membranes

Article Performance Comparison between Polyvinylidene Fluoride and Polytetrafluoroethylene Hollow Fiber Membranes for Direct Contact Membrane Distillation

Frank Y. C. Huang * and Allie Arning

Department of Civil and Environmental Engineering, New Mexico Tech, Socorro, NM 87801, USA; [email protected] * Correspondence: [email protected]; Tel.: +1-575-835-5592

 Received: 17 March 2019; Accepted: 8 April 2019; Published: 11 April 2019 

Abstract: Increasing water demand coupled with projected climate change puts the Southwestern United States at the highest risk of water sustainability by 2050. Membrane distillation offers a unique opportunity to utilize the substantial, but largely untapped geothermal brackish groundwater for desalination to lessen the stress. Two types of hydrophobic, microporous hollow fiber membranes (HFMs), including polytetrafluoroethylene (PTFE) and polyvinylidene fluoride (PVDF), were evaluated for their effectiveness in direct contact membrane distillation (DCMD). Water flux and salt rejection were measured as a function of module packing density and length in lab-scale systems. The PVDF HFMs generally exhibited higher water flux than the PTFE HFMs possibly due to thinner membrane wall and higher porosity. As the packing density or module length increased, water flux declined. The water production rate per module, however, increased due to the larger membrane surface area. A pilot-scale DCMD system was deployed to the 2nd largest geothermally-heated greenhouse in the United States for field testing over a duration of about 22 days. The results demonstrated the robustness of the DCMD system in the face of environmental fluctuation at the facility.

Keywords: DCMD; membrane distillation; desalination; hollow fiber membrane; PVDF; PTFE; water flux; field performance; pilot plant; geothermal; greenhouse

1. Introduction Traditionally, the bulk of the United States (U.S.) water demand has been satisfied by a combination of surface water and fresh groundwater. However, as climate and land use change, resources deplete, and population grows, these sources are not expected to suffice; thus, unconventional water resources are needed to prevent serious socio-economic consequences. Brackish groundwater represents a substantial, but largely untapped resource [1]. A conservative low estimate for the volume of extractible brackish groundwater is about 3.7 trillion·m3 (3 billion acre-feet), more than 35 times the amount of fresh groundwater used in the U.S. during 2010 [2]. The utilization of this unconventional water resource however is very limited. Data from United States Geological Survey (USGS) indicates that an estimated 12.3 million·m3/day (10,000 acre-feet/day) of brackish groundwater was used in the U.S. in 2010, which constituted only 4% of total groundwater use [2]. In the Western U.S., substantial amounts of freshwater are used for agriculture activities. For example, in California, Arizona, and other western states where pressures on water availability are especially intense, up to 80% of total water use is consumed by irrigated agriculture [3]. The strain on the water supply is further exacerbated by the projected climate change, which may result in a 10–20% reduction of the groundwater recharge across southwestern aquifers by 2100 [2]. Therefore, it is especially urgent

Membranes 2019, 9, 52; doi:10.3390/membranes9040052 www.mdpi.com/journal/membranes Membranes 2019, 9, 52 2 of 16 in this region to explore technologies that can utilize this vast quantity of unconventional resource as a supplement or replacement for freshwater supply. Reverse osmosis (RO) is a well-established technology for brackish water desalination. However, it is relatively energy-intensive, at times requires higher levels of pre-treatment to reduce membrane fouling, and has limited options for in-land concentrate management. For the western states of the U.S. where untapped brackish groundwater and geothermal resources are abundant, membrane distillation (MD) offers a unique opportunity to desalinate geothermal brackish groundwater with innate heat to supplement freshwater demand. In contrast to RO, MD is not a pressure-driven process and relies mainly on the vapor pressure gradient across hydrophobic membranes to drive the production of distilled water. Membrane distillation is especially attractive to horticulture and aquaculture operations that utilize geothermal brackish water as heating sources and where the demand of quality water for irrigation or cultivation is high. DCMD is the simplest and the most commonly used configuration of MD for water desalination [4,5]. In DCMD, both the hot brackish feed and cold distilled permeate are in direct contact with a micro-porous membrane. Liquid water is prevented from wetting the membrane pores by controlling the pore size and hydrophobicity of the membrane. Water vapor generated from the hot feed is transported across the membrane via diffusion induced by the trans-membrane vapor pressure gradient, and condenses on the cold permeate side. High hydrophobicity is one of the key parameters in selection to prevent pore wetting and subsequent performance deterioration in desalination. commonly used as membrane materials in DCMD include polyvinylidene fluoride (PVDF), polypropylene (PP), and polytetrafluoroethylene (PTFE). PTFE and PP exhibit higher hydrophobicity than PVDF; however, most researchers use PVDF to fabricate DCMD membranes due to its solubility in common solvents [5]. PTFE and PP, on the other hand, will have to be thermally extruded to make MD membranes. Many lab-scale DCMD studies have been conducted to demonstrate the ability of PVDF-based flat-sheet or hollow-fiber membranes for water desalination; however, limited data is available for their long-term performance in the field [6–13]. Gryta and Barancewicz evaluated the performance of PVDF capillary membranes with sponge-like or mixed sponge-/finger-like structures for lab-scale DCMD [14]. Significant decline (~60%) of water flux was observed for these systems after 1000 h of operation. The decline in water flux was attributed to progressive wetting of the pores. Pore wetting can significantly compromise the water flux and salt rejection of a DCMD system, leading to system failure. Inclusion of nanoparticles, such as PTFE, TiO2, SiO2, and carbon nanotube, and applications of extrinsic coatings with low surface energy have been investigated by researchers to increase the surface hydrophobicity of PVDF membranes, thereby reducing pore wetting [15–26]. These modifications, however, often suffer from complex manufacturing procedures, loss of the coatings in long-term operation, and potential health impact to human and the environment [27,28]. Compared to PVDF, PTFE possesses higher material hydrophobicity and lower chemical reactivity for applications in DCMD. Its performance reported in the literature, however, is almost exclusively for flat-sheet PTFE membranes; seldom for hollow fiber membranes (HFMs) [29–38]. Relative to flat sheet, the hollow-fiber configuration provides higher specific surface areas for membrane modules (e.g., HFM modules: up to 9000 m2/m3 versus spiral-wound modules: 1200 m2/m3), thus resulting in a higher productivity [39]. The hollow-fiber configuration is also self-supporting mechanically and therefore provides better flexibility during module fabrication and system operation [40]. This paper provides the much-needed performance data of PTFE HFMs for performance comparison. Preliminary field data collected for PVDF-based membrane modules at Masson Farms, the 2nd largest geothermally heated commercial greenhouse in the U.S., is also presented. The implication for system design and operation in the field is discussed. Membranes 2019, 9, 52 3 of 16

2. Materials and Methods

2.1. Membrane Fabrication and Characterization The PVDF hollow fiber membranes were fabricated using a dry jet-wet spinning process described in details elsewhere [13]. After spinning, the nascent fibers were submerged in de-ionized water for 24 h and subsequently freeze-dried to complete post-treatment before usage. The PTFE hollow fiber membranes fabricated using a melt-extrusion process were acquired from Markel Corporation (Plymouth Meeting, Montgomery County, PA, USA). Morphologies of the hollow fiber membranes were characterized using scanning electron microscopy, SEM (Hitachi S3200N, Chiyoda-ku, Tokyo, Japan). Pore sizes and distributions were determined using capillary flow porometry (Quantachrome 3G zh, Boynton Beach, FL, USA). Membrane porosities were estimated using a gravimetric method. Mechanical properties of the fibers were measured at 20 ◦C using a tensile tester (Mark-10 ESM303, Copiague, NY, USA) equipped with a 100 N digital force gauge (Mark-10 Model M5-20, Copiague, NY, USA). Detailed methodologies for characterization were described in prior publications [13,41].

2.2. Module Fabrication Individual fibers were hand sorted for defects before being fitted inside polycarbonate tubes with an inner diameter of 0.953 cm to achieve fiber packing densities up to 50%. Polycarbonate was chosen for its high thermal resistance and good optical clarity. The packing density of the hollow fiber membrane modules was calculated using Equation (1).

A f Packing Density (%) = × 100% (1) Am where A f is the total cross-sectional area of the fibers in a module and Am is the maximum packable cross-sectional area of the module determined by the mathematical software program Packomania [42]. Epoxy resin (Devcon Corporation) was then used to pot each module. The epoxy was cured at room temperature for 24 h followed by an additional 2–3 h at 50 ◦C. Prior to use, a section of the cured epoxy was removed from each end of the module to expose the lumen side of the hollow fiber membranes. Two types of membrane modules were prepared for testing with an effective fiber length of 9 cm and 24 cm, respectively. Each module underwent a QA/QC procedure in the lab after fabrication to check for module integrity before usage.

2.3. Experimental Setup for Lab-Scale Testing For the lab-scale membrane performance testing, all modules were evaluated in a co-current DCMD configuration with the experimental setup shown in Figure1. Membrane water flux and the associated salt rejection were quantified at various average water-vapor pressure gradients (VPG). Water flux, salt rejection, and the average water-vapor pressure gradient were defined in Equations (2)–(4).  liter  ∆V Water Flux , LMH = (2) m2 − hr A·∆t ! Cp Salt Rejection (%) = 1 − × 100% (3) Cf

 − −  Pf , i Pp, i + Pf ,o Pp,o  MPa  δ δ Average Vapor Pressure Gradient = (4) cm 2 where ∆V is the volume of the permeate, A is the membrane surface area per module, and ∆t is the measurement time interval. Cp and Cf are the TDS concentrations in the permeate and feed, respectively. Pf ,i, Pf ,o, Pp,i, Pp,o are the water-vapor pressures at the feed inlet and outlet, and those at Membranes 2019, 9, x FOR PEER REVIEW 4 of 16

129 of temperature gradient, across a membrane is a better process parameter for DCMD since it is scaled 130 linearly with water flux. Temperature gradient, on the other hand, is scaled non-linearly with water 131 flux due to the exponential increase of water vapor pressure with temperature and consequently, the 132 same temperature gradient may represent different driving forces for water diffusion [43]. 133 The main objective of the lab-scale study was to quantify water flux and the associated rate of 134 water production per module for the PTFE and PVDF HFMs. The impact of packing density and 135 temperature decline along the module on membrane performance was also investigated. Three 136 packing densities, including 10%, 25%, and 50%, were employed for PTFE and PVDF modules. For 137 all the testing, the feed-side and permeate-side fluid velocities were maintained at 0.06 m/s and 0.2 138 m/s respectively, regardless of the fiber packing density. The intent was to maintain the same level 139 of temperature polarization at the membrane surfaces for data comparison. The temperature 140 polarization (TP) is defined as the temperature difference between the bulk stream and the membrane 141 surface, and is a strong function of the fluid velocity [44]. Severe TP can limit the availability of 142 thermal energy in the hot feed for membrane distillation, leading to reduced water flux. For the effect 143 of temperature decline along the module length, the water-flux values generated for PVDF modules 144 with effective fiber lengths of 9 cm and 24 cm were compared. A minimum of 3 membrane modules 145 per packing density were evaluated to account for variance in operating conditions and fabrication 146 consistency. Sodium chloride solution with a TDS concentration of 5000 mg/L was used as the feed 147 for the modules. To maintain a constant feed concentration as permeate was generated, de-ionized 148 water was dripped from a supplementing reservoir into the feed reservoir to make up the loss. Hot 149 brackish water was run on the feed (shell) side of the membrane while cool, purified, water flowed 150 through the lumen. The cold stream was pumped through a stainless-steel heat exchanger with a 151 peristaltic pump (model: 77800-60, Cole-Parmer, Vernon Hills, IL, USA) before entering the permeate 152 (lumen) side of a module. The hot stream was pumped through the shell-side of a module with a 153 rotary piston pump (model Q, FMI, Syosset, NY, USA) and the feed temperature was controlled using 154 Membranesa hot water2019 bath, 9, 52 (model: 2335, Fisher Scientific, Hampton, NH, USA). Flow rates through a module4 of 16 155 were monitored with rotameters (Cole-Parmer, model T-03219-31). Condensate generated in the 156 lumen was collected in a permeate reservoir, then recirculated into the module. The permeate the permeate inlet and outlet. δ is the membrane thickness. Water-vapor pressure gradient, instead of 157 reservoir was placed on a balance and cumulative mass was measured over time to estimate water temperature gradient, across a membrane is a better process parameter for DCMD since it is scaled 158 flux. Temperature and mass were measured and recorded at 5-minute intervals for the duration of linearly with water flux. Temperature gradient, on the other hand, is scaled non-linearly with water 159 the test. Water samples from the feed and permeate were taken periodically to measure the flux due to the exponential increase of water vapor pressure with temperature and consequently, 160 conductivity of both streams with conductivity cells (Cole-Parmer, K = 10 and 0.1, 10 kΩ ATC). For the same temperature gradient may represent different driving forces for water diffusion [43]. 161 each VPG, the test was run for at least 45 hours after an equilibration period.

162 Figure 1. The lab-scale experimental setup for direct contact membrane distillation (DCMD). 163 Figure 1. The lab-scale experimental setup for direct contact membrane distillation (DCMD). The main objective of the lab-scale study was to quantify water flux and the associated rate of water production per module for the PTFE and PVDF HFMs. The impact of packing density and temperature decline along the module on membrane performance was also investigated. Three packing densities, including 10%, 25%, and 50%, were employed for PTFE and PVDF modules. For all the testing, the feed-side and permeate-side fluid velocities were maintained at 0.06 m/s and 0.2 m/s respectively, regardless of the fiber packing density. The intent was to maintain the same level of temperature polarization at the membrane surfaces for data comparison. The temperature polarization (TP) is defined as the temperature difference between the bulk stream and the membrane surface, and is a strong function of the fluid velocity [44]. Severe TP can limit the availability of thermal energy in the hot feed for membrane distillation, leading to reduced water flux. For the effect of temperature decline along the module length, the water-flux values generated for PVDF modules with effective fiber lengths of 9 cm and 24 cm were compared. A minimum of 3 membrane modules per packing density were evaluated to account for variance in operating conditions and fabrication consistency. Sodium chloride solution with a TDS concentration of 5000 mg/L was used as the feed for the modules. To maintain a constant feed concentration as permeate was generated, de-ionized water was dripped from a supplementing reservoir into the feed reservoir to make up the loss. Hot brackish water was run on the feed (shell) side of the membrane while cool, purified, water flowed through the lumen. The cold stream was pumped through a stainless-steel heat exchanger with a peristaltic pump (model: 77800-60, Cole-Parmer, Vernon Hills, IL, USA) before entering the permeate (lumen) side of a module. The hot stream was pumped through the shell-side of a module with a rotary piston pump (model Q, FMI, Syosset, NY, USA) and the feed temperature was controlled using a hot water bath (model: 2335, Fisher Scientific, Hampton, NH, USA). Flow rates through a module were monitored with rotameters (Cole-Parmer, model T-03219-31). Condensate generated in the lumen was collected in a permeate reservoir, then recirculated into the module. The permeate reservoir was placed on a balance and cumulative mass was measured over time to estimate water flux. Temperature and mass were Membranes 2019, 9, 52 5 of 16 measured and recorded at 5-min intervals for the duration of the test. Water samples from the feed and permeate were taken periodically to measure the conductivity of both streams with conductivity cells (Cole-Parmer, K = 10 and 0.1, 10 kΩ ATC). For each VPG, the test was run for at least 45 h after an equilibration period.

2.4. Field Testing Field testing of the DCMD system for desalination was conducted at Masson Farms with geothermal brackish groundwater. Masson Farms, located in Radium Springs New Mexico, is the second largest geothermally heated industrial greenhouse in the United States [13]. A well descending about 240 m is used to extract geothermal water from a highly fractured rhyolite dike for space heating of the greenhouse at a flow rate of about 6540 m3/day. Temperature and TDS of the geothermal water are 92 ◦C and 3800 mg/L, respectively. The geothermal water is circulated through plate and frame heat exchangers that transfer heat from the geothermal stream to a recirculating freshwater stream for space heating. The geothermal water has a temperature of about 70 ◦C after the heat exchangers and is then pumped back into the rhyolite dike reservoir downstream. Many of the plant species grown at the greenhouse are extremely sensitive to the salt concentration in water. Therefore, water from permitted freshwater wells and the Rio Grande River must be treated to reduce TDS from 1800 to 300–400 mg/L before it may be used for irrigation. Currently, the greenhouse relies on reverse osmosis (RO) to produce 872 m3/day of permeate. Chemical treatment is applied to the feed water for inorganic fouling control, which increases the energy footprint and operating cost of the greenhouse. Desalination of the brackish groundwater using DCMD for irrigation was evaluated at the site for cost reduction. Innate heat in the post-heat-exchanger geothermal fluid was used as the energy source for the DCMD process. Composition of the geothermal brackish groundwater is listed in Table1[13].

Table 1. Composition of the geothermal brackish groundwater *.

Parameter Masson Farms Ca+ 104 Mg2+ 11.4 K+ 191 Na+ 1221 Li+ 1.1 Fe2+ 0.2 Sr2+ 2.3 F− 5.5 Cl− 2022 SO2− 27.6 4 − HCO3 11.6 B 0.9 Si (as SiO2) 58.2 pH 7.5 TDS 3800 Temperature (◦C) 92 * Units are in mg/L unless otherwise noted.

Field DCMD System A mobile DCMD system was constructed and housed in a 6-m trailer for field testing. The geothermal raw water under the line pressure at the facility was filtered through a coarse (~100 µm) screen filter to remove potential particulates. The pressurized water then filled the source-water tank for usage. The filling was controlled by a pair of float switches coupled with a motorized ball valve at the inlet of the tank. A conductivity probe was mounted inside the tank to monitor TDS of the raw water. For DCMD experiments, the raw water was pumped from the tank using a centrifugal Membranes 2019, 9, 52 6 of 16

pump controlled by a variable frequency drive (VFD) into fabricated hollow-fiber membrane modules. There were four banks located at the shell-side and lumen-side inlets as well as outlets of the module array. Each sensor bank included a pressure transducer and a RTD (Pt100) temperature sensor, with the exception of the outlet banks, which also included conductivity probes for TDS measurements. TheMembranes post-module 2019, 9, concentrate x FOR PEER REVIEW entered a hot-water holding tank from where the fluid was injected6 of 16 via a centrifugal pump into the facility’s process line downstream of the source water intake. A check 196 measurements. The post-module concentrate entered a hot-water holding tank from where the fluid 197valve was was injected installed via betweena centrifugal the concentratepump into the discharge facility’s pumpprocess and line the downstream facility’s processof the source line to water prevent 198backflow. intake. TheA check permeate valve was generated installed flowed between directly the conc intoentrate a clean-water discharge pump holding and the tank facility’s where process the water 199accumulated line to prevent over backflow. time was The monitored permeate by generated weight (balance)flowed directly or water into level a clean-water (ultrasonic holding level tank sensor). 200A centrifugalwhere the water pump accumulated took the warmed over time water was monitore from thed by clean-water weight (balance) holding or water tank level and (ultrasonic circulated it 201through levelan sensor). air-cooled A centrifugal radiator topump reduce took the the water warm temperatureed water from before the clean-water returning it holding back to tank the modules and 202for membranecirculated it distillation. through an air-cooled Periodically, radiator the accumulatedto reduce the water water temperature in the clean-water before returning tank was it pumpedback 203to theto the hot-water modules tank for membrane for disposal. distillation. For geothermally-heated Periodically, the accumulated horticulture water in operations, the clean-water pumping tank of 204geothermal was pumped fluid forto the space hot-water heating tank will for transition disposal. from For geothermally-heated a continuous mode tohorticulture an intermittent operations, mode as 205 the weatherpumping getsof geothermal warmer. A fluid hot-water for space recirculation heating will loop transition through from the membrane a continuous module mode arrayto an was 206 intermittent mode as the weather gets warmer. A hot-water recirculation loop through the membrane incorporated in the design to minimize frequent system shutdown during this period, which often 207 module array was incorporated in the design to minimize frequent system shutdown during this led to equipment failure. The feed and permeate flow rates were monitored via either infrared-based 208 period, which often led to equipment failure. The feed and permeate flow rates were monitored via 209turbine either infrared-based or ultrasonic turbine flow sensors sensors. or A ultrason pipingic and flow instrumentation sensors. A piping diagram and forinstrumentation the field DCMD 210system diagram is provided for the field in Figure DCMD2. system is provided in Figure 2.

211 212 FigureFigure 2. Piping2. Piping and and instrumentation instrumentation diagramdiagram of of the the field field DCMD DCMD system system..

2133. Results3. Results and and Discussion Discussion

2143.1. Membrane3.1. Membrane Characteristics Characteristics 215 TheThe fabricated fabricated PVDF PVDF membrane membrane possesses possesses an asymmetrican asymmetric configuration, configuration, consisting consisting of an of external an 216sponge external layer sponge and an layer internal and an macro-void internal macro-void layer as shownlayer as in shown Figure in3 .Figure During 3. During DCMD, DCMD, hot brackish hot 217feed brackish was in contact feed was with in contact the exterior with the (shell-side) exterior (she ofll-side) the HFMs of the while HFMs cold while distilled cold distilled water was water in was contact 218with in the contact interior with (lumen-side) the interior (lumen-side) of the HFMs. of the The HFMs. intention The intention was to was use to the use tight the tight pore pore structure structure of the 219 spongeof the layer sponge to prevent layer to the prevent brackish the water brackish from wa intrudingter from intointruding the membranes into the membranes and low tortuosityand low of 220 tortuosity of the macro-void layer to facilitate water-vapor transport through the membranes. The 221 PTFE membrane has a rather symmetric pore structure as shown in Figure 3. The uni-axial tension 222 applied during the fabrication process created elliptically shaped pores on the external surfaces of 223 the membranes.

Membranes 2019, 9, 52 7 of 16

the macro-void layer to facilitate water-vapor transport through the membranes. The PTFE membrane has a rather symmetric pore structure as shown in Figure3. The uni-axial tension applied during the fabricationMembranes 2019 process, 9, x FOR created PEER REVIEW elliptically shaped pores on the external surfaces of the membranes.7 of 16

224

225 FigureFigure 3.3. ScanningScanning electron electron micros microscopycopy (SEM) (SEM) images images of the of hollow the hollow fiber membranes fiber membranes (a) 226 (a) polytetrafluoroethylene (PTFE) (PTFE) fiber, fiber, (b) PTFE (b) PTFE cross-section, cross-section, (c) PTFE (c) PTFEexternal external surface,surface, (d) 227 (dpolyvinylidene) polyvinylidene fluoride fluoride (PVDF) (PVDF) fiber, fiber, (e) (PVDFe) PVDF cross-section, cross-section, and and(f) PVDF (f) PVDF external external surface. surface.

228 TableTable2 shows2 shows the the characteristics characteristics of of thethe hollowhollow fiberfiber membranes. The The wall wall thickness thickness of ofthe the PTFE PTFE 229 membranemembrane is is about about 1.5 1.5 timestimes thickerthicker than that of of the the PVDF PVDF membrane membrane and and the the associated associated porosity porosity is is 230 aboutabout 5/8 5/8 ofof thatthat forfor thethe PVDFPVDF membrane.membrane. Therefor Therefore,e, it it was was anticipated anticipated that that the the PVDF PVDF membrane membrane 231 wouldwould produce produce a a higher higher water water flux flux than than the the PTFE PTFE membranemembrane forfor aa similarsimilar pair of inlet temperature in 232 thein feedthe feed and and permeate permeate to the to lab-scalethe lab-scale DCMD DCMD system. system. The The PTFE PTFE membrane membrane has has a nominala nominal pore pore size 233 aboutsize 1.5about times 1.5 oftimes that forof that the PVDFfor the membrane PVDF membrane and also and possesses also possesses a much wider a much pore-size wider pore-size distribution. 234 Thisdistribution. is likely the This reason is likely why the the reason values why of the liquid values entry of liquid pressure entry for pressure water (LEP for water) for (LEP the twow) for types the of 235 ° w membranestwo types areof similarmembranes at 22 are◦C despitesimilar theat 22 higherC despite hydrophobicity the higher for hydrophobicity PTFE. It is important for PTFE. to note It is that 236 important to note that LEPw was observed to decrease with increase of water temperature and can LEPw was observed to decrease with increase of water temperature and can constrain the operation of 237 constrain the operation of DCMD systems at elevated temperature [45]. For mechanical properties, DCMD systems at elevated temperature [45]. For mechanical properties, the PTFE fiber has a much 238 the PTFE fiber has a much higher Young’s modulus than the PVDF fiber, suggesting that the former 239 is more brittle and has a higher degree of crystallinity. Its failure stress is also substantially higher 240 than that of the PVDF fiber and can facilitate the module-making process.

Membranes 2019, 9, 52 8 of 16 higher Young’s modulus than the PVDF fiber, suggesting that the former is more brittle and has a higher degree of crystallinity. Its failure stress is also substantially higher than that of the PVDF fiber and can facilitate the module-making process.

Table 2. Characteristics of the PTFE and PVDF Hollow Fiber Membranes.

Membrane Characteristic PVDF PTFE Outer diameter (µm) a 841 ± 5 1799 ± 50 Wall thickness (µm) a 122 ± 22 178 ± 12 Macrovoid to sponge ratio b 1.08 N/A Pore size 0.319/0.333/0.422 c 0.385/0.495/0.831 d Porosity 0.79 ± 0.05 0.50 ± 0.04 Failure stress 1.32 >21.5 Young’s modulus 15.66 348 at 22 ◦C 1.32 1.37 Liquid entry pressure, LEP (bar) w at 81 ◦C 0.53 - a nominal size based on post-processing of SEM images. b ratio of the macrovoid-layer thickness to the sponge-layer thickness. c minimum pore size/mean pore size/maximum pore size. d based on the short axis of the elliptical pore.

3.2. Lab-Scale Membrane Performance Water flux of the membrane modules at packing densities of 10, 25, and 50% was measured as a function of the average water-vapor pressure gradient imposed on the modules. Since the vapor pressure gradient is the driving force for water diffusion across the membranes, it was observed to scale linearly with the water flux for a specific packing density. Figure4 shows the correlations for PTFE membrane modules with an effective fiber length of 9 cm. The membrane surface area for a 10% packed module was about 0.0011 m2 and increased proportionally with packing density. As the packing density increased, the observed water flux declined. This reduction of water flux was probably attributed to the decline of the total heat flow into the module while maintaining the average shell-side fluid velocity. Randomly-packed hollow fibers with an increasing packing density could also result in progressively uneven distribution of fluid flow and thus heat flow among the fibers. A similar trend with packing density was also observed for the PVDF membrane modules. For a 10% packing density, the PVDF-based modules had about 0.00264 m2 of membrane surface area per module. Figure5 presents a performance comparison between the PTFE and PVDF membrane modules at an average vapor pressure gradient of 1 MPa/cm (e.g., hot inlet: 55–60 ◦C and cold inlet: 22–25 ◦C). For low packing density (i.e., 10%) when the thermal energy for distillation is abundant, the PVDF modules exhibited a significantly higher water flux (~1.6 times) than the PTFE module due to the larger membrane surface area per module. As the packing density increased from 10 to 50%, similar water flux, however, was observed for both types of modules while the ratio of the membrane surface area between PVDF and PTFE remained about the same at 2.5. This convergence of observed water flux at high packing density was likely in part due to the lack of thermal energy for distillation as the total membrane surface area in a module grew. Nevertheless, the module water production rate for the PVDF membranes was still 2.5 times of that for the PTFE membranes at 50% packing density. This result signifies the importance of reducing fiber diameter in maximizing the water production rate per module for DCMD. The water flux values observed in this study for PVDF and PTFE HFMs were also compared to those reported in the literature for PTFE membranes as shown in Table3[ 38,46,47]. Water flux of the fabricated PVDF HFM in this study is comparable to those reported for commercial PTFE flat-sheet membranes under similar inlet temperatures. The water flux values observed for the PTFE HFM tested in this study are higher than that reported in the literature for a PTFE HFM. However, depending on the module packing density, these values were about 15% to 68% lower than those reported for commercial PTFE flat-sheet membranes. It is important to note that the active-layer thickness for the PTFE HFM in this study is 2.7 to 6 times thicker than those for Membranes 2019, 9, 52 9 of 16

the commercial PTFE flat-sheet membranes and therfore generated substantially higher resistance for massMembranes transfer 2019 of, water9, x FOR vapor.PEER REVIEW 9 of 16

14

10% * 25% 50% 12 * module packing density -hr) -hr) 2 2 10 (liter/m (liter/m w w 8

6 Water Flux, J Water Flux, J 4 Effective PTFE fiber length: 9 cm Co-current flow configuration 2 00.511.52 Average Vapor Pressure Gradient, VPG (MPa/cm) 280

281 FigureFigure 4. Water 4. Water flux flux as as a functiona function of of the the average average water-vaporwater-vapor pressure pressure gradient gradient for for10, 10,25, 25,and and50% 50% 282 modulemodule packing packing densities densities in in PTFE PTFE membrane membrane modules.modules. Other Other experimental experimental conditions: conditions: Lab Labtesting, testing, 283 feed:feed: NaCl NaCl solution solution with with TDS TDS of of 5000 5000 mg/L, mg/L, effective fiber fiber length: length: 9 cm, 9 cm, co-current co-current flow, flow, stable stable water water 284 fluxflux for 45forh, 45salt hours, rejection: salt rejection: over 99%. over 99%.

14 5.0 Effective fiber length: 9 cm VPG: 1 MPa/cm 12 4.0 -hr) -hr) 2 2 10

3.0 8 (liter/m (liter/m w w 6 2.0

4 Water Flux, J Water Flux, J 1.0 Water Production Rate (liter/day) Rate Production Water 2 (liter/day) Rate Production Water

0 0.0 0 102030405060 Module Packing Density (%) Flux (PVDF) Flux (PTFE) Water Production (PVDF) Water Production (PTFE) 285 286 FigureFigure 5. 5.Module Module performanceperformance in in water water flux flux and and production production rate for rate PTFE for and PTFE PVDF and hollow PVDF fiber hollow 287 fibermembranes membranes as a as function a function of packing of packing density density at a vapor at a pressure vapor pressure gradient gradientof 1 MPa/cm. of 1 Other MPa/cm. 288 experimental conditions: Lab testing, feed: NaCl solution with TDS of 5000 mg/L, effective fiber 288 Otherexperimental experimental conditions:conditions: Lab Labtesting, testing, feed: feed: NaCl NaCl solution solution with with TDS TDS of 5000 of 5000 mg/L, mg/L, effective effective fiber fiber 289 length: 9 cm, co-current flow, stable water flux for 45 hours, salt rejection: over 99%. length: 9 cm, co-current flow, stable water flux for 45 h, salt rejection: over 99%.

Membranes 2019, 9, 52 10 of 16

Table 3. Comparisons of Properties and Performance for Selected PVDF and PTFE MD Membranes.

Membrane Membrane This Study This Study Zhang et al. a Millipore b GE Osmonics c Characteristic Solutions c Hollow Type Hollow fiber Hollow fiber Flat sheet Flat sheet Flat sheet fiber Material PTFE PVDF PTFE PTFE PTFE PTFE Membrane Sponge/ Active layer/ Active layer/ Symmetric Symmetric N/A configuration macrovoid fabric scrim Active 178 122 365 N/A 30 67 Thickness Support N/A N/A N/A N/A 185 97 Nominal pore size 0.5 0.3 0.3 0.25 1 0.45 of active layer (µm) Porosity (active 0.5 0.79 0.85 0.7 0.92 0.88 layer) Feed TDS (mg/L) 5000 5000 10,000 29,250 10,000 10,000 Flow configuration Co-current Co-current Counter-current Counter-current Counter-current Counter-current Module packing 10 50 10 50 N/A N/A N/A N/A density (%) Feed inlet (◦C) 65 64 62 64 60 65 60 60 Permeate inlet (◦C) 22 31 28 36 20 15 20 20 Water flux (LMH) d 9.4 5.3 17.3 5.7 4 12.6 11 16.5 a based on reference [38]. b FGLP 1425, based on reference [45]. c based on reference [46]. d LMH = liters per m2 of membrane area, per hour.

Water flux for PVDF membrane modules with an effective fiber length of 24 cm was also recorded for different packing densities as shown in Figure6. The results for the long and short (i.e., 24 cm vs. 9 cm) PVDF membrane modules were compared to examine the impact of temperature decline along the module length. As expected, increasing module length led to lower water flux at a constant packing density. For the long modules, severe temperature decline along the module length as a result of insufficient heat flow diminished the rate of water distillation for membrane area closer to the module outlet, leading to a lower average water flux for the module. At 50% packing density, the long module provided about 53% of the water flux for the short module and yet delivered 42% more in the module water production rate due to 2.7 times of membrane surface area. For the long modules, the water flux and associated production rate can be further boosted by increasing the fluid flow rate on the shell side of the modules at the expense of higher energy costs for pumping. Salt rejection was recorded for all the experiments and was always greater than 99%. Membranes 2019, 9, 52 11 of 16 Membranes 2019, 9, x FOR PEER REVIEW 11 of 16

18

16 10% * 25% 50% * module packing density 14 -hr) 2 12

10 (liter/m w 8

6

Water Flux, J 4

2 Effective PVDF fiber length: 24 cm Co-current flow configuration 0 00.511.522.53 Average Vapor Pressure Gradient, VPG (MPa/cm) 307

308 FigureFigure 6. Water 6. Water flux flux as as a functiona function of of the the average average water-vaporwater-vapor pressure pressure gradient gradient for for10, 10,25, 25,and and50% 50% 309 modulemodule packing packing densities densities in in PVDF PVDF membrane membrane modules.modules. Other Other experimental experimental conditions: conditions: Lab Labtesting, testing, 310 feed:feed: NaCl NaCl solution solution with with TDS TDS of of 5000 5000 mg/L, mg/L, effectiv effectivee fiber fiber length: length: 24 24 cm, cm, co-current co-current flow, flow, stable stable water water 311 fluxflux for 45forh, 45 salt hours, rejection: salt rejection: over 99%. over 99%.

3123.3. 3.3. Field Field Performance Performance 313 PVDFPVDF membrane membrane modules modules fabricated fabricated with an an effective effective fiber fiber leng lengthth of of24 24cm cm and and a 10% a 10% packing packing 314density density were were tested tested at at MassonMasson Farms Farms for for their their effectiveness effectiveness of desalination of desalination using the using field DCMD the field system. DCMD 315system. The geothermal The geothermal brackish brackish groundwater groundwater on site was onconveyed site was to the conveyed field system to the post field heat systemexchangers post for heat 316exchangers water desalination. for water Temperature desalination. of Temperaturethe groundwater of at the the groundwater inlet of the DCMD at the system inlet offluctuated the DCMD between system 317 ° ° fluctuated35 to 75 between C with most 35 to of 75 the◦C values with mostconcentrating of the values within concentrating 50 to 60 C. Temperature within 50 toof 60the◦ C.recirculating Temperature 318 permeate ranged from 10 to 33 °C during the field testing. Performance of the modules was monitored for of the recirculating permeate ranged from 10 to 33 ◦C during the field testing. Performance of 319 more than 22 days and was shown in Figure 7. Water flux was observed to track closely with the water- the modules was monitored for more than 22 days and was shown in Figure7. Water flux was 320 vapor pressure gradient imposed on the modules. The salt rejection was consistently about 99% 321observed throughout to track the closelyperiod. The with significant the water-vapor variability pressure of the observed gradient water-vapor imposed onpressure the modules. gradient Thewas salt 322rejection resulted was mainly consistently from the about temperature 99% throughout fluctuation the of the period. feed into The the significant DCMD system. variability The offlexible the observed hose 323water-vapor connecting pressure the facility gradient process-water was resulted pipe to mainly the inlet from of the temperatureDCMD system fluctuation was not insulated of the feed and into 324the DCMDconsequently system. subjected The flexibleto the influence hose connecting of the atmospheric the facility temperature. process-water The projected pipe towater the flux inlet was of the 325DCMD calculated system based was on not the insulatedlab-scale correlation and consequently presented in subjected Figure 6 for to modules the influence of a similar ofthe sizeatmospheric and with 326temperature. a 10% packingThe density. projected However,water the flux DCMD was calculated process em basedployed onduring the lab-scalethe field testing correlation was in a presented counter- in 327Figure current6 for flow modules configuration. of a similar This size configuration and with was a 10% shown packing to exhibit density. higher However, water fluxes the DCMD than the process co- 328employed current duringconfiguration the field used testing to establish was in the a counter-currentlab-scale correlation. flow Therefore, configuration. a multiplier This of configuration 1.21, obtained was 329shown from to prior exhibit studies, higher was water applied fluxes to the thanlab-scale the correl co-currentation to configuration account for the used increase to establishof water flux the under lab-scale 330 counter-current flow configuration [18]. A comparison of the water flux data in Figure 8 showed that the correlation. Therefore, a multiplier of 1.21, obtained from prior studies, was applied to the lab-scale 331 membrane performance in water production did not deteriorate over time. The observed water flux, correlation to account for the increase of water flux under counter-current flow configuration [18]. 332 however, was consistently lower than the projected water flux. In order to estimate the average percentage A comparison of the water flux data in Figure8 showed that the membrane performance in water 333 of reduction, the observed water-flux data was sorted in a descending order and separated into different 334production groups based did noton the deteriorate associated overaverage time. vapor The pressure observed gradient water (VPG). flux, The however, water flux was data consistently in each group lower 335than with the a projected VPG span water of 0.05 flux.MPa/cm, In order was then to estimate averaged. the The average results shown percentage in Figure of 8, reduction, indicated an the average observed 336water-flux reduction data of 10% was in sorted water flux in a when descending compared order to the and corrected separated lab-scale into correlation. different groupsThe reduction based was on the 337associated probably average caused vaporby the pressureparticulate gradient foulants (VPG).from the The corrosion water fluxof cast data-iron in pipes each in group the facility with aand VPG the span 338of 0.05magnitude MPa/cm, of reduction was then was averaged. consistent The with results the data shown observed in Figure during8, lab-scale indicated fouling an average testing [18]. reduction of 10% in water flux when compared to the corrected lab-scale correlation. The reduction was probably caused by the particulate foulants from the corrosion of cast-iron pipes in the facility and the magnitude of reduction was consistent with the data observed during lab-scale fouling testing [18]. Membranes 2019, 9, 52 12 of 16

MembranesDuring 2019 the, 9, fieldx FOR testing,PEER REVIEW deposition of reddish particulate matter on membrane surfaces12 of 16 was observed over time. It was evidenced that the 100-µm pre-filter at the inlet of the field DCMD system 339 was notDuring sufficient the tofield capture testing, all deposi the particulatetion of reddish matter. particulate In order matter to determine on membrane the nature surfaces and surfacewas 340 observed over time. It was evidenced that the 100-µm pre-filter at the inlet of the field DCMD system was loading of the particulate matter, the hollow fiber membranes were submerged in de-ionized water and 341 not sufficient to capture all the particulate matter. In order to determine the nature and surface loading of sonicated to dislodge the particulate matter at the end of the field testing. The resulted water was then 342 the particulate matter, the hollow fiber membranes were submerged in de-ionized water and sonicated to filtered with a 5-µm filter, which was subsequently dried at 70 ◦C for an hour before being weighed. 343 dislodge the particulate matter at the end of the field testing. The resulted water was then filtered with a Based on the results, the surface loading was estimated to be about 300 mg per m2 of membrane surface 344 5-µm filter, which was subsequently dried at 70 °C for an hour before being weighed. Based on the results, area. The dried particulate matter was analyzed using X-ray diffraction (XRD) and was determined to 345 the surface loading was estimated to be about 300 mg per m2 of membrane surface area. The dried 346 beparticulate mainly iron matter oxides, was probably analyzed ausing corrosion X-ray productdiffraction of (XRD) the cast-iron and was pipes determined used at to Masson be mainly Farms. iron 347 oxides,Based probably on the fielda corrosion experience, product implication of the cast-iron for system pipes used design at Masson and operation Farms. was provided for field 348 DCMDBased systems on withthe field PVDF-based experience, HFMs implication that utilize for system geothermal design brackish and operation groundwater was provided as the source. for 349 Fromfield the DCMD system systems design with perspective, PVDF-based HFM HFMs modules that utilize should geothermal preferrably brackish be oriented groundwater in an as upright the 350 orientationsource. From with the both system the feed design and perspective, permeate flowing HFM modules co-currently should upward. preferrably Gas evolution be oriented is common in an 351 forupright geothermal orientation groundwater with both and the feed can createand permea gas lockte flowing in improper co-currently module upward. setup Gas to evolution obstruct flowis 352 throughcommon the for modules. geothermal A recirculation groundwater loop and shouldcan create be ga incorporateds lock in improper in the systemmodule designsetup to to obstruct minimize 353 membraneflow through fouling the and modules. equipment A recirculation failure during loop the should warmer be incorporated seasons when in geothermal the system groundwater design to 354 is availableminimize intermittentlymembrane fouling for space and heating.equipment Chilling failure permeate during the to warmer increase seasons the vapor when pressure geothermal gradient 355 forgroundwater water production is available can beintermittently energy intensive. for space Heat heating. exchnagers Chilling using permeate air cooling to increase are thethe primaryvapor 356 candidatespressure basedgradient on for economic water production considerations. can be For energy the field intensive. operation, Heat the exchnagers hightest frequencyusing air cooling of system 357 failureare the occurred primary during candidates startup based and on shutdown economic whenconsiderations. the hydraulic For the pressure field operation, differential the hightest across the 358 membranesfrequency fluctuatedof system thefailure most. occurred Consequently, during st pressureartup and monitoring shutdown and when proper the selectionhydraulic ofpressure pumping 359 systemsdifferential are needed across to the minimize membranes the risk. fluctuated Generally, the groundwatermost. Consequently, with higher pressure temperature monitoring is chosen and as 360 proper selection of pumping systems are needed to minimize the risk. Generally, groundwater with the source for DCMD systems to increase the water flux. However, LEPw can be significantly reduced 361 higher temperature is chosen as the source for DCMD systems to increase the water flux. However, with the increase of water temperature. For example, the LEPw of the fabricated PVDF HFMs in this 362 LEPw can be significantly reduced with the increase of water temperature. For example, the LEPw of study was reduced by 60% from 1.32 to 0.53 bar as the water temperature increased from 22 to 81 ◦C. 363 the fabricated PVDF HFMs in this study was reduced by 60% from 1.32 to 0.53 bar as the water This reduction of LEPw increases the risk of brackish water intrusion and subsequent loss of water flux 364 temperature increased from 22 to 81 °C. This reduction of LEPw increases the risk of brackish water 365 duringintrusion the systemand subsequent startup and loss shutdown.of water flux during the system startup and shutdown.

12 3

10 2.5 -hr)

2 8 2

6 1.5

4 1 Water Flux (liter/m Flux Water

2 0.5 VPG (MPa/cm) orConductivity (mS/cm) 0 0 0 50 100 150 200 250 300 350 400 450 500 550 Elapsed Time (hour)

366 Observed Water Flux Projected Water Flux VPG Permeate Conductivity

367 FigureFigure 7. 7.Observed Observed and and projected projected waterwater fluxflux with the as associatedsociated vapor vapor pressure pressure gradient gradient (VPG) (VPG) over over 368 timetime for for the the PVDF PVDF membrane membrane modules modules tested tested in the in field.the field. Other Other experimental experimental conditions: conditions: Field Field testing, 369 feed:testing, geothermal feed: geothermal groundwater groundwater with TDS with of 3800TDS mg/L,of 3800 effectivemg/L, effective fiber length: fiber length: 24 cm, 24 10% cm, packing 10% 370 packing density, counter-current flow, water flux observed for 530 hours, salt rejection: 99%. density, counter-current flow, water flux observed for 530 h, salt rejection: 99%.

Membranes 2019, 9, 52 13 of 16 Membranes 2019, 9, x FOR PEER REVIEW 13 of 16

12

Observed Water Flux 10 Projected Water Flux -hr) 2 8 (liter/m

w 6

Jw, o bs e rv e d = 6.13 (VPG) + 2.01 4 R² = 0.9053 Water Flux, J 2

10% Module Packing Density 0 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 Average Vapor Pressure Gradient, VPG (MPa/cm) 371

372 Figure 8. ComparisonComparison between between the the observed observed and and predicte predictedd water water flux flux as as a function of the average 373 vapor pressurepressure gradient gradient for for the field-testedthe field-tested PVDF PVDF membrane membrane modules. modules. Other experimental Other experimental conditions: 374 conditions:Field testing, Field feed: testing, geothermal feed: geothermal groundwater ground withwater TDS with of 3800 TDS mg/L, of 3800 effective mg/L, effective fiber length: fiber length: 24 cm, 375 2410% cm, packing 10% packing density, density, counter-current counter-current flow, water flow, flux wa observedter flux observed for 530 h, for salt 530 rejection: hours, 99%.salt rejection: 376 99%. 4. Conclusions 377 4. ConclusionsPVDF hollow fiber membranes (HFMs) were fabricated and compared to commercially available 378 PTFEPVDF HFMs hollow in their fiber effectiveness membranes for (H desalinationFMs) were fabricated using DCMD. and compared The PVDF to commercially HFMs exhibited available much 379 PTFEhigher HFMs water in flux their than effectiveness the PTFE for HFMs desalinat at 10% packingusing DCMD. density, The resulting PVDF HFMs mainly exhibited from thinner much 380 highermembrane water wall flux and than higher the PTFE porosity. HFMs Both at types10% packing of HFMs, density, however, resulting showed mainly similar from water thinner flux 381 membraneat 50% packing wall densityand higher possibly porosity. due toBoth insufficient types of HFMs, thermal however, energy in showed the feed. similar Relative water to commercialflux at 50% 382 packingPTFE flat-sheet density membranes,possibly due the to insufficient PTFE HFMs thermal possessed energy water in the flux feed. that Relative was about to 15%commercial to 68% lower,PTFE 383 flat-sheetdepending membranes, on the packing the PTFE density. HFMs The possessed deficiency water can mostflux that likely was be about rectified 15% by to reducing 68% lower, the 384 dependingactive-layer on thickness the packing of the density. PTFE The HFMs, deficiency which can is about most likely 2.7 to be 6 timesrectified thicker by reducing than those the foractive- the 385 layercommercial thickness PTFE of the flat-sheet PTFE HFMs, membranes. which Increasing is about 2.7 module to 6 times packing thicker density than andthose length for the had commercial a negative 386 PTFEimpact flat-sheet on water membranes. flux. For example,Increasing as module the packing packing density density of and the length short PVDF had a modulesnegative impact increased on 387 waterfrom 10flux. to 50%,For example, the water as flux the decreasedpacking density by 60% of fromthe short 13.4 PVDF to 5.4 LMHmodules at an increased average from vapor 10 pressure to 50%, 388 thegradient waterof flux 1 MPa/cm. decreased Despite by 60% thefrom declining 13.4 to 5.4 of waterLMH flux,at an the average water vapor production pressure rate gradient per module of 1 389 MPa/cm.increased Despite gradually the with declining the increase of water of total flux, membrane the water surface production area from rate raised per packingmodule densityincreased or 390 graduallymodule length. with the The increase results affirmedof total membrane the needs surface to reduce area fiber from diameter, raised packing maximize density packing or module density, 391 length.and increase The results module affirmed lengthfor the water needs production. to reduce Itfiber is anticipated diameter, thatmaximize increasing packing the feeddensity, flow and rate 392 increasewould further module boost length the for water water flux, production. however, It atis an theticipated expense that of a increasing higher pumping the feed cost.flow Membranerate would 393 furtherperformance boost of the PVDF water modules flux, withhowever, an effective at the fiber expense length of of a 24 higher cm and pumping a 10% packing cost. densityMembrane was 394 performancealso evaluated of atPVDF the 2nd modules largest, with geothermally-heated, an effective fiber length commercial of 24 cm greenhouse and a 10% packing in the United density States was 395 alsowith evaluated on-site geothermal at the 2nd brackish largest, groundwater.geothermally-h Theeated, water commercial flux and salt greenhouse rejection observedin the United were States fairly 396 withstable on-site over moregeothermal than 22 brackish days in gr theoundwater. presence of The particulate water flux foulants and salt and rejection severe observed temperature were swing. fairly 397 stableThe results over more demonstrated than 22 days the robustness in the presence of the of DCMD particulate system foulants in the field.and severe temperature swing. 398 The results demonstrated the robustness of the DCMD system in the field. Author Contributions: Conceptualization, F.Y.C.H. and A.A.; methodology, A.A. and F.Y.C.H.; formal analysis, 399 A.A. and F.Y.C.H.; investigation, A.A.; writing—original draft preparation, F.Y.C.H. and A.A.; writing—review and editing, A.A. and F.Y.C.H.; funding acquisition, F.Y.C.H.

Membranes 2019, 9, 52 14 of 16

Funding: This work was supported in part by U.S. National Science Foundation award #IIA-1301346 to NM EPSCoR and its Osmotic Power Research Group. Additional funding was provided by the Desalination and Water Purification Research (DWPR) Program of Bureau of Reclamation, U.S. Department of Interior. Acknowledgments: The authors would like to express our gratitude to Masson Farms for providing information. Conflicts of Interest: The authors declare no conflict of interest.

References

1. Tidwell, V.C.; Moreland, B.D.; Zemlick, K.M.; Roberts, B.L.; Passell, H.D.; Jensen, D.; Forsgren, C.; Sehlke, G.; Cook, M.A.; King, C.W.; et al. Mapping water availability, projected use and cost in the western United States. Environ. Res. Lett. 2014, 9, 1–16. [CrossRef] 2. Stanton, J.S.; Anning, D.W.; Brown, C.J.; Moore, R.B.; McGuire, V.L.; Qi, S.L.; Harris, A.C.; Dennehy, K.F.; McMahon, P.B.; Degnan, J.R.; et al. Brackish Groundwater in the United States; USGS Professional Paper 1833; U.S. Geological Survey: Reston, VA, USA, 2017. 3. Gleick, P.H. Roadmap for sustainable water resources in southwestern North America. Proc. Natl. Acad. Sci. USA 2012, 107, 21300–21305. [CrossRef][PubMed] 4. Simone, S.; Figoli, A.; Criscuoli, A.; Carnevale, M.C.; Rosselli, A.; Drioli, E. Preparation of hollow fibre membranes from PVDF/PVP blends and their application in VMD. J. Membr. Sci. 2010, 364, 219–232. [CrossRef] 5. Wang, P.; Chung, T. Recent advances in membrane distillation processes: Membrane development, configuration design and application exploring. J. Membr. Sci. 2015, 474, 39–56. [CrossRef] 6. Wang, K.Y.; Chung, T.S.; Gryta, M. Hydrophobic PVDF hollow fiber membranes with narrow pore size distribution and ultra-thin skin for fresh water production through membrane distillation. Chem. Eng. Sci. 2008, 63, 2587–2594. [CrossRef] 7. Teoh, M.M.; Chung, T.S. Membrane distillation with hydrophobic macrovoid-free PVDF-PTFE hollow fiber membranes. Sep. Purif. Technol. 2009, 66, 229–236. [CrossRef] 8. Zhang, J.; Li, J.D.; Duke, M.; Xie, Z.; Gray, S. Performance of asymmetric hollow fibre membranes in membrane distillation under various configurations and vacuum enhancement. J. Membr. Sci. 2010, 362, 517–528. [CrossRef] 9. Wang, P.; Teoh, M.M.; Chung, T. Morphological architecture of dual-layer hollow fiber for membrane distillation with higher desalination performance. Water Res. 2011, 45, 5489–5500. [CrossRef] 10. Xiao, T.; Wang, P.; Yang, X.; Cai, X.; Lu, J. Fabrication and characterization of novel asymmetric polyvinylidene fluoride (PVDF) membranes by the nonsolvent thermally induced phase separation (NTIPS) method for membrane distillation applications. J. Membr. Sci. 2015, 489, 160–174. [CrossRef] 11. Nejati, S.; Boo, C.; Osuji, C.O.; Elimelech, M. Engineering flat sheet microporous PVDF films for membrane distillation. J. Membr. Sci. 2015, 492, 355–363. [CrossRef] 12. Li, L.; Sirkar, K.K. Influence of microporous membrane properties on the desalination performance in direct contact membrane distillation. J. Membr. Sci. 2016, 513, 280–293. [CrossRef] 13. Huang, F.Y.C.; Laumbach, A.; Reprogle, R.; Medin, C. Geothermal Membrane Distillation in Industrial Greenhouse Applications: Membrane Fabrication and Characterization. Environ. Eng. Sci. 2018, 35, 815–828. [CrossRef] 14. Gryta, M.; Barancewicz, M. Influence of morphology of PVDF capillary membranes on the performance of direct contact membrane distillation. J. Membr. Sci. 2010, 358, 158–167. [CrossRef] 15. Liao, Y.; Loh, C.H.; Wang, R.; Fane, A.G. Electrospun Superhydrophobic Membranes with Unique Structures for Membrane Distillation. ACS Appl. Mater. Interfaces 2014, 6, 16035–16048. [CrossRef] 16. Lee, E.J.; An, A.K.; Hadi, P.; Lee, S.; Woo, Y.C.; Shon, H.K. Advanced multi-nozzle electrospun functionalized titanium dioxide/polyvinylidene fluoride-co-hexafluoropropylene (TiO2/PVDF-HFP) composite membranes for direct contact membrane distillation. J. Membr. Sci. 2017, 524, 712–720. [CrossRef] 17. Tijing, L.D.; Woo, Y.C.; Shim, W.G.; He, T.; Choi, J.S.; Kim, S.H.; Shon, H.K. Superhydrophobic nanofiber membrane containing carbon nanotubes for high-performance direct contact membrane distillation. J. Membr. Sci. 2016, 502, 158–170. [CrossRef] 18. Cho, H.; Jeong, J.; Kim, W.; Choi, D.; Lee, S.; Hwang, W. Conformable superoleophobic surfaces with multi-scale structures on polymer substrates. J. Mater. Chem. A 2016, 4, 8272–8282. [CrossRef] Membranes 2019, 9, 52 15 of 16

19. Boo, C.; Lee, J.; Elimelech, M. Omniphobic Polyvinylidene Fluoride (PVDF) Membrane for Desalination of Shale Gas Produced Water by Membrane Distillation. Environ. Sci. Technol. 2016, 50, 12275–12282. [CrossRef] 20. Li, F.; Wang, Z.; Huang, S.; Pan, Y.; Zhao, X. Flexible, Durable, and Unconditioned Superoleophobic/ Superhydrophilic Surfaces for Controllable Transport and Oil-Water Separation. Adv. Funct. Mater. 2018, 28, 1706867. [CrossRef] 21. Yue, X.; Li, J.; Zhang, T.; Qiu, F.; Yang, D.; Xue, M. In situ one-step fabrication of durable Superhydrophobic-superoleophilic cellulose/LDH membrane with hierarchical structure for efficiency oil/water separation. Chem. Eng. J. 2017, 328, 117–123. [CrossRef] 22. Lee, E.J.; Deka, B.J.; Guo, J.; Woo, Y.C.; Shon, H.K.; An, A.K. Engineering the re-Entrant Hierarchy and Surface Energy of PDMS-PVDF Membrane for Membrane Distillation Using a Facile and Benign Microsphere Coating. Environ. Sci. Technol. 2017, 51, 10117–10126. [CrossRef] 23. Liu, H.; Huang, J.; Chen, Z.; Chen, G.; Zhang, K.Q.; Al-Deyab, S.S. Robust translucent Superhydrophobic PDMS/PMMA film by facile one-step spray for self-cleaning and efficient emulsion separation. Chem. Eng. J. 2017, 330, 26–35. [CrossRef]

24. Woo, Y.C.; Chen, Y.; Tijing, L.D.; Phuntsho, S.; He, T.; Choi, J.S.; Kim, S.H.; Shon, H.K. CF4 plasma-modified omniphobic electrospun nanofiber membrane for produced water brine treatment by membrane distillation. J. Membr. Sci. 2017, 529, 234–242. 25. Ray, S.S.; Gandhi, M.; Chen, S.S.; Chang, H.M.; Dan, C.T.N.; Le, H.Q. Anti-wetting behaviour of a superhydrophobic octadecyltrimethoxysilane blended PVDF/recycled composite membrane for enhanced desalination. Environ. Sci. Water Res. Technol. 2018, 4, 1612–1623. [CrossRef] 26. Ray, S.S.; Chen, S.S.; Sangeetha, D.; Chang, H.M.; Thanh, C.N.D.; Le, Q.H.; Ku, H.M. Developments in forward osmosis and membrane distillation for desalination of waters. Environ. Chem. Lett. 2018, 16, 1247–1265. [CrossRef] 27. Li, L.; Li, B.; Dong, J.; Zhang, J. Roles of silanes and silicones in forming Superhydrophobic and superoleophobic materials. J. Mater. Chem. A 2016, 4, 13677–13725. [CrossRef] 28. Gao, S.; Dong, X.; Huang, J.; Li, S.; Li, Y.; Chen, Z. Rational construction of highly transparent Superhydrophobic coatings based on a non-particle, fluorine-free and water-rich system for versatile oil-water separation. Chem. Eng. J. 2018, 333, 621–629. [CrossRef] 29. Martinetti, C.R.; Childress, A.E.; Cath, T.Y. High recovery of concentrated RO brines using forward osmosis and membrane distillation. J. Membr. Sci. 2009, 331, 31–39. [CrossRef] 30. Hwang, H.J.; He, K.; Gray, S.; Zhang, J.; Moon, S. Direct contact membrane distillation (DCMD): Experimental study on the commercial PTFE membrane and modeling. J. Membr. Sci. 2011, 371, 90–98. [CrossRef] 31. Zhang, J.; Gray, S.; Li, J.D. Modeling heat and mass transfers in DCMD using compressible membranes. J. Membr. Sci. 2012, 387, 7–16. [CrossRef] 32. Adnan, S.; Hoang, M.; Wang, H.; Xie, Z. Commercial PTFE membranes for membrane distillation application: Effect of microstructure and support material. Desalination 2012, 284, 297–308. [CrossRef] 33. Francis, L.; Ghaffour, N.; Alsaadi, A.S.; Nunes, S.P.; Amy, G.L. Performance evaluation of the DCMD desalination process under bench scale and large scale module operating conditions. J. Membr. Sci. 2014, 455, 103–112. [CrossRef] 34. Lin, S.; Nejati, S.; Boo, C.; Hu, Y.; Osuji, C.O.; Elimelech, M. Omniphobic Membrane for Robust Membrane Distillation. Environ. Sci. Technol. Lett. 2014, 1, 443–447. [CrossRef] 35. Lee, J.G.; Kim, Y.D.; Kim, W.S.; Francis, L.; Amy, G.; Ghaffour, N. Performance modeling of direct contact membrane distillation (DCMD) seawater desalination process using a commercial composite membrane. J. Membr. Sci. 2015, 478, 85–95. [CrossRef] 36. Mostafa, M.G.; Zhu, B.; Cran, M.; Dow, N.; Milne, N.; Desai, D.; Duke, M. Membrane Distillation of Meat Industry Effluent with Hydrophilic Polyurethane Coated Polytetrafluoroethylene Membranes. Membranes 2017, 7, 55. [CrossRef] 37. Garcia, J.V.; Dow, N.; Milne, N.; Zhang, J.; Naidoo, L.; Gray, S.; Duke, M. Membrane Distillation Trial on Textile Wastewater Containing Surfectants Using Hydrophobic and Hydrophilic-Coated Polytetrafluoroethylene (PTFE) Membranes. Membranes 2018, 8, 31. [CrossRef] 38. Zhang, J.; Li, J.D.; Gray, S. Effect of applied pressure on performance of PTFE membrane in DCMD. J. Membr. Sci. 2011, 369, 514–525. [CrossRef] Membranes 2019, 9, 52 16 of 16

39. Ghaleni, M.M.; Bavarian, M.; Nejati, S. Model-guided design of high-performance membrane distillation modules for water desalination. J. Membr. Sci. 2018, 563, 794–803. [CrossRef] 40. Wang, K.Y.; Chung, T.S. Polybenzimidazole Nanofiltration Hollow Fiber for Cephalexin Separation. AIChE J. 2006, 52, 1363–1377. [CrossRef] 41. Huang, F.Y.C.; Reprogle, R. Thermal Conductivity of Polyvinylidene Fluoride Membranes for Direct Contact Membrane Distillation. Environ. Eng. Sci. 2018.[CrossRef] 42. Pakomania. Available online: http://www.packomania.com (accessed on 11 March 2019). 43. Reid, R.C.; Prausnitz, J.M.; Poling, B.E. The Properties of Gases and Liquids, 4th ed.; McGraw-Hill: New York, NY, USA, 1988. 44. Sanchez, C. Computational Fluid Dynamic Modeling of Geothermal Membrane Distillation. Master’s Thesis, New Mexico Institute of Mining and Technology, Socorro, NM, USA, 2018. 45. Huang, F.Y.C.; Medin, C.; Arning, A. Mechanical Vibration for the Control of Membrane Fouling in Direct Contact Membrane Distillation. Symmetry 2019, 11, 126. [CrossRef] 46. Qtaishat, M.; Rana, D.; Khayet, M.; Matsuura, T. Preparation and characterization of novel hydrophobic/ hydrophilic polyetherimide composite membranes for desalination by direct contact membrane distillation. J. Membr. Sci. 2009, 327, 264–273. [CrossRef] 47. Zhang, J.; Dow, N.; Ostarcevic, E.; Li, J.D.; Gray, S. Identification of material and physical features of membrane distillation membranes for high performance desalination. J. Membr. Sci. 2010, 349, 295–303. [CrossRef]

© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).