However, the addition of antiscalant in pure water feed did not show any significant change in the mass transfer coefficient for both membranes. At such high concentrations, the CaSO 4 was expected to quickly foul the membrane and the temperature dependence graph did not show an exponential increment pattern as observed using pure water as feed, as the scaling rate increased at higher temperatures [1,48]. A slight increase in initial flux was due to the antiscaling effect of antiscalant for short time periods (1 h). At a temperature of  Figure 5b illustrates the performance of the PP and CNIM as a function of feed flow rate at 100, 150, and 200 mL/min, while the permeate side flow rate was maintained constant at 200 mL/min at a feed temperature of 70 °C. Results indicate that the water vapor flux was enhanced by feed flow rate for all membranes. With increase in feed flow rate from 100 to 200 mL/min, the Reynolds number increased from 662 to 1324, which improved the heat transfer efficiency from the bulk to the membrane surface [49]. The increase in feed flow rate reduced the boundary layer resistance at the feed solution-membrane interface, and generated more water vapor per unit time [50]. The use of antiscalants in feed solution with CNIM did not show any significant effect with respect to flow rate. The conductivity of the permeated water was found 2-10 µS/cm and salt rejection was determined to be high (>99%) for both membranes.

Membrane Scaling
The scaling of PP membrane and CNIM was studied with highly saline feed solutions, namely CaSO4 (2.95 g/L), CaCO3 (3.5 g/L), and BaSO4 (2.5 g/L). The high concentrations were deliberately  Figure 5b illustrates the performance of the PP and CNIM as a function of feed flow rate at 100, 150, and 200 mL/min, while the permeate side flow rate was maintained constant at 200 mL/min at a feed temperature of 70 • C. Results indicate that the water vapor flux was enhanced by feed flow rate for all membranes. With increase in feed flow rate from 100 to 200 mL/min, the Reynolds number increased from 662 to 1324, which improved the heat transfer efficiency from the bulk to the membrane surface [49]. The increase in feed flow rate reduced the boundary layer resistance at the feed solution-membrane interface, and generated more water vapor per unit time [50]. The use of antiscalants in feed solution with CNIM did not show any significant effect with respect to flow rate. The conductivity of the permeated water was found 2-10 µS/cm and salt rejection was determined to be high (>99%) for both membranes.

Membrane Scaling
The scaling of PP membrane and CNIM was studied with highly saline feed solutions, namely CaSO 4 (2.95 g/L), CaCO 3 (3.5 g/L), and BaSO 4 (2.5 g/L). The high concentrations were deliberately selected so that the membranes would foul quickly. The membrane scaling was evaluated by the reduction of permeated water flux with time during the operation. Figure 6a- Figure 6 depicts the decline in water vapor fluxes with time for both the membranes as an outcome of scaling. It is clear from the figures that the CNIM exhibited higher antiscaling properties in comparison with PP membrane. This may be due to additional screening effect of CNTs which reduced pore blocking from salt deposition (as shown in Figure 1). Figure 6a shows the short-term scaling behavior of CaSO4 salt solutions with PP membrane and CNIM with and without antiscalant. The water vapor flux declined with time as expected. However, the use of antiscalant lowered the scaling tendency, hence generating higher water vapor flux.    Figure 1). Figure 6a shows the short-term scaling behavior of CaSO 4 salt solutions with PP membrane and CNIM with and without antiscalant. The water vapor flux declined with time as expected. However, the use of antiscalant lowered the scaling tendency, hence generating higher water vapor flux.
It can be seen from Figure 6b that for CaSO 4 solution, the flux declined to 13.4 from 39.4 kg/m 2 ·h for PP, and to 30.5 from 47.6 kg/m 2 ·h for the CNIM after 10 h of operation. The results show that by using CNIM, the water vapor flux after 10 h was still 126.7% higher than the PP membrane. The lower scaling tendency of CNIM can be explained via its 'screening effect', where the net-like presence of CNTs prevent salt clusters from depositing on the membrane surface or pores. Furthermore, the experiments were carried out using antiscalant (PAA) to study its effect on CNIM. It was observed that the use of AS in the feed solution improved the antiscaling behavior of the membranes and the water vapor flux after 10 h of operation was 26.8 and 36 kg/m 2 ·h for PP membrane and CNIM, respectively, which is 100% and 18% higher compared to the system without AS. This may be due to the fact that the antiscalant delays the clustering process and prevents the precipitation of salt on the membrane surface [51][52][53][54]. The presence of AS was found to be more effective with the PP membrane than the CNIM.
Similar trends were observed with the other two salts and are shown in Figure 6c,d. As can be seen from Figure 6c Figure 6d), the flux declined to 23.7 from 47.7 kg/m 2 ·h for PP membrane, and to 37.7 from 49.4 kg/m 2 ·h with CNIM. The results exhibited higher membrane antiscaling performance using CNIM and the water vapor flux was found to be 59% higher for CNIM compared to PP membrane after 10 h of operation. However, by using the antiscalant materials with CNIM, the membrane scaling was reduced further and the flux increased bỹ 18%. Among all thee salts, CaSO 4 fouled the membrane most drastically and the use of AS in the feed solution also was found to be more effective for CaSO 4 . In general, by using CNIM, the scaling on the membrane surface was reduced and the MD performance enhanced. The addition of AS further increased the antiscaling properties of the system.
The normalized flux declination (FDn) for different membrane systems with various salt solutions are shown in Table 1. It is clear from the table that the CNIM exhibited lower flux decline for all salts compared to PP, indicating a clear lowering of the scaling tendency. The use of AS further improved the antiscaling property of both membranes. For CaSO 4 , the flux decline for CNIM-AS and CNIM were found to be 30.8% and 36%, respectively, which were 23% and 45% lower than that of the PP membrane.

Deposition of Salts on the Membranes
The deposition of the salt crystals on the membrane surface was measured by weighing the amount of salt on the membrane before and after the 6 h run, and then drying the membrane overnight in an oven at 70 • C. The weight measurements were done very carefully to avoid any loss of deposited salt from the surface. The results are shown in Table 2. From Table 2, it is clear that the amount of salts deposited on the membrane surface was lesser for CNIM than the PP membrane for all cases. The lower salt deposition on CNIM may be attributable to the screening effect of CNTs [55]. The table also demonstrates the advantage of using antiscalant in reducing the salt deposition on the membrane surface [56,57]. The AS influenced the PP membrane-based desalination slightly more than the CNIM, where the salt deposition was reduced by 25.6% and 28.2%, respectively, for CNIM and PP for CaSO 4 .
SEM images of the deposition of various salt crystals on different membrane surfaces with and without using antiscalant are shown in Figure 7. The SEM images clearly show the difference in crystal configuration of different salt clusters. It is also revealed from the images that the use of AS significantly reduced the scaling layer on the membrane surface [1,35,58]. The foulants interact with each other and with the membrane surface to form deposits [59][60][61][62][63], The antiscalant interact with the foulants and with the membrane surface to breakdown the salt crystals (foulants) and reduce the scaling as is shown in Figure 7a-c for all salts.   Table 2, it is clear that the amount of salts deposited on the membrane surface was lesser for CNIM than the PP membrane for all cases. The lower salt deposition on CNIM may be attributable to the screening effect of CNTs [55]. The table also demonstrates the advantage of using antiscalant in reducing the salt deposition on the membrane surface [56,57]. The AS influenced the PP membrane-based desalination slightly more than the CNIM, where the salt deposition was reduced by 25.6% and 28.2%, respectively, for CNIM and PP for CaSO4.
SEM images of the deposition of various salt crystals on different membrane surfaces with and without using antiscalant are shown in Figure 7. The SEM images clearly show the difference in crystal configuration of different salt clusters. It is also revealed from the images that the use of AS significantly reduced the scaling layer on the membrane surface [1,35,58]. The foulants interact with each other and with the membrane surface to form deposits [59][60][61][62][63], The antiscalant interact with the foulants and with the membrane surface to breakdown the salt crystals (foulants) and reduce the scaling as is shown in Figure 7a-c for all salts.

Membrane Regeneration and Stability
The regenerability of the fouled membranes using CaSO 4 as feed were studied with or without AS in the system. The MD experimentations were continued for 6 h followed by washing of the fouled membrane with circulating DI water at 70 • C for 30 min and then the MD experiments were continued again for another 6 h with the freshly prepared feed mixture. The water vapor flux after washing was compared with the initial flux. Table 3 shows the regenerability of the membranes with CaSO 4 with and without AS. It is clear from the table that the CNIM was able to remove the deposited salts and attained around 97% of its initial water vapor flux. By contrast, the PP membrane only reached up to 85% of its original value, which clearly demonstrated the superiority of CNIM in terms of membrane regeneration. The outcomes were further confirmed by using SEM and measuring the weight of the salts that remained on the membrane surface after washing. Figure 8a-d show the surface SEM images of the CaSO 4 fouled PP membrane and CNIM with and without AS after washing. The salt precipitation was found to be less on the CNIM surface after washing, which further demonstrated the washability and regenerability of the CNIM. BaSO4 deposition on (i) PP, (ii) PP-AS, (iii) CNIM, and (iv) CNIM-AS after running the experiment given in Figure 6d.

Membrane Regeneration and Stability
The regenerability of the fouled membranes using CaSO4 as feed were studied with or without AS in the system. The MD experimentations were continued for 6 h followed by washing of the fouled membrane with circulating DI water at 70 °C for 30 min and then the MD experiments were continued again for another 6 h with the freshly prepared feed mixture. The water vapor flux after washing was compared with the initial flux. Table 3 shows the regenerability of the membranes with CaSO4 with and without AS. It is clear from the table that the CNIM was able to remove the deposited salts and attained around 97% of its initial water vapor flux. By contrast, the PP membrane only reached up to 85% of its original value, which clearly demonstrated the superiority of CNIM in terms of membrane regeneration. The outcomes were further confirmed by using SEM and measuring the weight of the salts that remained on the membrane surface after washing. Figure 8a-d show the surface SEM images of the CaSO4 fouled PP membrane and CNIM with and without AS after washing. The salt precipitation was found to be less on the CNIM surface after washing, which further demonstrated the washability and regenerability of the CNIM.  Table 4 shows the amount of salt remained on the membrane surface after washing the fouled membranes. The result showed the amount of salt remained on the membrane surface was significantly lower than that of the fouled membrane. Among PP and CNIM, the CNIM showed higher washability of the deposited salts. While running the experiment with CNIM, no CNTs were detected in the feed solutions that have been recycled. The CNIM was also run with aqueous solutions for 30 days (6 h per day) at 70 • C and then inspected to see is there was any loss of CNTs. However, any appreciable loss of CNTs was not observed, which demonstrates the ability of CNTs to be retained on the membrane surface [45].

Conclusions
This paper successfully demonstrates the enhanced antiscaling behavior of CNIM over pristine PP membrane. The addition of antiscalant to the feed solution resulted in a reduction in scaling on the surface of the membrane. The desalination performance of PP and CNIM were compared. The CNIM exhibited lower flux decline for all the salts. The washability and regenerability of the CNIM was observed to be superior to the pristine PP membrane. The addition of antiscalant materials and use of CNIM in MD was found to be highly effective in enhancing the membrane performance and membrane regenerability when treating with high concentration scaling salts. Funding: This study was partially supported by a grant from the Chemical, Bioengineering, Environmental, and Transport Systems Division, National Science Foundation (grant number CBET-1603314) and King Abdullah Scholarship Program, Ministry of Higher education, Kingdom of Saudi Arabia for their financial support. The work also partially supported from the Ada Fritts Chair position funding at NJIT. The Otto York Center for membrane characterization was also acknowledged for the use of instrumentation.

Conflicts of Interest:
The authors declare no conflict of interest.