Enhanced desalination using a three-layer OTMS based superhydrophobic membrane for a membrane distillation process

Superhydrophobic membranes are essential for improved seawater desalination. This study presents the successful casting of a three-layered membrane composed of a top superhydrophobic coating onto a polypropylene (PP) mat through simple sol–gel processing of octadecyltrimethoxysilane (OTMS), and the bottom layer was casted with hydrophilic poly(vinyl alcohol) (PVA) by using a knife casting technique; this membrane represents a novel class of improved-performance membranes consisting of a top superhydrophobic coating onto a hydrophobic PP mat and a hydrophilic layer (PVA) at the bottom. OTMSs are well known low-surface-energy materials that enhance superhydrophobicity, and they were observed to be the ideal chemical group for increasing the hydrophobicity of the PP mat. The PVA layer acted as base layer absorbing the condensed vapor and thus enhancing the vapor flux across the membrane. The hybrid three-layered membrane exhibited superhydrophobicity, with an average contact angle of more than 160°, and demonstrated high performance in terms of rejection and water flux. This study also examined the pore size distribution, surface roughness, surface area, tensile strength, water flux, and salt rejection of the fabricated membrane. The salt rejection level was calculated to be 99.7%, and a high permeate flux of approximately 6.7 LMH was maintained for 16 h.


Introduction
The demand for fresh water has increased gradually in the past 20 years. Membrane distillation (MD) is one of the most effective technologies for seawater desalination. MD is a thermally driven separation process in which only vapor molecules can pass through a porous hydrophobic membrane. 1,2 Typically, MD possesses many unique features such as requiring lower operating temperatures compared with those encountered in conventional processes, and placing lower demands on membrane mechanical strength. In addition, the hydrostatic pressure encountered in MD is much lower than that in pressure-driven membrane processes such as reverse osmosis. Hence, MD is a cost-effective process that places lower demands on membrane characteristics. [2][3][4] Hydrophobic materials such as polyvinylidene uoride (PVDF), polypropylene (PP), and polytetrauoroethylene (PTFE) are generally utilized in the MD process, and they are fabricated through processes such as stretching, electrospinning, thermally induced phase separation, and phase inversion. 5,6 Direct contact membrane distillation (DCMD) conguration is a type of MD in which an aqueous solution at a lower temperature is in direct contact with the permeate stream of the membrane. DCMD conguration has been widely studied because of its convenience and simplicity. 7 To prevent membrane wetting and the subsequent formation of liquid-lled pores with higher mass transfer resistance, hydrophobic membranes composed of PVDF and PTFE are commonly utilized in MD. 8 Commercial membranes such as PP, PTFE, and PVDF demonstrate higher hydrophobicity and chemical and thermal resistance, compared with other membranes. Recently, membrane researchers have further modied these MD membranes into superhydrophobic membranes to avoid membrane fouling by reducing the direct contact between membrane foulants and membrane surface at the Cassie-Baxter state. 9 Superhydrophobic-specialized surfaces with versatile features have gained increased attention in the eld of desalination and ltration. In general, MD processes must generate pure water with lower conductivity continuously. However, liquid penetration and pore wetting during long-term operation lead to lower salt rejection. According to some researchers, an increase in the hydrophobicity of MD membranes can reduce membrane pore wetting effectively. 10,11 Therefore, researchers are attempting to develop superhydrophobic membranes by utilizing hydrophobic additives. 12 Superhydrophobicity introduces an air gap between water molecules as well as on the membrane surface, which can potentially increase the allowable pore sizes before pore wetting occurs, resulting in a higher mass ux. 13 In addition, superhydrophobic modication could reduce heat loss through conduction across the membrane 14 .
In this study, a three-layered membrane was fabricated for enhanced desalination in MD applications. Octadecyltrimethoxysilane (OTMS) is well known as a low-surface-energy material. OTMS have 18 carbon atom hydrocarbon chains and tail end group of OTMS molecule is the hydrophobic alkylsilane group, and the monolayers of OTMS self-assemble on a solid surface, with their molecular axis perpendicular to the surface. 15 Typically, OTMS layers are applied to lower the surface energy of solid surfaces to render superhydrophobic characteristics. OTMS materials are utilized as one of the coprecursors in solgel processing to achieve superhydrophobic coatings or layers. 16,17 In this study, a simple and single-step sol-gel processing of long-chain OTMS was performed to fabricate a superhydrophobic layer on a PP mat. The fabricated coatings were transparent, were chemically durable, and exhibited an attractive self-cleaning property. The OTMS coating was the topmost layer and was determined to be superhydrophobic, the PP mat was the middle layer and was observed to be hydrophobic as the contact angle was found to be 99.5 and in addition to that the pore size and thickness was evaluated to be 20 nm and 25 mm respectively. Poly(vinyl alcohol) (PVA) was the bottom layer and was determined to be hydrophilic; the PVA layer was casted by using a knife casting technique. Here, as mentioned earlier OTMS coating was the top superhydrophobic layer as the contact angle was determined to be more than 160 , the PP mat acted as a porous supportive layer improving the mechanical stability during the long-term MD operation. Additionally, the PVA layer could absorb condensed vapor, thus directly enhancing the vapor ux across the membrane. 18 The hydrophilic bottom layer could facilitate the absorption of the condensed vapor from the superhydrophobic top layer, thus preventing pore wetting. Moreover, the combination of the superhydrophobic/hydrophilic layer could provide low resistance to mass ux and also prevent conductive heat loss. Therefore, the three-layered OTMS-PP/PVA contributed to improved mass ux. Fig. 1 demonstrates the detailed mechanism of the three-layered surface membrane composed of OTMS-PP/PVA. The gure also indicates the reasons for placing the superhydrophobic OTMS-PP layer on the feed solution side.
Recently in market, microporous hydrophobic membranes made of PP, PTFE, and PVDF are commercially available. Typically, PTFE is the most ideal polymer for MD membrane because of its hydrophobic characteristic, good chemical resistance and thermal stability. Various commercially available membranes used in MD process has been summarised in Table 1. The general properties of the proposed membrane have been compared with that of other commercially available membranes based on membrane thickness, average pore size, contact angle. The present composite membrane achieves comparable membrane thickness and average pore size which makes it feasible for direct application in MD system.
The fabrication of the composite membrane is simple and includes only three steps: (a) deposition of PP mat in OTMS solution and dried for 12 h at room temperature; (b) PVA solution was used to fabricate a hydrophilic layer by utilizing knife casting device and (c) the composite membrane OTMS-PP/ PVA undergone heat-pressing treatment. An elaborate discussion has been provided in Section 2.2.
Current research has focused on improving membranes by modifying the surface chemistry. 23,24 The current study examined factors such as high permeate water ux and higher salt rejection ($99%); therefore, it also conrmed the feasibility of the proposed three-layered OTMS-PP/PVA membranes for MD applications. The overall MD performance of the fabricated membrane was carefully analyzed using a thermally driven MD process to calculate the permeate water ux and salt rejection of 30 000 ppm NaCl aqueous solution. Furthermore, this paper reports the fundamental concepts for casting other novel three and bilayered superhydrophobic membranes for MD applications.

Starting material
OTMS with a molecular weight of 586.30 was ordered from Matrix Scientic. PVA with a molecular weight of M n 146 000-186 000 and 99.9% hydrolyzed was purchased from Sigma-Aldrich. Furthermore, PP mats (non-commercial) were provided by the BenQ Material Corporation, Taiwan.

Experimental
2.2.1 Preparation of superhydrophobic material. First, 1 mL of distilled water and 0.5 mL of ammonia were mixed thoroughly with 25 mL of ethanol and stirred for 10 min in a magnetic stirrer. Subsequently, 200 mL of OTMS was slowly added to this solution under constant stirring. Ethanol was utilized as a solvent, whereas ammonia was applied as a catalyst to initiate the sol-gel process, However, water was used to accelerate the hydrolysis reaction. Typically, a functional methoxy group of OTMS undergoes hydrolysis and polycondensation reactions to form network-like structures. Finally, aer 10 min, a homogenous solution appeared, conrming the network structure formation in the sol.
2.2.2 Preparation of superhydrophobic coating by sol-gel process. First, the PP mats were cleaned thoroughly and then immersed in the aforementioned solution for deposition times of 5, 15, and 30 min. The superhydrophobic coatings prepared with deposition times of 5, 15, and 30 min were named as OTMS (5), OTMS (15), and OTMS(30), respectively. A thin, watery coating was formed on the precleaned PP mats. Finally, the superhydrophobic coatings were dried for 12 h at room temperature.
2.2.3 Preparation of PVA solution for knife casting. PVA was dissolved in distilled water of 8% w/v concentration inside a water bath maintaining a temperature of up to 120-140 C under constant stirring of 700-800 rpm in a magnetic stirrer. A thick solution was obtained aer maintaining the same condition for 10-12 h. Because the molecular weight of PVA is high, dissolving PVA polymers higher than 8% w/v is very difficult. 25 The controlled conditions for the membrane fabrication are mentioned in Table 2.
2.2.4 Membrane distillation application. MD was performed at lab scale as indicated in Fig. 2. The MD experimental test cell lab-scale setup was acquired from Sterlitech Corporation Limited. The observations were carefully analyzed for different durations from 1 until 16 h, and the results were evaluated for different temperature differences. Fig. 2 presents the schematic lab setup of the MD process.
The water ux J w (L m À2 h À1 ; LMH) was evaluated in terms of the total volume of the permeate water by using eqn (1) 2,26-28 where V is the total volume of the permeate over a specic time period Dt (h) and A is the effective surface area of the membrane  utilized in the MD process. The salt rejection of the system, indicated as the percentage of NaCl retained by the membrane, was measured by using eqn (2). 26 where C P (g L À1 ) is the solute concentration in the permeate stream side and C f (g L À1 ) is the solute concentration on the feed stream side.

Characterization
The surface and cross-sectional morphologies were thoroughly examined through scanning electron microscopy (SEM; JEOL JSM-5900, Japan). The average contact angle of the membrane was studied using an optical surface analyzer (OSA60-G, Ningbo NB Scientic Instruments Co., Ltd., China) to measure its hydrophobicity and hydrophilicity. The surface roughness was analyzed through atomic force microscopy (AFM). Membrane characteristics such as pore diameter, pore width, surface area, and pore volume were examined using an adsorption/ desorption analyzer (Micromeritics N 2 , ASAP 2020, USA).

Quality control
To obtain reliable data, the experiments were executed in triplicate and the mean values were recorded. Error bars are based on the standard errors of the three replicate results.

Results and discussions
3.1 Characterization of the membranes 3.1.1 Average contact angle evaluation. When an interface occurs in between liquid and solid, the angle between the liquid surface and the outline of the contact surface is indicated as the contact angle. Generally, contact angle is the measure of wettability of a solid by liquid. In complete wetting, the contact angle tends to 0 , whereas, the solid is wettable in between 0 and 90 , and above 90 it is considered to be non-wettable. Thus, the materials with higher contact angle (superhydrophobic) are supposed to have anti-wetting property. 29 The average contact angle of the PP mat and modied PP membranes (OTMS-PP/PVA) were carefully analyzed using an optical surface analyzer (OSA60-G, Ningbo NB Scientic Instruments Co., Ltd., China). Typically, the average contact angle is considered to illustrate the nature of the membrane surface in terms of hydrophobicity or hydrophilicity. In this study, the average contact angle of the PP mat was observed to be hydrophobic. Aer the surface chemistry modication of the PP mat with OTMS, the average contact angle demonstrated a superhydrophobic nature. The contact angle of the modied surface of the PP mat increased as the deposition time of OTMS increased from 5 to 30 min through sol-gel processing. Finally, the bottom layer was cast through a knife casting device by using PVA, which was observed to be highly hydrophilic. Fig. 3 indicates the average contact angle analysis of the membrane surfaces used in the MD application. The contact diameter was measured, and the data suggested that a higher average contact angle was correlated with a lower average contact diameter. Fig. 3 indicates that OTMS (30 min)-PP had the highest average contact angle (163 ) and the lowest contact diameter (3.0 mm).
3.1.2 SEM morphology analysis. The uniformity and morphology of the coatings were thoroughly characterised by SEM and SEM micrographs are indicated in Fig. 4. The morphology of the fabricated membranes was examined by using eld emission SEM (JOEL, JSM 7600 F, Japan). The OTMS layer must be placed on top to avoid them being covered by the PP mat. Fig. 4 indicates the morphological study results, presenting multiple SEM images to illustrate the effect of surface There is no such change in the morphology of PP mat modied with OTMS as the coating of OTMS tends to be very thin selfassembled monolayers. Thus, it can be seen that the morphology appears to be quite similar before and aer sol gel treatment technique. Additionally, the bottom PVA layer was analyzed through SEM. The cross-sectional view of the membrane is also added to this gure. This gure clearly indicates the successful fabrication of the OTMS-PP/PVA membrane.
3.1.3 Surface roughness analysis by AFM. Typically, lower surface energy and surface roughness play an important role in the wetting features of solid surfaces. The surface morphologies of the coatings of the modied membranes are depicted in Fig. 5. A wrinkled, rough, mountain-like nanostructure appeared before and aer the coating of the PP mat with OTMS. The gure indicates that the surface roughness doesn't show signicant change and increased a bit from 25.5 to 35 nm. The OTMS (30 min)-PP/PVA membrane had a maximum contact angle of 163 . Furthermore, the surface roughness lowers the total area of solid-liquid interface and increases the air pockets on the membrane surface, which makes the water droplets move more easily and hence, reduces the water sliding angles. Typically, lower the sliding angle the easier to clean the surface. 30 3.1.4 Membrane surface analysis through Barrett-Joyner-Halenda adsorption/desorption technique. The pore size distribution provides quantitative information on the pore size ranges present in a membrane sample. Typically, it indicates a more specic description of the particle sizes likely to be removed or retained by the membrane. [31][32][33] The membrane surface was examined using a Micromeritics N 2 adsorption/ desorption analyzer (ASAP 2020, USA). The data included the Langmuir surface area, Brunauer-Emmett-Teller (BET) surface area, pore volume, pore width, and pore diameter. The Langmuir and BET surface areas appeared to increase with the deposition time of OTMS onto PP/PVA; very little differences in the pore volume were observed. Fig. 6 presents the surface areato-pore volume ratio. Notably, because of the high surface areato-pore volume ratio, these novel types of fabricated MD membranes could overcome the low ux limitation of the PP mat. This effect could be observed in the permeate water ux, which is discussed in the next section.
The membrane thickness of the PP mat was measured to be 20 mm, which is very thin compared with other commercial MD membranes. Subsequently, the thickness of the fabricated membrane was observed to increase by four times. Typically, the membrane thickness has an effect on water ux and reduce the thermal resistance by lowering the heat efficiency or interface temperature difference as the membrane becomes thinner. Additionally, higher surface area to pore volume ratio plays an important role in lowering the mass transfer resistance. 34 This effect can be observed in permeate water ux, which has been discussed in next section.   7 indicates the dynamic mechanical analysis of different membranes in terms of storage modulus and loss modulus. The loss modulus is the measure of energy dissipation, though as a modulus, the mechanical strength of a material is dened by its hardness or stiffness. 35 The thickness of the surface-modied MD membranes was determined to be comparable to those of commercially available membranes for MD processes (range: 80-170 mm).
Loss modulus is the basic property to characterise the viscoelastic performance of a material. Moreover, higher storage modulus denotes more solid-like property and in simple words, higher the storage modulus indicates higher strength or mechanical rigidity. Infact, this is directly co-related to the degree of material integrity. Thus, higher the degree of material integrity the greater the storage modulus. 36 In this study, we analyzed both the pore width and pore diameter. The pore width and diameters of the PP mat and modied OTMS-PP/PVA membranes were examined using Barrett-Joyner-Halenda adsorption/desorption techniques. The differences observed in pore diameter were not signicant. However, the OTMS-PP/PVA membrane was expected to exhibit a higher salt rejection than the PP mat due to its lower pore diameter range. To support the previous results, the average pore widths of these fabricated membranes were also examined, and OTMS-PP/PVA was observed to possess a very small pore width of 30 nm compared with that of the PP mat. Fig. 8 denotes the average pore width and average pore diameter of all the fabricated membranes including the commercial PP mat. Interestingly, the adsorption/desorption average pore diameters and pore widths were found to be similar for different OTMS-PP/PVA membranes as the composition and fabrication technique of all these membranes are similar. Though the superhydrophobic OTMS coating prepared with deposition times of  Membrane wetting can be examined by average contact angle. Typically, lower pore size, higher contact angle and surface tension increase liquid entry pressure (LEP). As earlier mentioned, membrane wetting pores leads to reduced permeate quality; hence, it is necessary to utilize membranes with greater LEP value. 37 Franken et al. 38 has suggested a model to evaluate LEP value based on Cantor-Laplace equation 38 eqn (3): where LEP represents liquid entry pressure of pure water in Pa, B represents dimensionless geometrical factor which includes the irregularities of the pores (B ¼ 1 for assumed cylindrical pores), g L represents the liquid surface tension in N m À1 (in this case water at 25 C, 0.07199 N m À1 ), cos q represents the contact angle in degree, r max is the maximal pore (non-closed) radius in m. Table 3 represents the LEP values that was calculated using the Cantor-Laplace equation. Interestingly, the fabricated membranes OTMS-PP/PVA show higher degree of LEP than compared to PP mat which indirectly indicates reduced wetting probability. This effect can be clearly seen in Fig. 9 and 10.

Membrane performance in membrane distillation
Lab-scale MD was performed as shown in Fig. 2. NaCl solution with a concentration of 30 000 ppm was used as the feed stream solution. The temperature of the feed stream was maintained using a water bath and was varied between 40 and 70 C. By contrast, the temperature of the distillate (cooling) stream was controlled at 20 C by circulating a large quantity of distilled water through an air-cooling system. The membrane area was evaluated to be 100 cm 2 (dimensions: 10 cm Â 10 cm). The salt concentration of the feed stream and the permeate water quality were analyzed by a professional series and a conductivity meter/ TDS/DO (YSI Quatro, USA). The overall outcomes of the MD process indicated a steep increase in the permeate water ux with an increasing temperature difference up to 70 C (DT ¼ 50 C).
The permeate water ux varied from 3.5 to 6.7 LMH, indicating a stronger temperature inuence. OTMS-PP/PVA demonstrated the highest permeate water ux, compared with the PP mat.    Paper reveals an increase in the permeate water ux with an increase in temperature difference for the different fabricated membranes used in the MD process. The MD process was continued for up to 16 h for the different fabricated membranes to examine their long-term stability performance. However, the permeate water ux and salt rejection decreased slightly as the time increased. OTMS (30 min)-PP/PVA demonstrated the highest water ux of 6.7 LMH and salt rejection of 99.89%. From Fig. 10, we can conclude that the negligible decline in water ux over time clearly demonstrates good stability and durability of the OTMS and PVA sublayers on the PP mat. Specically, the interconnectivity and stability of these two sub layers (top layer and bottom layer) on the PP mat can be demonstrated by the stable water ux and rejection until 16 h of operation. Furthermore, the salt rejection of all the fabricated membranes was found to be 99.3-99.89%, illustrating good permeating quality on the permeate side as compared with the PP mat, which can be clearly seen in Fig. 10. Fig. 11 shows the reusability of the OTMS-PP/PVA membrane which was examined by evaluating the decrease in water ux aer physical cleaning of the membrane. In this experiment, the same membrane was reused for 16 h to analyse the longterm stability as well as reusability of the membrane. Interestingly, only 2.8% decrease in water ux was evaluated. Therefore,  it can be concluded that the reduction in water ux seems to be insignicant aer reusing it for 16 h which directly denotes the self-cleaning property of this OTMS-PP/PVA superhydrophobic membrane.

Conclusion
In this study, a three-layered membrane consisting of a top superhydrophobic coating was cast onto a PP mat through a simple one-step sol-gel processing of long-chain OTMS, and a base layer of PVA was fabricated through a knife casting technique, which has been successfully utilized for MD desalination. The PVA layer acted as a bottom layer absorbing water molecule (condensed vapor) and thus enhancing the vapor ux across the membrane. The results demonstrate that the hybrid membrane composed of a superhydrophobic coating of OTMS onto a PP mat exhibited superhydrophobicity, with an average contact angle of approximately 163 . Additionally, the membrane's morphology, average contact angle, pore size, surface area, and surface roughness were thoroughly examined and compared with those of a standard PP mat. The water ux and salt rejection percentage were also analyzed using 30 g L À1 of NaCl solution in the feed stream. Changes in the surface chemistry of the modied three-layered OTMS-PP/PVA membranes resulted in a higher pore volume, producing a higher permeate ux when compared with the PP mat. Furthermore, this three-layered membrane achieved a salt rejection level of 99.7%. In conclusion, the OTMS-PP/PVA membrane achieved an improved salt rejection and a stable permeate ux, conrming the feasibility of a three-layered membrane for long-term MD operations.

Conflicts of interest
There are no conicts to declare.