Environ Sci Pollut Res DOI 10.1007/s11356-015-4333-x
RESEARCH ARTICLE
Purifying fluoride-contaminated water by a novel forward osmosis design with enhanced flux under reduced concentration polarization Madhubonti Pal & Sankha Chakrabortty & Parimal Pal & Lassi Linnanen
Received: 23 November 2014 / Accepted: 6 March 2015 # Springer-Verlag Berlin Heidelberg 2015
Abstract For purifying fluoride-contaminated water, a new forward osmosis scheme in horizontal flat-sheet cross flow module was designed and investigated. Effects of pressure, cross flow rate, draw solution and alignment of membrane module on separation and flux were studied. Concentration polarization and reverse salt diffusion got significantly reduced in the new hydrodynamic regime. This resulted in less membrane fouling, better solute separation and higher pure water flux than in a conventional module. The entire scheme was completed in two stages—an upstream forward osmosis for separating pure water from contaminated water and a downstream nanofiltration operation for continuous recovery and recycle of draw solute. Synchronization of these two stages of operation resulted in a continuous, steady-state process. From a set of commercial membranes, two polyamide composite membranes were screened out for the upstream and downstream filtrations. A 0.3-M NaCl solution was found to be the best one for forward osmosis draw solution. Potable water with less than 1 % residual fluoride could be produced at a high flux of 60–62 L m−2 h−1 whereas more than 99 % draw solute could be recovered and recycled in the downstream nanofiltration stage from where flux was 62–65 L m−2 h−1.
Responsible editor: Bingcai Pan M. Pal : S. Chakrabortty : P. Pal (*) Environment and Membrane Technology Laboratory, Department of Chemical Engineering, National Institute of Technology Durgapur, Durgapur 713209, India e-mail: [email protected] P. Pal e-mail: [email protected] L. Linnanen Energy and Environment Department, Lappeenranta University of Technology, 53851 Lappeenranta, Finland
Keywords Fluoride separation . Forward osmosis . Nanofiltration . Cross flow module . Concentration polarization
Nomenclature A Water permeability constant of the membrane D Diffusivity coefficient DS Draw solution ECP External concentration polarization FS Feed solution ICP Internal concentration polarization MCL Maximum contaminant level PRO Pressure retarded osmosis TDS Total dissolved solid (mg L−1) TH Total hardness (mg L−1) TMP Transmembrane pressure (bar)
Introduction The problem of high fluoride concentration in groundwater is one of the most serious toxicological and geo environmental issues in several countries featuring India as the most dominant one (Gizaw 1996; Kartikeyan and Shunmugasundaraj 2000; Gupta et al. 2006; Patra et al. 2010). Intake of high concentration (>10 mg L−1) of fluoride may cause ‘fluorosis’ while consumption of less fluoride can cause ‘dental caries’. WHO has already recommended safeguard limit of fluoride in drinking water to be 1.5 mg/L (WHO 1993). To treat contaminated groundwater with fluoride level of as high as 30 mg L−1 in many cases to the level of WHO recommendation has thrown a challenge to the scientific community. This calls for extremely high degree of separation efficiency that no conventional treatment system can ensure.
Environ Sci Pollut Res
Substantial research on fluoride removal has been conducted on the areas encompassing coagulation–precipitation, adsorption, ion exchange and membrane separation (Choi et al. 2001; Hu and Dickson 2006; Tahaikt et al. 2007; Hou et al. 2010; Chakrabortty et al. 2013; Jiang et al. 2013; Sasaki et al. 2013). Still millions continue to drink fluoride-contaminated unsafe water in a world where around one billion people do not have access to safe drinking water. Search is still continuing for simple, low-cost and easy-to-implement technologies for producing fluoride-free water in the vast fluoride-affected regions of the world. In this context, forward osmosis assumes significance. Forward osmosis (FO) is now an emerging technology that promises high degree of separation of contaminants from water involving low energy consumption, low fouling and high recovery in easy and simple design. The principle of FO desalination lies in transporting water through a semi-permeable membrane by natural osmotic process using a highly concentrated solution (called as draw solution (DS)) that draws water from the feed solution (FS). Depending on the end use of the product water from the FO desalination process, the diluted DS may or may not require posttreatment. Forward osmosis (FO) has progressed substantially towards establishing itself as a viable technology of water purification through the works of several other research groups (Wang et al. 2010; Zhao and Zou 2011; Sant’Anna et al. 2012; Lutchmiah et al. 2014) in the context of seawater/ brackish water desalination, wastewater treatment and liquid food processing. Though FO has widely been studied in the context of desalination, other possible separation areas such as removal of fluoride from groundwater have received very little attention. In the backdrop of the problem of groundwater contamination by fluoride in many countries, this study assumes significance. A FO scheme may be considered as a complete and sustainable one by operating it in continuous mode keeping constant compositions of feed and DSs. This can be ensured only with periodic discharge of concentrated fluoride rejects for subsequent stabilization with addition of fresh feed water to the feeding tank and economical separation of draw solute for recycle. Reported studies have hardly considered the total scheme of continuous steady-state operation, and low flux resulting from internal and external concentration polarization remains a concern (Wang et al. 2010; Zhao and Zou 2011; Sant’Anna et al. 2012; Lutchmiah et al. 2014). Attempt has been made to address the issues of concentration polarization through better designs of membranes and draw solutes. Widjojo et al. (2011) used 2-M NaCl DS and TFC polyamide membranes in PRO mode but found a flux of only 15 L m−2 h−1. Data on concentration polarization and fouling study over a reasonably long operation time are missing. Recovery of draw solute by RO involves high energy consumption that offsets the gain of low energy involvement in
the FO. Thus, the major problem of flux decline after a certain time of operation following concentration polarization or reverse salt diffusion still stands in the way of implementation of FO in water purification. Diffusion of draw solute is often accompanied by back diffusion of feed solute from the active membrane surface to the bulk FS. Thus, reverse salt diffusion progressively increases the osmotic pressure of the FS, thereby seriously limiting separation of the desired solute (Boo et al. 2012). This study through a novel design overcomes the major hurdles of FO in the context of removal of fluoride from contaminated groundwater. This study for the first time proposes a complete scheme consisting of an upstream fluoride separation loop by FO and recovery of draw solute for recycling in a nanofiltration module. The objective of the study is to examine effectiveness of a new FO system design in horizontal alignment and in steady-state continuous mode with synchronization of one upstream FO loop with a downstream nanofiltration (NF) loop.
Materials and methods Chemicals and membranes All the chemical reagents (NaCl and MgSO4) used in the investigation were of reagent grade, and no further purification was done, and these were procured mostly from Merck (Germany). Sodium hydroxide pellets and HCl used for maintaining the pH during analysis were procured from Merck. Thin-film composite polyamide nanofiltration (NF-1 and NF-2) membranes were procured from Sepro Membranes Inc. (USA) whereas asymmetric structure-based cellulose triacetate FO membrane (OsMem™ CTA-NW) was purchased from Hydration Technology Innovations (USA). The membrane surface area of each module was 100 cm2. Major characteristics of the membranes used in the investigation are presented in Table 1. The groundwater used during investigation was collected from some fluoride-affected areas of the Eastern part of India. In such water, fluoride concentration varied from 18 to 20 ppm. Physico-chemical characterization of this groundwater has been presented in Table 2. Experimental set-up Two loops—one for FO and the other for nanofiltration—used in the investigation were connected in a single line in the whole experimental set-up. The FO loop consists of a flatsheet cross flow membrane module with necessary accessories like feed water tank and DS tank made of stainless steel and connected to the FO membrane system via circulating pumps, flow meters and pressure gauges as shown in Fig. 1. Peristaltic pumps are used (Miclins India, VSP-100, India) for
Environ Sci Pollut Res Table 1 Characteristics of three commercial membrane used in investigation
Characteristics
Geometry Materials Thickness pH resistance Max. temperature Max. pressure Pore radius Solute rejection MgSO4 NaCl Water flux in
Unit
Polyamide composite membrane (at 10 bar)
FO membrane
NF-1
NF-2
(at 0.35 bar) OsMem™ CTA-NW
– – cm – K bar nm
Flat sheet Polyamide 0.0165 2–11 323 83 0.53
Flat sheet Polyamide 0.0165 2–11 323 83 0.57
Flat sheet Cellulose triacetate – 3–8 344 0.7 –
% % L m−2 h−1
99.5 90 110
97 50 195
– 99 4
circulating the feed solution (FS) and DS through the FO module while a diaphragm pump is used for circulating diluted DS through the downstream nanofiltration module. The upstream and downstream pressure gauges indicate the transmembrane pressures. The cross flow rate through the system is monitored and controlled using rotameter and by-pass valves. The module is so designed that the flat-sheet membrane remains in a horizontal plane on a perforated stainless steel support while feed water and DS flow tangentially in counter-current directions along the top and bottom surfaces of the membrane, respectively. The bottom chamber of the module is so designed that DS enters the same through one of its inlets at one end and, on crossing tangentially along the membrane surface, leaves the chamber through an outlet maintained at a much higher level than the level of the membrane. This ensures complete immersion of the bottom surface of the membrane in the DS while allowing a continuous sweeping fluid action along its bottom surface. The feed water that flows tangentially along the top surface of the membrane ensures sweeping fluid action on the top layer. This very
Table 2 Characteristics of fluoride-contaminated groundwater before and after treatments
design minimizes concentration polarization, membrane fouling and also back diffusion of the DS, thereby ensuring a high unidirectional volumetric flux that is water transport from feed to DS only. As continuous FO results in dilution of the DS, concentration of the same needs to be done to sustain the FO process. Thus, a downstream nanofiltration membrane module in flatsheet cross flow mode is operated simultaneously for recovery of pure water from the DS while recycling the draw solute to the system. A diaphragm pump (Milton Roy India Pvt. Ltd.) operated at 10–14 bar maintains circulation of the DS through the flat-sheet cross flow nanofiltration module. Operating pressure and cross flow rates are monitored through control valves, rotameter and pressure gauges as illustrated in Fig. 1. Experimental procedure All investigations were carried out in continuous flow mode using flat-sheet cross flow membrane modules with real groundwater from some fluoride-affected areas of
Water parameters
Units
Feed water compositions
Finally treated DS/NaCl (0.3 M)
Finally treated water DS/MgSO4 (0.3 M)
Max. water permissible limit (WHO)
pH Salinity Conductivity TDS TH Chloride Sulphate Iron Fluoride
– mg L−1 μs cm−1 mg L−1 mg L−1 mg L−1 mg L−1 mg L−1 mg L−1
9.8 0.19 670 340 280 85 2.9 0.7 19.8
7.4 2.65 4.8 4.6 5.6 196 0.1 0.03 0.25
7.6 2.32 4.7 4.9 6.1 3.4 174 0.05 1.1
8.5 1000 – 500 600 250 500 1.0 1.5
Environ Sci Pollut Res
Fig. 1 Schematic diagram of pilot-scale forward osmosis–nanofiltration continuous system for removal of fluoride from contaminated groundwater: (1) Make-up feed water tank, (2) water extraction pump, (3) rotameter 1, (4) pH probe 1, (5) concentrated retantate, (6) rotameter (7) pressure gauge 1, (8) pressure gauge 2, (9) pressure gauge 3, (10) rotameter 3, (11) rotameter 4, (12) recycle, (13) DS tank, (14) pH probe
2, (15) draw solution inlet, (16) diluted draw solution, (17) peristaltic pump 1, (18) peristaltic pump 1, (19) diaphragm pump, (20) stabilization tank, (21) stabilized fluoride, (22) stabilizing agent, (23) nanofiltration system, (24) forward osmosis system, (25) feed tank, (26) concentrated fluoride rejects and (27) fluoride-free water
India. Effective filtration surface of each membrane module was 100 cm2. Fluoride-contaminated groundwater was pumped through the forward osmosis (FO) module from where water was extracted in a DS of either NaCl or MgSO4, leaving behind fluoride and other possible contaminants. The diluted DS was then passed on to another nanofiltration membrane module for recovering pure water as the filtrate and recycling the draw solute in the DS loop. Filtration modules were run over long hours in a continuous mode where feed water and DS compositions were maintained at the same level through continuous withdrawal of fluoride rejects and addition of fresh water in the FO loop and recovery of pure water from the downstream nanofiltration module at the same rate of dilution of the DS. Effects of transmembrane pressure, cross flow rates, concentration of the fluoride in the feed water and DSs were investigated during the continuous run spanning over
more than 120 h. Each run of FO experiment was carried out with the active layer of the membrane facing the FS and the support layer facing the DS and that was supposed to be able to afford higher rejection with higher water fluxes due to the reduced internal concentration polarization. The cross flow rates for both draw and the FS were varied between 21 and 42 cm s−1 whereas pressure of the FO module was in the range of 1 to 1.2 bar. The NF module was run at 10–14-bar operating pressure. Both the FS and the DS were maintained at the same temperature. Fluoride rejection and reverse salt flux were assessed by periodic analysis of the samples from the DS tank and feed tank, respectively. NF modules were initially run at different cross flow rates and at different applied pressures to arrive at the best possible cross flow rate and operating pressure. Thus, 13-bar operating pressure and 800 L h−1 cross flow rate were selected for
Environ Sci Pollut Res
continuous running of the system at steady state that ensured high retention of the draw solute while yielding highest possible pure water flux. Concentrated fluoride rejects were subjected to subsequent stabilization and disposal.
with relevant electrodes (Thermo Electron Corporation, USA) where total dissolved solids, conductivity and salinity were determined by InoLab Cond 720 and with electrode TetraCon 325 (WTW, Germany). Iron concentration was also measured by atomic absorption spectrophotometer (AAS100, PerkinElmer, Germany).
Analytics Chemical analysis
Analysis of membrane morphology
Analysis of fluoride has been carried out in Orion 4-Star pH ISE Benchtop Ion Meter of Thermo Electron Corporation, USA. The ion meter was first calibrated using the previously prepared fluoride samples of known strength. Removal efficiency of fluoride was calculated using the bulk fluoride concentration (CF,F) which is present in feed sample and the residual concentration (CP,F) in the permeate, respectively, by the equation below:
Membrane morphology was analyzed by scanning electron microscopy (SEM, Hitachi S-3000 N, Japan) before and after each filtration run (over long periods) to find out the degree of membrane fouling. During SEM analysis, membrane pieces were freeze-fractured in liquid nitrogen and then gold-coated in ion sputter at 15 kV.
R j; F
C P; F ð% Þ ¼ 1 − 100 C F; F
Measurement of water and reverse salt flux ð8Þ
Determination of pH and of concentrations of chloride and sodium was done by Orion 4-Star ISE Benchtop Ion Meter > Measured Water Flux ¼
The weight balance was used for determining the water flux by measuring the weight change of the permeate DS tank during each experimental running, and numerically, water flux can be calculated by the following equation:
Change of weight of the permeate density of the water effective membrane area time interval
Reverse salt flux was assessed by analyzing chloride and sulphate concentrations in the feed tank periodically.
Results and discussion Effects of DS on fluoride rejection and water flux in FO system The osmotic pressure depends on DS (DS) concentration that influences mass transport and overall process performance. A perfect DS should have high osmotic pressure, have non-toxic character and should be easily recoverable in the re-concentration step. Two types of DS were investigated in this study such as NaCl and MgSO4 with molar concentrations ranging from 0.1 to 1 M. Due to high solubility, high osmotic pressure and low cost, NaCl appears to be the most widely used DS in FO process (Cath et al. 2006; Achilli et al. 2010) whereas MgSO4 has been used as a draw solute due to its insignificant retarded forward diffusion resulting in negligible reverse solute flux. Figure 2 indicates how changes in concentration of DS
affect fluoride rejection (primary axis) and water flux (secondary axis) in two different membranes, and this also shows that the DS concentration has a strong positive correlation with volumetric flux. Figure 2a illustrates that at an optimum concentration of 0.3 M NaCl, the NF-2 membrane yields a volumetric water flux of 62 L m −2 h−1 whereas the water flux with MgSO4 as DS is 52 L m−2 h−1. Effects of DS concentration on fluoride rejection for two different membranes are illustrated in Fig. 2a. Fluoride rejection increases linearly with increasing DS concentration, and at a draw solute concentration of 0.3 M NaCl, almost 100 % fluoride rejection is observed by NF-2 and rejection reaches 98 % in case of 0.3 M MgSO4 at 1-bar pressure and 120 L h−1 feed flow rate. Similarly, the DS and the corresponding osmotic pressures influence mass transport and overall process performance significantly. A higher concentration of DS produces a higher osmotic pressure for water transport through the membrane. Higher concentration of DS and hence higher osmotic pressure that forces larger amount of water as water flux through the membrane eventually reduce solute flux across the membrane in an uncoupled transport
Environ Sci Pollut Res
Fig. 2 Fluoride rejection and water flux during forward osmosis under different draw solution concentrations of two different membranes. a Effects of DS concentration on fluoride removal and pure water in NF2 membrane. b Effects of DS concentration on fluoride removal and pure water in OsMem™ CTA-NW membrane. Fluoride rejection (DS/NaC1) (blue triangle), fluoride rejection (DS/ MgSO4) (red asterisk), flux (DS/ NaC1) (red broken line) and flux (DS/MgSO4) (orange broken line). Experimental conditions: total fluoride concentration of 19.8 mg L−1, pH 9.8, draw solution concentration range 0.1–0.5 mol L−1, pressure 1 bar, feed side cross flow rate 120 L h−1, draw solution side cross flow rate 10 L h−1 and temperature 308 K
process. This explains higher fluoride rejection following higher water flux. By virtue of its high solubility, NaCl produces a higher osmotic pressure than MgSO4 for the same concentration in a solution. Low diffusivity of MgSO4 compared to that of NaCl is another reason behind its significantly reduced reverse salt diffusion (Lutchmiah et al. 2014). Figure 2b shows the effects of draw solution concentrations on fluoride rejection and water flux in case of OsMem™ CTA-NW membrane. Almost 100 and 98 % fluoride rejection are observed at 0.3-M concentration for both of the draw solutes (NaCl and MgSO4). But, the water flux observed in case of OsMem™ CTA-NW membrane is much lower (5–7 L m−2 h−1) than that of NF-2 membrane at 0.6-bar pressure and 120 L h−1 feed flow rate. An operating pressure of 0.6 bar was maintained as the maximum
Fig. 3 Fluoride removal efficiency and water flux during forward osmosis under different draw solution concentrations of two different membranes. a Effects of hydraulic transmembrane pressure on fluoride removal efficiency and pure water in NF-2 membrane. Fluoride rejection (DS/NaC1) (blue triangle), fluoride rejection (DS/MgSO4) (red asterisk), flux (DS/NaC1) (blue broken line) and flux (DS/MgSO4) (red broken line). Experimental conditions: fluoride concentration of 19.8 mg L−1, pH 9.8, draw solution concentration range 0.3 mol L−1, pressure range 0–1.2 bar, feed side cross flow rate 120 L h−1, draw solution side cross flow rate 10 L h−1 and temperature 308 K. b Effects of feed cross flow rate on fluoride removal efficiency and pure water in NF-2 membrane. Fluoride rejection (DS/NaC1) (blue diamond), fluoride rejection (DS/ MgSO4) (red big dot), flux (DS/NaC1) (blue green broken line) and flux (DS/MgSO4) (orange broken line). Experimental conditions: total fluoride concentration of 19.8 mg L −1 , pH 9.8, draw solution concentration range 0.3 mol L−1, pressure range 1 bar, feed side cross flow rate range 30–150 L h−1, draw solution side cross flow rate 10 L h−1 and temperature 308 K
operating limit of pressure for CTA-NW membrane is 0.7 bar (HTI, USA). Effects of hydraulic transmembrane pressure and feed cross flow rate on fluoride rejection and water flux in upstream FO system Figure 3 shows the effect of hydraulic transmembrane pressure (TMP) and feed cross flow rate on fluoride rejection and water flux in a screened NF-2 membrane and two different
Environ Sci Pollut Res
draw solutions with fixed concentrations (0.3 M NaCl, MgSO4). Though intrinsically water transport in FO system is by virtue of osmotic pressure difference only, hydraulic TMP often plays a significant role in some particular modules like spiral wound or capillary type where such TMP helps overcome the hydraulic resistances of the flow channels. Considering that in a flat-sheet cross flow membrane module of this study, such hydraulic resistances are far less than in the other conventional modules, at 1-bar pressure, around 100 and 98 % fluoride were rejected (Fig. 3a) by NF-2 membrane at a feed flow rate of 120 L h−1 and draw solution flow rate of 10 L h−1 while using 0.3 M NaCl and 0.3 M MgSO4, respectively, as a draw solution. The anionic form of F ions gets repelled to some higher degree by Donnan exclusion mechanism of the composite polyamide membranes (negative–negative repulsion) (Chakrabortty et al. 2013). In the solution diffusion mechanism of transport through composite polyamide membranes, solute flux and solvent flux are uncoupled and this explains why under enhanced pressure, solvent flux increases with commensurate decrease of solute flux or rather increase of solute rejection (Kumar et al. 2011). Figure 3a (secondary axis) also shows the effect of TMP on volumetric water flux at a feed flow rate of 120 L h−1 and draw solution flow rate of 10 L h−1 in the FO system. A strong positive correlation with TMP and water flux has been shown in Fig. 3 where the highest flux of 62 and 52 L m−2 h−1 was achieved with 0.3 M NaCl and 0.3 M MgSO4 draw solution concentration, respectively. Figure 3b shows an effect of cross flow rate of feed water on fluoride rejection and permeate flux during FO of fluoridecontaminated groundwater. A positive correlation with cross flow rate on water flux as well as fluoride rejection was observed in all cases. At 120 L h−1 of feed flow rate, very close to 100 and 98 % of fluoride rejection were found with 0.3 M NaCl and 0.3 M MgSO4 draw solution concentration, respectively, by keeping fixed pressure 1 bar whereas flux was found 62 and 52 L m−2 h−1 at the same condition. Cross flow rate plays an important role in reducing concentration polarization and increasing turbulence on the membrane surface by its sweeping action and thus reducing fouling as well as enhancing the hydraulic TMP which in turn enhances solvent flux (Kumar et al. 2011). Uncoupled nature of the solute and solvent fluxes results in higher retention of fluoride as flux increases with increasing cross flow rate. Effects of applied pressure and draw solution concentration on RSF Reverse salt diffusion in terms of reverse salt flux (RSF) of NaCl and MgSO4 draw solutions has been shown in Fig. 4 under varying hydraulic TMP and DS concentrations. In Fig. 4a, an inversely proportional relationship between
Fig. 4 Investigation on reverse salt flux (RSF) during forward osmosis of NF-2 membrane. a Effects of hydraulic transmembrane pressure on reverse salt flux. Reverse salt flux (DS/NaC1) (blue triangle) and reverse salt flux (DS/MgSO4) (red asterisk). Experimental conditions: fluoride concentration of 19.8 mg L −1 , pH 9.8, draw solution concentration range 0.3 mol L−1, pressure range 0–2 bar, feed side cross flow rate 120 L h−1, draw solution side cross flow rate 10 L h−1 and temperature 308 K. b Effects of DS concentration on reverse salt flux. Reverse salt flux (DS/NaC1) (blue triangle) and reverse salt flux (DS/ MgSO4) (red asterisk). Experimental conditions: fluoride concentration of 19.8 mg L−1, pH 9.8, draw solution concentration range 0.1– 0.5 mol L−1, feed side cross flow rate 120 L h−1, draw solution side cross flow rate 10 L h−1, pressure 1 bar and temperature 308 K
hydraulic TMP and RSF is observed. In Fig. 4a, a very little change of RSF of MgSO 4 is observed with increasing pressure in NF-2 membrane whereas decline of RSF with increasing pressure is relatively high for NaCl draw solute. Monovalent ion of NaCl possessing a high diffusivity is the cause of such a high RSF whereas bivalent ion of MgSO4 possessing lower diffusivity, due its large hydrated radii, leads to its lower RSF (Cornelissen et al. 2008). The trend of decline in RSF with applied pressure may be attributed to physical changes in the membrane active
Environ Sci Pollut Res
Effects of hydraulic TMP and feed cross flow rate on draw solute recovery and pure water flux in downstream nanofiltration module
Fig. 5 Salt removal and pure water flux during nanofiltration under varying transmembrane pressures and cross flow rates. a Effects of transmembrane pressure on salt removal efficiency and pure water flux in NF-1 membrane. MgSO4 rejection (blue diamond), NaC1 rejection (big red dot) and flux (blue broken line). Experimental conditions: draw solution concentration of 0.3 mol L−1, pressure range 0–14 bar, pH 8, cross flow rate 750 L h−1 and temperature 308 K. b Effects of cross flow rate on salt removal efficiency and pure water flux in NF-1 membrane. MgSO4 rejection (blue diamond), NaC1 rejection (big red dot) and flux (blue broken line). Experimental conditions: draw solution concentration of 0.3 mol L−1, pressure 13 bar, pH 8, cross flow rate range 0–800 L h−1 and temperature 308 K
layer as a function of the applied pressure. Increasing pressure at the membrane surface leads to compression of the interface between the thin active layer and support layer of the membranes, thus reducing the possibility of RSF and salt permeability of the membrane, while slightly affecting water flux (Coday et al. 2013). Variation of RSF with different draw solution concentrations is observed in Fig. 4b. The figure shows that the RSF has a strong positive correlation with draw solution concentration. At high concentration of the draw solution, the possibility of ions permeating through the membrane through RSF increases. Effect of draw solution concentration on RSF is more distinct in case of NaCl than in case of MgSO4 because of differences in ionic characters.
Figure 5 exhibits strong positive correlation of recovery of both the draw solutes (MgSO4 and NaCl) with hydraulic TMP and feed cross flow rate during nanofiltration. Figure 5a (primary axis) shows that at 13-bar TMP, retention of MgSO4 and NaCl by nanofiltration membranes (NF-1) reaches 99 and 98.7 %, respectively, at a feed flow rate of 800 L h−1. Such solute retention behaviour of nanofiltration membrane is attributed to the very solution–diffusion mechanism of mass transport through nanofiltration membrane. In this mechanism, solute and solvent fluxes are uncoupled, and thus, when increased TMP causes an increase in solvent flux, the solute flux decreases resulting in its higher retention or recovery for recycling. The hydrated radii and nature of ions of MgSO4 are different from those of NaCl resulting in a higher retention in case of MgSO4 than in case of NaCl. According to Donnan exclusion principle, during transport of negatively charged ions over the polyamide composite membrane, retention of the draw solutes takes place because of negative–negative charge repulsion (Bowen and Welfoot 2002). Apart from Donnan exclusion mechanism, sieving mechanism also plays a vital role in separation of the species with large hydrated radii. Here again, ions of MgSO4 being much larger than their counterparts of NaCl are retained more resulting in very close to 100 % recovery and recycle of MgSO4 as draw solute. Recovery of NaCl is also very high (98.7 %) by the selected nanofiltration membrane (NF-1) that ultimately produces potable water with some residual chloride which is well below the WHO-prescribed level (250 mg L−1). The flux data as presented in Fig. 5a (secondary axis) shows that salt-free water flux increases with increase of TMP and reaches 62 L m−2 h−1 at 13-bar TMP for a feed flow rate of 800 L h−1. Such flux behaviour of the nanofiltration membranes with pressure during investigation is well established in the literature (Chakrabortty et al. 2014). Figure 5b shows a pronounced effect of feed cross flow rate on draw solute recovery and water flux during nanofiltration, and it shows a strong positive correlation with draw solute recovery and water flux. Figure 5b (primary axis) shows that the recovery efficiency of draw solute increases from 68 to almost 100 % as cross flow rate increases from 400 to 800 L h−1 for the nanofiltration membrane (NF-1) with magnesium sulphate as draw solute at 13-bar TMP. Similar trend is observed in case of NaCl as draw solute. The NF-1 membrane (Fig. 5b, secondary axis) produces a pure water flux of 68 L m−2 h−1 under the operating conditions of 13-bar TMP and 800 L h−1 cross flow rate. Cross flow rate plays an important role in reducing concentration polarization. With increase in the cross flow rate, sweeping action on the active membrane
Environ Sci Pollut Res
the solute and solvent fluxes, also results in higher removal efficiency in this case. Effects of operation time on water flux in upstream FO system and downstream NF system: a fouling study
Fig. 6 Effects of operational time on water flux in FO and NF system. Flux in FO system-NF-2 membrane (big red dot) and flux in NF systemNF-1 membrane (blue triangle). Experimental conditions: fluoride concentration of 19.8 mg L−1, pH 9.8, time range 0–120 h, draw solution concentration of 0.3 mol L−1, feed cross flow rate in FO system 120 L h−1, pressure in FO system 1 bar, feed cross flow rate in NF system 800 L h −1 , pH 8, pressure in NF system 13 bar and temperature 308 K
surface area also increases, thereby reducing concentration polarization. This leads to maximization of the effective membrane surface area available for a given separation. In other flow modes, increase in solvent flux normally accompanies increase in concentration polarization. However, in the very flat-sheet cross flow module such enhancement of flux, following increased cross flow does not really lead to any increase in concentration polarization. Again, reduction of concentration polarization develops convective force that in turn enhances the solvent flux and, due to the uncoupling nature of Fig. 7 Scanning electron microscope (SEM) image before and after investigation. a NF-2: before forward osmosis. b NF-2: after forward osmosis. c NF-1: before nanofiltration. d NF-1: after nanofiltration
Build-up of concentration polarization leading to membrane fouling and hence decline in flux is quite common in membrane filtration. But, such fouling can be largely eliminated through provision of appropriate hydrodynamic conditions associated with a module design. Flat-sheet cross flow membrane module is one such system that provides for sweeping action of fluid on the membrane surface, thus significantly reducing concentration polarization. A drop of 7 % of flux was observed in the module of this study after a long 120 h of operation (Fig. 6). This indicates a significant improvement of the new design over the existing ones that remarkably control concentration polarization and membrane fouling, eventually ensuring long time operation at a steady flux. Reported flux decline is quite high compared to that in this study. For example, Zhao and Zou (2011) and Tang and Yong Ng (2008) found 22–33 % water flux decline within only 15– 20 h of operation while using their module. Choi et al. (2009) studied a plate and frame module and observed around 20 % of flux decline in 100 h of operation which is against only 6 % flux decline in this design over the same duration of filtration. Figure 7 represents the SEM images of the NF-1 and NF-2 membranes before and after the FO and nanofiltration runs, respectively. The figure shows that the membranes do not undergo major morphological changes possibly due to the very hydrodynamic design of the cross flow module. Whatever small fouling that occurs, it could be removed by
Environ Sci Pollut Res
rinsing with 0.1 N NaOH and 0.01 M HNO3. This establishes reversible nature of fouling that occurs on NF-1 membrane.
of India (DST-INSPIRE fellowship program (DST/INSPIRE Fellowship/ 2012/271 and DST-FIST program).
Stabilization of fluoride rejects for safe disposal
References
Stabilization of fluoride rejects was done by adding calcium chloride into the concentrated rejects that formed an insoluble calcium fluoride compound. For safe disposal, fluoride is stabilized in a solid matrix through chemical precipitation using appropriate doses of calcium chloride under response surface (RSM) optimized conditions of the major governing parameters (initial concentration of the fluoride, dosses of calcium chloride, pH and stirring time). Fluoride stabilized in solid matrix is tested for possible leaching out following toxic characteristic leaching procedure (TCLP) at optimum pH condition. Concentrated fluoride rejects were stabilized in a solid matrix of calcium fluoride (CaF2) under a set of optimum conditions determined through response surface optimization technique of Design Expert Software (version 8.1). The maximum fluoride stabilization efficiency of 98 % was achieved under such conditions. Concentration of fluoride after 1 month running of the unit stands at 500 mg L−1, which requires a calcium chloride dose of 10 g L−1, stirring time of 10 min and pH of 5. The weight of sludge was around 4.5 kg (for 150-m3 drinking water production). The leachate concentration of fluoride in the TCLP test was found to be 25 mg L−1 which was well below the maximum leachate concentration (150 mg L−1). This stabilized solid calcium fluoride passed the TCLP test indicating its class of non-hazardous waste to be disposed safely through landfill.
Conclusion The study culminated in development of a new FO– nanofiltration integrated system capable of producing safe potable water from fluoride-contaminated groundwater. This is an approach different from the existing ones in alignment of membrane modules, in integration of the appropriate membranes and in synchronized operation of two stages of filtration. The major hurdle of low flux and reverse salt diffusion could be effectively reduced. The system configuration ensures water transport at high fluxes under significantly reduced concentration polarization and efficient recovery of draw solute at high flux. Both stages involve low energy. Findings of the investigation using fluoride-contaminated real groundwater instead of a model solution are expected to raise scale-up confidence and pave the way for implementing such a total scheme in fluoride-affected regions of the world. Acknowledgments The authors acknowledge with thanks financial support from the Department of Science and Technology, Government
Achilli A, Cath TY, Childress AE (2010) Selection of inorganic based draw solutions for forward osmosis applications. J Membr Sci 364(1–2):233–241 Boo C, Lee S, Elimelech M, Meng Z, Hong S (2012) Colloidal fouling in forward osmosis: role of reverse salt diffusion. J Membr Sci 390– 391:277–284 Bowen WR, Welfoot JS (2002) Modeling the performance of membrane nanofiltration—critical assessment and model development. Chem Eng Sci 57:1121–1137 Cath TY, Childress AE, Elimelech M (2006) Forward osmosis: principles, applications, and recent developments. J Membr Sci 281(1–2): 70–87 Chakrabortty S, Roy M, Pal P (2013) Removal of fluoride from contaminated groundwater by cross flow nanofiltration: transport modelling and economic evaluation. Desalination 313:115–124 Chakrabortty S, Sen M, Pal P (2014) Arsenic removal from contaminated groundwater by membrane-integrated hybrid plant: optimization and control using visual basic platform. Environ Sci Pollut Res 21: 3840–3857 Choi S, Yun A, Hong S, Ahn K (2001) The effect of co-existing ions and surface characteristics of nanomembrane on the removal of nitrate and fluoride. Desalination 133:53–64 Choi Y, Choi J, Oh H, Lee S, Yang DR, Kim JH (2009) Toward a combined system of forward osmosis and reverse osmosis for seawater desalination. Desalination 247:239–246 Coday BD, Heil DM, Xu P, Cath TY (2013) Effects of transmembrane hydraulic pressure on performance of forward osmosis membranes. Environ Sci Technol 47:2386–2393 Cornelissen ER, Harmsen D, de Korte KF, Ruiken CJ, Qin JJ, Oo H, Wessels LP (2008) Membrane fouling and process performance of forward osmosis membranes on activated sludge. J Membr Sci 319(1–2):158–168 Gizaw B (1996) The origin of high bicarbonate and fluoride concentration in waters of the main Ethiopian Rift Valley, East African Rift System. J Afr Earth Sci 22:391–402 Gupta S, Banerjee S, Saha DJK, Mondal N (2006) Fluoride geochemistry of groundwater in Nalhati-1 block of the Birbhum District, West Bengal, India, Research report. Fluoride 39(4): 318–320 Hou D, Wang J, Zhao C, Wang B, Luan Z, Sun X (2010) Fluoride removal from brackish groundwater by direct contact membrane distillation. J Env Sci 22(12):1860–1867 Hu K, Dickson JM (2006) Nanofiltration membrane performance on fluoride removal from water. J Membr Sci 279:528–529 Jiang K, Zhou K, Yang Y, Du H (2013) A pilot-scale study of cryolite precipitation from high fluoride-containing wastewater in a reaction-separation integrated reactor. J Env Sci 25(7):1331–1337 Kartikeyan G, Shunmugasundaraj A (2000) Isopleth mapping and in-situ fluoride dependence on water quality in the Krishnagiri block of Tamil Nadu in South India. Fluoride 33:121–127 Kumar R, Bhakta P, Chakraborty S, Pal P (2011) Separating cyanide from coke wastewater by cross flow nanofiltration. Sep Sci Technol 46: 2119–2127 Lutchmiah K, Verliefde ARD, Roest K, Rietveld LC, Cornelissen ER (2014) Forward osmosis for application in wastewater treatment: a review. Water Res 58:179–197
Environ Sci Pollut Res Patra PK, Mandal B, Chakrabortty S (2010) The hydro geochemistry of fluoride rich groundwater in Birbhum District of West Bengal, India. Ecoscan 4(2&3):209–211 Sant’Anna V, Marczak LDF, Tessaro IC (2012) Membrane concentration of liquid foods by forward osmosis: process and quality view. J Food Eng 111:483–489 Sasaki K, Yoshida M, Ahmmad B, Fukumoto N, Hirajima T (2013) Sorption of fluoride on partially calcined dolomite. Colloids Surf A: Physicochem Eng Asp 435:56–62 Tahaikt M, Habbani RE, AitHaddou A, Achary I, Amor Z, Taky M, Alami A, Boughriba A, Hafsi M, Elmidaoui A (2007) Fluoride removal from groundwater by nanofiltration. Desalination 212:46–53 Tang W, Yong Ng H (2008) Concentration of brine by forward osmosis: performance and influence of membrane structure. Desalination 224:143–153
Wang KY, Ong RC, Chung TS (2010) Double–skinned forward osmosis membranes for reducing internal concentration polarization within the porous sublayer. Indus Eng Chem Res 49: 4824–4831 Widjojo N, Chung TS, Weber M, Maletzko C, Warzelhan V (2011) The role of sulphonated polymer and macrovoid-free structure in the support layer for thin-film composite (TFC) forward osmosis (FO) membranes. J Membr Sci 383:214–223 World Health Organization (WHO), Guidelines for drinking water quality, 1, World Health Organization, Geneva, 1993, pp. 45– 46 Zhao S, Zou L (2011) Effects of working temperature on separation performance, membrane scaling and cleaning in forward osmosis desalination. Desalination 278:157–164
RESEARCH ARTICLE
Purifying fluoride-contaminated water by a novel forward osmosis design with enhanced flux under reduced concentration polarization Madhubonti Pal & Sankha Chakrabortty & Parimal Pal & Lassi Linnanen
Received: 23 November 2014 / Accepted: 6 March 2015 # Springer-Verlag Berlin Heidelberg 2015
Abstract For purifying fluoride-contaminated water, a new forward osmosis scheme in horizontal flat-sheet cross flow module was designed and investigated. Effects of pressure, cross flow rate, draw solution and alignment of membrane module on separation and flux were studied. Concentration polarization and reverse salt diffusion got significantly reduced in the new hydrodynamic regime. This resulted in less membrane fouling, better solute separation and higher pure water flux than in a conventional module. The entire scheme was completed in two stages—an upstream forward osmosis for separating pure water from contaminated water and a downstream nanofiltration operation for continuous recovery and recycle of draw solute. Synchronization of these two stages of operation resulted in a continuous, steady-state process. From a set of commercial membranes, two polyamide composite membranes were screened out for the upstream and downstream filtrations. A 0.3-M NaCl solution was found to be the best one for forward osmosis draw solution. Potable water with less than 1 % residual fluoride could be produced at a high flux of 60–62 L m−2 h−1 whereas more than 99 % draw solute could be recovered and recycled in the downstream nanofiltration stage from where flux was 62–65 L m−2 h−1.
Responsible editor: Bingcai Pan M. Pal : S. Chakrabortty : P. Pal (*) Environment and Membrane Technology Laboratory, Department of Chemical Engineering, National Institute of Technology Durgapur, Durgapur 713209, India e-mail: [email protected] P. Pal e-mail: [email protected] L. Linnanen Energy and Environment Department, Lappeenranta University of Technology, 53851 Lappeenranta, Finland
Keywords Fluoride separation . Forward osmosis . Nanofiltration . Cross flow module . Concentration polarization
Nomenclature A Water permeability constant of the membrane D Diffusivity coefficient DS Draw solution ECP External concentration polarization FS Feed solution ICP Internal concentration polarization MCL Maximum contaminant level PRO Pressure retarded osmosis TDS Total dissolved solid (mg L−1) TH Total hardness (mg L−1) TMP Transmembrane pressure (bar)
Introduction The problem of high fluoride concentration in groundwater is one of the most serious toxicological and geo environmental issues in several countries featuring India as the most dominant one (Gizaw 1996; Kartikeyan and Shunmugasundaraj 2000; Gupta et al. 2006; Patra et al. 2010). Intake of high concentration (>10 mg L−1) of fluoride may cause ‘fluorosis’ while consumption of less fluoride can cause ‘dental caries’. WHO has already recommended safeguard limit of fluoride in drinking water to be 1.5 mg/L (WHO 1993). To treat contaminated groundwater with fluoride level of as high as 30 mg L−1 in many cases to the level of WHO recommendation has thrown a challenge to the scientific community. This calls for extremely high degree of separation efficiency that no conventional treatment system can ensure.
Environ Sci Pollut Res
Substantial research on fluoride removal has been conducted on the areas encompassing coagulation–precipitation, adsorption, ion exchange and membrane separation (Choi et al. 2001; Hu and Dickson 2006; Tahaikt et al. 2007; Hou et al. 2010; Chakrabortty et al. 2013; Jiang et al. 2013; Sasaki et al. 2013). Still millions continue to drink fluoride-contaminated unsafe water in a world where around one billion people do not have access to safe drinking water. Search is still continuing for simple, low-cost and easy-to-implement technologies for producing fluoride-free water in the vast fluoride-affected regions of the world. In this context, forward osmosis assumes significance. Forward osmosis (FO) is now an emerging technology that promises high degree of separation of contaminants from water involving low energy consumption, low fouling and high recovery in easy and simple design. The principle of FO desalination lies in transporting water through a semi-permeable membrane by natural osmotic process using a highly concentrated solution (called as draw solution (DS)) that draws water from the feed solution (FS). Depending on the end use of the product water from the FO desalination process, the diluted DS may or may not require posttreatment. Forward osmosis (FO) has progressed substantially towards establishing itself as a viable technology of water purification through the works of several other research groups (Wang et al. 2010; Zhao and Zou 2011; Sant’Anna et al. 2012; Lutchmiah et al. 2014) in the context of seawater/ brackish water desalination, wastewater treatment and liquid food processing. Though FO has widely been studied in the context of desalination, other possible separation areas such as removal of fluoride from groundwater have received very little attention. In the backdrop of the problem of groundwater contamination by fluoride in many countries, this study assumes significance. A FO scheme may be considered as a complete and sustainable one by operating it in continuous mode keeping constant compositions of feed and DSs. This can be ensured only with periodic discharge of concentrated fluoride rejects for subsequent stabilization with addition of fresh feed water to the feeding tank and economical separation of draw solute for recycle. Reported studies have hardly considered the total scheme of continuous steady-state operation, and low flux resulting from internal and external concentration polarization remains a concern (Wang et al. 2010; Zhao and Zou 2011; Sant’Anna et al. 2012; Lutchmiah et al. 2014). Attempt has been made to address the issues of concentration polarization through better designs of membranes and draw solutes. Widjojo et al. (2011) used 2-M NaCl DS and TFC polyamide membranes in PRO mode but found a flux of only 15 L m−2 h−1. Data on concentration polarization and fouling study over a reasonably long operation time are missing. Recovery of draw solute by RO involves high energy consumption that offsets the gain of low energy involvement in
the FO. Thus, the major problem of flux decline after a certain time of operation following concentration polarization or reverse salt diffusion still stands in the way of implementation of FO in water purification. Diffusion of draw solute is often accompanied by back diffusion of feed solute from the active membrane surface to the bulk FS. Thus, reverse salt diffusion progressively increases the osmotic pressure of the FS, thereby seriously limiting separation of the desired solute (Boo et al. 2012). This study through a novel design overcomes the major hurdles of FO in the context of removal of fluoride from contaminated groundwater. This study for the first time proposes a complete scheme consisting of an upstream fluoride separation loop by FO and recovery of draw solute for recycling in a nanofiltration module. The objective of the study is to examine effectiveness of a new FO system design in horizontal alignment and in steady-state continuous mode with synchronization of one upstream FO loop with a downstream nanofiltration (NF) loop.
Materials and methods Chemicals and membranes All the chemical reagents (NaCl and MgSO4) used in the investigation were of reagent grade, and no further purification was done, and these were procured mostly from Merck (Germany). Sodium hydroxide pellets and HCl used for maintaining the pH during analysis were procured from Merck. Thin-film composite polyamide nanofiltration (NF-1 and NF-2) membranes were procured from Sepro Membranes Inc. (USA) whereas asymmetric structure-based cellulose triacetate FO membrane (OsMem™ CTA-NW) was purchased from Hydration Technology Innovations (USA). The membrane surface area of each module was 100 cm2. Major characteristics of the membranes used in the investigation are presented in Table 1. The groundwater used during investigation was collected from some fluoride-affected areas of the Eastern part of India. In such water, fluoride concentration varied from 18 to 20 ppm. Physico-chemical characterization of this groundwater has been presented in Table 2. Experimental set-up Two loops—one for FO and the other for nanofiltration—used in the investigation were connected in a single line in the whole experimental set-up. The FO loop consists of a flatsheet cross flow membrane module with necessary accessories like feed water tank and DS tank made of stainless steel and connected to the FO membrane system via circulating pumps, flow meters and pressure gauges as shown in Fig. 1. Peristaltic pumps are used (Miclins India, VSP-100, India) for
Environ Sci Pollut Res Table 1 Characteristics of three commercial membrane used in investigation
Characteristics
Geometry Materials Thickness pH resistance Max. temperature Max. pressure Pore radius Solute rejection MgSO4 NaCl Water flux in
Unit
Polyamide composite membrane (at 10 bar)
FO membrane
NF-1
NF-2
(at 0.35 bar) OsMem™ CTA-NW
– – cm – K bar nm
Flat sheet Polyamide 0.0165 2–11 323 83 0.53
Flat sheet Polyamide 0.0165 2–11 323 83 0.57
Flat sheet Cellulose triacetate – 3–8 344 0.7 –
% % L m−2 h−1
99.5 90 110
97 50 195
– 99 4
circulating the feed solution (FS) and DS through the FO module while a diaphragm pump is used for circulating diluted DS through the downstream nanofiltration module. The upstream and downstream pressure gauges indicate the transmembrane pressures. The cross flow rate through the system is monitored and controlled using rotameter and by-pass valves. The module is so designed that the flat-sheet membrane remains in a horizontal plane on a perforated stainless steel support while feed water and DS flow tangentially in counter-current directions along the top and bottom surfaces of the membrane, respectively. The bottom chamber of the module is so designed that DS enters the same through one of its inlets at one end and, on crossing tangentially along the membrane surface, leaves the chamber through an outlet maintained at a much higher level than the level of the membrane. This ensures complete immersion of the bottom surface of the membrane in the DS while allowing a continuous sweeping fluid action along its bottom surface. The feed water that flows tangentially along the top surface of the membrane ensures sweeping fluid action on the top layer. This very
Table 2 Characteristics of fluoride-contaminated groundwater before and after treatments
design minimizes concentration polarization, membrane fouling and also back diffusion of the DS, thereby ensuring a high unidirectional volumetric flux that is water transport from feed to DS only. As continuous FO results in dilution of the DS, concentration of the same needs to be done to sustain the FO process. Thus, a downstream nanofiltration membrane module in flatsheet cross flow mode is operated simultaneously for recovery of pure water from the DS while recycling the draw solute to the system. A diaphragm pump (Milton Roy India Pvt. Ltd.) operated at 10–14 bar maintains circulation of the DS through the flat-sheet cross flow nanofiltration module. Operating pressure and cross flow rates are monitored through control valves, rotameter and pressure gauges as illustrated in Fig. 1. Experimental procedure All investigations were carried out in continuous flow mode using flat-sheet cross flow membrane modules with real groundwater from some fluoride-affected areas of
Water parameters
Units
Feed water compositions
Finally treated DS/NaCl (0.3 M)
Finally treated water DS/MgSO4 (0.3 M)
Max. water permissible limit (WHO)
pH Salinity Conductivity TDS TH Chloride Sulphate Iron Fluoride
– mg L−1 μs cm−1 mg L−1 mg L−1 mg L−1 mg L−1 mg L−1 mg L−1
9.8 0.19 670 340 280 85 2.9 0.7 19.8
7.4 2.65 4.8 4.6 5.6 196 0.1 0.03 0.25
7.6 2.32 4.7 4.9 6.1 3.4 174 0.05 1.1
8.5 1000 – 500 600 250 500 1.0 1.5
Environ Sci Pollut Res
Fig. 1 Schematic diagram of pilot-scale forward osmosis–nanofiltration continuous system for removal of fluoride from contaminated groundwater: (1) Make-up feed water tank, (2) water extraction pump, (3) rotameter 1, (4) pH probe 1, (5) concentrated retantate, (6) rotameter (7) pressure gauge 1, (8) pressure gauge 2, (9) pressure gauge 3, (10) rotameter 3, (11) rotameter 4, (12) recycle, (13) DS tank, (14) pH probe
2, (15) draw solution inlet, (16) diluted draw solution, (17) peristaltic pump 1, (18) peristaltic pump 1, (19) diaphragm pump, (20) stabilization tank, (21) stabilized fluoride, (22) stabilizing agent, (23) nanofiltration system, (24) forward osmosis system, (25) feed tank, (26) concentrated fluoride rejects and (27) fluoride-free water
India. Effective filtration surface of each membrane module was 100 cm2. Fluoride-contaminated groundwater was pumped through the forward osmosis (FO) module from where water was extracted in a DS of either NaCl or MgSO4, leaving behind fluoride and other possible contaminants. The diluted DS was then passed on to another nanofiltration membrane module for recovering pure water as the filtrate and recycling the draw solute in the DS loop. Filtration modules were run over long hours in a continuous mode where feed water and DS compositions were maintained at the same level through continuous withdrawal of fluoride rejects and addition of fresh water in the FO loop and recovery of pure water from the downstream nanofiltration module at the same rate of dilution of the DS. Effects of transmembrane pressure, cross flow rates, concentration of the fluoride in the feed water and DSs were investigated during the continuous run spanning over
more than 120 h. Each run of FO experiment was carried out with the active layer of the membrane facing the FS and the support layer facing the DS and that was supposed to be able to afford higher rejection with higher water fluxes due to the reduced internal concentration polarization. The cross flow rates for both draw and the FS were varied between 21 and 42 cm s−1 whereas pressure of the FO module was in the range of 1 to 1.2 bar. The NF module was run at 10–14-bar operating pressure. Both the FS and the DS were maintained at the same temperature. Fluoride rejection and reverse salt flux were assessed by periodic analysis of the samples from the DS tank and feed tank, respectively. NF modules were initially run at different cross flow rates and at different applied pressures to arrive at the best possible cross flow rate and operating pressure. Thus, 13-bar operating pressure and 800 L h−1 cross flow rate were selected for
Environ Sci Pollut Res
continuous running of the system at steady state that ensured high retention of the draw solute while yielding highest possible pure water flux. Concentrated fluoride rejects were subjected to subsequent stabilization and disposal.
with relevant electrodes (Thermo Electron Corporation, USA) where total dissolved solids, conductivity and salinity were determined by InoLab Cond 720 and with electrode TetraCon 325 (WTW, Germany). Iron concentration was also measured by atomic absorption spectrophotometer (AAS100, PerkinElmer, Germany).
Analytics Chemical analysis
Analysis of membrane morphology
Analysis of fluoride has been carried out in Orion 4-Star pH ISE Benchtop Ion Meter of Thermo Electron Corporation, USA. The ion meter was first calibrated using the previously prepared fluoride samples of known strength. Removal efficiency of fluoride was calculated using the bulk fluoride concentration (CF,F) which is present in feed sample and the residual concentration (CP,F) in the permeate, respectively, by the equation below:
Membrane morphology was analyzed by scanning electron microscopy (SEM, Hitachi S-3000 N, Japan) before and after each filtration run (over long periods) to find out the degree of membrane fouling. During SEM analysis, membrane pieces were freeze-fractured in liquid nitrogen and then gold-coated in ion sputter at 15 kV.
R j; F
C P; F ð% Þ ¼ 1 − 100 C F; F
Measurement of water and reverse salt flux ð8Þ
Determination of pH and of concentrations of chloride and sodium was done by Orion 4-Star ISE Benchtop Ion Meter > Measured Water Flux ¼
The weight balance was used for determining the water flux by measuring the weight change of the permeate DS tank during each experimental running, and numerically, water flux can be calculated by the following equation:
Change of weight of the permeate density of the water effective membrane area time interval
Reverse salt flux was assessed by analyzing chloride and sulphate concentrations in the feed tank periodically.
Results and discussion Effects of DS on fluoride rejection and water flux in FO system The osmotic pressure depends on DS (DS) concentration that influences mass transport and overall process performance. A perfect DS should have high osmotic pressure, have non-toxic character and should be easily recoverable in the re-concentration step. Two types of DS were investigated in this study such as NaCl and MgSO4 with molar concentrations ranging from 0.1 to 1 M. Due to high solubility, high osmotic pressure and low cost, NaCl appears to be the most widely used DS in FO process (Cath et al. 2006; Achilli et al. 2010) whereas MgSO4 has been used as a draw solute due to its insignificant retarded forward diffusion resulting in negligible reverse solute flux. Figure 2 indicates how changes in concentration of DS
affect fluoride rejection (primary axis) and water flux (secondary axis) in two different membranes, and this also shows that the DS concentration has a strong positive correlation with volumetric flux. Figure 2a illustrates that at an optimum concentration of 0.3 M NaCl, the NF-2 membrane yields a volumetric water flux of 62 L m −2 h−1 whereas the water flux with MgSO4 as DS is 52 L m−2 h−1. Effects of DS concentration on fluoride rejection for two different membranes are illustrated in Fig. 2a. Fluoride rejection increases linearly with increasing DS concentration, and at a draw solute concentration of 0.3 M NaCl, almost 100 % fluoride rejection is observed by NF-2 and rejection reaches 98 % in case of 0.3 M MgSO4 at 1-bar pressure and 120 L h−1 feed flow rate. Similarly, the DS and the corresponding osmotic pressures influence mass transport and overall process performance significantly. A higher concentration of DS produces a higher osmotic pressure for water transport through the membrane. Higher concentration of DS and hence higher osmotic pressure that forces larger amount of water as water flux through the membrane eventually reduce solute flux across the membrane in an uncoupled transport
Environ Sci Pollut Res
Fig. 2 Fluoride rejection and water flux during forward osmosis under different draw solution concentrations of two different membranes. a Effects of DS concentration on fluoride removal and pure water in NF2 membrane. b Effects of DS concentration on fluoride removal and pure water in OsMem™ CTA-NW membrane. Fluoride rejection (DS/NaC1) (blue triangle), fluoride rejection (DS/ MgSO4) (red asterisk), flux (DS/ NaC1) (red broken line) and flux (DS/MgSO4) (orange broken line). Experimental conditions: total fluoride concentration of 19.8 mg L−1, pH 9.8, draw solution concentration range 0.1–0.5 mol L−1, pressure 1 bar, feed side cross flow rate 120 L h−1, draw solution side cross flow rate 10 L h−1 and temperature 308 K
process. This explains higher fluoride rejection following higher water flux. By virtue of its high solubility, NaCl produces a higher osmotic pressure than MgSO4 for the same concentration in a solution. Low diffusivity of MgSO4 compared to that of NaCl is another reason behind its significantly reduced reverse salt diffusion (Lutchmiah et al. 2014). Figure 2b shows the effects of draw solution concentrations on fluoride rejection and water flux in case of OsMem™ CTA-NW membrane. Almost 100 and 98 % fluoride rejection are observed at 0.3-M concentration for both of the draw solutes (NaCl and MgSO4). But, the water flux observed in case of OsMem™ CTA-NW membrane is much lower (5–7 L m−2 h−1) than that of NF-2 membrane at 0.6-bar pressure and 120 L h−1 feed flow rate. An operating pressure of 0.6 bar was maintained as the maximum
Fig. 3 Fluoride removal efficiency and water flux during forward osmosis under different draw solution concentrations of two different membranes. a Effects of hydraulic transmembrane pressure on fluoride removal efficiency and pure water in NF-2 membrane. Fluoride rejection (DS/NaC1) (blue triangle), fluoride rejection (DS/MgSO4) (red asterisk), flux (DS/NaC1) (blue broken line) and flux (DS/MgSO4) (red broken line). Experimental conditions: fluoride concentration of 19.8 mg L−1, pH 9.8, draw solution concentration range 0.3 mol L−1, pressure range 0–1.2 bar, feed side cross flow rate 120 L h−1, draw solution side cross flow rate 10 L h−1 and temperature 308 K. b Effects of feed cross flow rate on fluoride removal efficiency and pure water in NF-2 membrane. Fluoride rejection (DS/NaC1) (blue diamond), fluoride rejection (DS/ MgSO4) (red big dot), flux (DS/NaC1) (blue green broken line) and flux (DS/MgSO4) (orange broken line). Experimental conditions: total fluoride concentration of 19.8 mg L −1 , pH 9.8, draw solution concentration range 0.3 mol L−1, pressure range 1 bar, feed side cross flow rate range 30–150 L h−1, draw solution side cross flow rate 10 L h−1 and temperature 308 K
operating limit of pressure for CTA-NW membrane is 0.7 bar (HTI, USA). Effects of hydraulic transmembrane pressure and feed cross flow rate on fluoride rejection and water flux in upstream FO system Figure 3 shows the effect of hydraulic transmembrane pressure (TMP) and feed cross flow rate on fluoride rejection and water flux in a screened NF-2 membrane and two different
Environ Sci Pollut Res
draw solutions with fixed concentrations (0.3 M NaCl, MgSO4). Though intrinsically water transport in FO system is by virtue of osmotic pressure difference only, hydraulic TMP often plays a significant role in some particular modules like spiral wound or capillary type where such TMP helps overcome the hydraulic resistances of the flow channels. Considering that in a flat-sheet cross flow membrane module of this study, such hydraulic resistances are far less than in the other conventional modules, at 1-bar pressure, around 100 and 98 % fluoride were rejected (Fig. 3a) by NF-2 membrane at a feed flow rate of 120 L h−1 and draw solution flow rate of 10 L h−1 while using 0.3 M NaCl and 0.3 M MgSO4, respectively, as a draw solution. The anionic form of F ions gets repelled to some higher degree by Donnan exclusion mechanism of the composite polyamide membranes (negative–negative repulsion) (Chakrabortty et al. 2013). In the solution diffusion mechanism of transport through composite polyamide membranes, solute flux and solvent flux are uncoupled and this explains why under enhanced pressure, solvent flux increases with commensurate decrease of solute flux or rather increase of solute rejection (Kumar et al. 2011). Figure 3a (secondary axis) also shows the effect of TMP on volumetric water flux at a feed flow rate of 120 L h−1 and draw solution flow rate of 10 L h−1 in the FO system. A strong positive correlation with TMP and water flux has been shown in Fig. 3 where the highest flux of 62 and 52 L m−2 h−1 was achieved with 0.3 M NaCl and 0.3 M MgSO4 draw solution concentration, respectively. Figure 3b shows an effect of cross flow rate of feed water on fluoride rejection and permeate flux during FO of fluoridecontaminated groundwater. A positive correlation with cross flow rate on water flux as well as fluoride rejection was observed in all cases. At 120 L h−1 of feed flow rate, very close to 100 and 98 % of fluoride rejection were found with 0.3 M NaCl and 0.3 M MgSO4 draw solution concentration, respectively, by keeping fixed pressure 1 bar whereas flux was found 62 and 52 L m−2 h−1 at the same condition. Cross flow rate plays an important role in reducing concentration polarization and increasing turbulence on the membrane surface by its sweeping action and thus reducing fouling as well as enhancing the hydraulic TMP which in turn enhances solvent flux (Kumar et al. 2011). Uncoupled nature of the solute and solvent fluxes results in higher retention of fluoride as flux increases with increasing cross flow rate. Effects of applied pressure and draw solution concentration on RSF Reverse salt diffusion in terms of reverse salt flux (RSF) of NaCl and MgSO4 draw solutions has been shown in Fig. 4 under varying hydraulic TMP and DS concentrations. In Fig. 4a, an inversely proportional relationship between
Fig. 4 Investigation on reverse salt flux (RSF) during forward osmosis of NF-2 membrane. a Effects of hydraulic transmembrane pressure on reverse salt flux. Reverse salt flux (DS/NaC1) (blue triangle) and reverse salt flux (DS/MgSO4) (red asterisk). Experimental conditions: fluoride concentration of 19.8 mg L −1 , pH 9.8, draw solution concentration range 0.3 mol L−1, pressure range 0–2 bar, feed side cross flow rate 120 L h−1, draw solution side cross flow rate 10 L h−1 and temperature 308 K. b Effects of DS concentration on reverse salt flux. Reverse salt flux (DS/NaC1) (blue triangle) and reverse salt flux (DS/ MgSO4) (red asterisk). Experimental conditions: fluoride concentration of 19.8 mg L−1, pH 9.8, draw solution concentration range 0.1– 0.5 mol L−1, feed side cross flow rate 120 L h−1, draw solution side cross flow rate 10 L h−1, pressure 1 bar and temperature 308 K
hydraulic TMP and RSF is observed. In Fig. 4a, a very little change of RSF of MgSO 4 is observed with increasing pressure in NF-2 membrane whereas decline of RSF with increasing pressure is relatively high for NaCl draw solute. Monovalent ion of NaCl possessing a high diffusivity is the cause of such a high RSF whereas bivalent ion of MgSO4 possessing lower diffusivity, due its large hydrated radii, leads to its lower RSF (Cornelissen et al. 2008). The trend of decline in RSF with applied pressure may be attributed to physical changes in the membrane active
Environ Sci Pollut Res
Effects of hydraulic TMP and feed cross flow rate on draw solute recovery and pure water flux in downstream nanofiltration module
Fig. 5 Salt removal and pure water flux during nanofiltration under varying transmembrane pressures and cross flow rates. a Effects of transmembrane pressure on salt removal efficiency and pure water flux in NF-1 membrane. MgSO4 rejection (blue diamond), NaC1 rejection (big red dot) and flux (blue broken line). Experimental conditions: draw solution concentration of 0.3 mol L−1, pressure range 0–14 bar, pH 8, cross flow rate 750 L h−1 and temperature 308 K. b Effects of cross flow rate on salt removal efficiency and pure water flux in NF-1 membrane. MgSO4 rejection (blue diamond), NaC1 rejection (big red dot) and flux (blue broken line). Experimental conditions: draw solution concentration of 0.3 mol L−1, pressure 13 bar, pH 8, cross flow rate range 0–800 L h−1 and temperature 308 K
layer as a function of the applied pressure. Increasing pressure at the membrane surface leads to compression of the interface between the thin active layer and support layer of the membranes, thus reducing the possibility of RSF and salt permeability of the membrane, while slightly affecting water flux (Coday et al. 2013). Variation of RSF with different draw solution concentrations is observed in Fig. 4b. The figure shows that the RSF has a strong positive correlation with draw solution concentration. At high concentration of the draw solution, the possibility of ions permeating through the membrane through RSF increases. Effect of draw solution concentration on RSF is more distinct in case of NaCl than in case of MgSO4 because of differences in ionic characters.
Figure 5 exhibits strong positive correlation of recovery of both the draw solutes (MgSO4 and NaCl) with hydraulic TMP and feed cross flow rate during nanofiltration. Figure 5a (primary axis) shows that at 13-bar TMP, retention of MgSO4 and NaCl by nanofiltration membranes (NF-1) reaches 99 and 98.7 %, respectively, at a feed flow rate of 800 L h−1. Such solute retention behaviour of nanofiltration membrane is attributed to the very solution–diffusion mechanism of mass transport through nanofiltration membrane. In this mechanism, solute and solvent fluxes are uncoupled, and thus, when increased TMP causes an increase in solvent flux, the solute flux decreases resulting in its higher retention or recovery for recycling. The hydrated radii and nature of ions of MgSO4 are different from those of NaCl resulting in a higher retention in case of MgSO4 than in case of NaCl. According to Donnan exclusion principle, during transport of negatively charged ions over the polyamide composite membrane, retention of the draw solutes takes place because of negative–negative charge repulsion (Bowen and Welfoot 2002). Apart from Donnan exclusion mechanism, sieving mechanism also plays a vital role in separation of the species with large hydrated radii. Here again, ions of MgSO4 being much larger than their counterparts of NaCl are retained more resulting in very close to 100 % recovery and recycle of MgSO4 as draw solute. Recovery of NaCl is also very high (98.7 %) by the selected nanofiltration membrane (NF-1) that ultimately produces potable water with some residual chloride which is well below the WHO-prescribed level (250 mg L−1). The flux data as presented in Fig. 5a (secondary axis) shows that salt-free water flux increases with increase of TMP and reaches 62 L m−2 h−1 at 13-bar TMP for a feed flow rate of 800 L h−1. Such flux behaviour of the nanofiltration membranes with pressure during investigation is well established in the literature (Chakrabortty et al. 2014). Figure 5b shows a pronounced effect of feed cross flow rate on draw solute recovery and water flux during nanofiltration, and it shows a strong positive correlation with draw solute recovery and water flux. Figure 5b (primary axis) shows that the recovery efficiency of draw solute increases from 68 to almost 100 % as cross flow rate increases from 400 to 800 L h−1 for the nanofiltration membrane (NF-1) with magnesium sulphate as draw solute at 13-bar TMP. Similar trend is observed in case of NaCl as draw solute. The NF-1 membrane (Fig. 5b, secondary axis) produces a pure water flux of 68 L m−2 h−1 under the operating conditions of 13-bar TMP and 800 L h−1 cross flow rate. Cross flow rate plays an important role in reducing concentration polarization. With increase in the cross flow rate, sweeping action on the active membrane
Environ Sci Pollut Res
the solute and solvent fluxes, also results in higher removal efficiency in this case. Effects of operation time on water flux in upstream FO system and downstream NF system: a fouling study
Fig. 6 Effects of operational time on water flux in FO and NF system. Flux in FO system-NF-2 membrane (big red dot) and flux in NF systemNF-1 membrane (blue triangle). Experimental conditions: fluoride concentration of 19.8 mg L−1, pH 9.8, time range 0–120 h, draw solution concentration of 0.3 mol L−1, feed cross flow rate in FO system 120 L h−1, pressure in FO system 1 bar, feed cross flow rate in NF system 800 L h −1 , pH 8, pressure in NF system 13 bar and temperature 308 K
surface area also increases, thereby reducing concentration polarization. This leads to maximization of the effective membrane surface area available for a given separation. In other flow modes, increase in solvent flux normally accompanies increase in concentration polarization. However, in the very flat-sheet cross flow module such enhancement of flux, following increased cross flow does not really lead to any increase in concentration polarization. Again, reduction of concentration polarization develops convective force that in turn enhances the solvent flux and, due to the uncoupling nature of Fig. 7 Scanning electron microscope (SEM) image before and after investigation. a NF-2: before forward osmosis. b NF-2: after forward osmosis. c NF-1: before nanofiltration. d NF-1: after nanofiltration
Build-up of concentration polarization leading to membrane fouling and hence decline in flux is quite common in membrane filtration. But, such fouling can be largely eliminated through provision of appropriate hydrodynamic conditions associated with a module design. Flat-sheet cross flow membrane module is one such system that provides for sweeping action of fluid on the membrane surface, thus significantly reducing concentration polarization. A drop of 7 % of flux was observed in the module of this study after a long 120 h of operation (Fig. 6). This indicates a significant improvement of the new design over the existing ones that remarkably control concentration polarization and membrane fouling, eventually ensuring long time operation at a steady flux. Reported flux decline is quite high compared to that in this study. For example, Zhao and Zou (2011) and Tang and Yong Ng (2008) found 22–33 % water flux decline within only 15– 20 h of operation while using their module. Choi et al. (2009) studied a plate and frame module and observed around 20 % of flux decline in 100 h of operation which is against only 6 % flux decline in this design over the same duration of filtration. Figure 7 represents the SEM images of the NF-1 and NF-2 membranes before and after the FO and nanofiltration runs, respectively. The figure shows that the membranes do not undergo major morphological changes possibly due to the very hydrodynamic design of the cross flow module. Whatever small fouling that occurs, it could be removed by
Environ Sci Pollut Res
rinsing with 0.1 N NaOH and 0.01 M HNO3. This establishes reversible nature of fouling that occurs on NF-1 membrane.
of India (DST-INSPIRE fellowship program (DST/INSPIRE Fellowship/ 2012/271 and DST-FIST program).
Stabilization of fluoride rejects for safe disposal
References
Stabilization of fluoride rejects was done by adding calcium chloride into the concentrated rejects that formed an insoluble calcium fluoride compound. For safe disposal, fluoride is stabilized in a solid matrix through chemical precipitation using appropriate doses of calcium chloride under response surface (RSM) optimized conditions of the major governing parameters (initial concentration of the fluoride, dosses of calcium chloride, pH and stirring time). Fluoride stabilized in solid matrix is tested for possible leaching out following toxic characteristic leaching procedure (TCLP) at optimum pH condition. Concentrated fluoride rejects were stabilized in a solid matrix of calcium fluoride (CaF2) under a set of optimum conditions determined through response surface optimization technique of Design Expert Software (version 8.1). The maximum fluoride stabilization efficiency of 98 % was achieved under such conditions. Concentration of fluoride after 1 month running of the unit stands at 500 mg L−1, which requires a calcium chloride dose of 10 g L−1, stirring time of 10 min and pH of 5. The weight of sludge was around 4.5 kg (for 150-m3 drinking water production). The leachate concentration of fluoride in the TCLP test was found to be 25 mg L−1 which was well below the maximum leachate concentration (150 mg L−1). This stabilized solid calcium fluoride passed the TCLP test indicating its class of non-hazardous waste to be disposed safely through landfill.
Conclusion The study culminated in development of a new FO– nanofiltration integrated system capable of producing safe potable water from fluoride-contaminated groundwater. This is an approach different from the existing ones in alignment of membrane modules, in integration of the appropriate membranes and in synchronized operation of two stages of filtration. The major hurdle of low flux and reverse salt diffusion could be effectively reduced. The system configuration ensures water transport at high fluxes under significantly reduced concentration polarization and efficient recovery of draw solute at high flux. Both stages involve low energy. Findings of the investigation using fluoride-contaminated real groundwater instead of a model solution are expected to raise scale-up confidence and pave the way for implementing such a total scheme in fluoride-affected regions of the world. Acknowledgments The authors acknowledge with thanks financial support from the Department of Science and Technology, Government
Achilli A, Cath TY, Childress AE (2010) Selection of inorganic based draw solutions for forward osmosis applications. J Membr Sci 364(1–2):233–241 Boo C, Lee S, Elimelech M, Meng Z, Hong S (2012) Colloidal fouling in forward osmosis: role of reverse salt diffusion. J Membr Sci 390– 391:277–284 Bowen WR, Welfoot JS (2002) Modeling the performance of membrane nanofiltration—critical assessment and model development. Chem Eng Sci 57:1121–1137 Cath TY, Childress AE, Elimelech M (2006) Forward osmosis: principles, applications, and recent developments. J Membr Sci 281(1–2): 70–87 Chakrabortty S, Roy M, Pal P (2013) Removal of fluoride from contaminated groundwater by cross flow nanofiltration: transport modelling and economic evaluation. Desalination 313:115–124 Chakrabortty S, Sen M, Pal P (2014) Arsenic removal from contaminated groundwater by membrane-integrated hybrid plant: optimization and control using visual basic platform. Environ Sci Pollut Res 21: 3840–3857 Choi S, Yun A, Hong S, Ahn K (2001) The effect of co-existing ions and surface characteristics of nanomembrane on the removal of nitrate and fluoride. Desalination 133:53–64 Choi Y, Choi J, Oh H, Lee S, Yang DR, Kim JH (2009) Toward a combined system of forward osmosis and reverse osmosis for seawater desalination. Desalination 247:239–246 Coday BD, Heil DM, Xu P, Cath TY (2013) Effects of transmembrane hydraulic pressure on performance of forward osmosis membranes. Environ Sci Technol 47:2386–2393 Cornelissen ER, Harmsen D, de Korte KF, Ruiken CJ, Qin JJ, Oo H, Wessels LP (2008) Membrane fouling and process performance of forward osmosis membranes on activated sludge. J Membr Sci 319(1–2):158–168 Gizaw B (1996) The origin of high bicarbonate and fluoride concentration in waters of the main Ethiopian Rift Valley, East African Rift System. J Afr Earth Sci 22:391–402 Gupta S, Banerjee S, Saha DJK, Mondal N (2006) Fluoride geochemistry of groundwater in Nalhati-1 block of the Birbhum District, West Bengal, India, Research report. Fluoride 39(4): 318–320 Hou D, Wang J, Zhao C, Wang B, Luan Z, Sun X (2010) Fluoride removal from brackish groundwater by direct contact membrane distillation. J Env Sci 22(12):1860–1867 Hu K, Dickson JM (2006) Nanofiltration membrane performance on fluoride removal from water. J Membr Sci 279:528–529 Jiang K, Zhou K, Yang Y, Du H (2013) A pilot-scale study of cryolite precipitation from high fluoride-containing wastewater in a reaction-separation integrated reactor. J Env Sci 25(7):1331–1337 Kartikeyan G, Shunmugasundaraj A (2000) Isopleth mapping and in-situ fluoride dependence on water quality in the Krishnagiri block of Tamil Nadu in South India. Fluoride 33:121–127 Kumar R, Bhakta P, Chakraborty S, Pal P (2011) Separating cyanide from coke wastewater by cross flow nanofiltration. Sep Sci Technol 46: 2119–2127 Lutchmiah K, Verliefde ARD, Roest K, Rietveld LC, Cornelissen ER (2014) Forward osmosis for application in wastewater treatment: a review. Water Res 58:179–197
Environ Sci Pollut Res Patra PK, Mandal B, Chakrabortty S (2010) The hydro geochemistry of fluoride rich groundwater in Birbhum District of West Bengal, India. Ecoscan 4(2&3):209–211 Sant’Anna V, Marczak LDF, Tessaro IC (2012) Membrane concentration of liquid foods by forward osmosis: process and quality view. J Food Eng 111:483–489 Sasaki K, Yoshida M, Ahmmad B, Fukumoto N, Hirajima T (2013) Sorption of fluoride on partially calcined dolomite. Colloids Surf A: Physicochem Eng Asp 435:56–62 Tahaikt M, Habbani RE, AitHaddou A, Achary I, Amor Z, Taky M, Alami A, Boughriba A, Hafsi M, Elmidaoui A (2007) Fluoride removal from groundwater by nanofiltration. Desalination 212:46–53 Tang W, Yong Ng H (2008) Concentration of brine by forward osmosis: performance and influence of membrane structure. Desalination 224:143–153
Wang KY, Ong RC, Chung TS (2010) Double–skinned forward osmosis membranes for reducing internal concentration polarization within the porous sublayer. Indus Eng Chem Res 49: 4824–4831 Widjojo N, Chung TS, Weber M, Maletzko C, Warzelhan V (2011) The role of sulphonated polymer and macrovoid-free structure in the support layer for thin-film composite (TFC) forward osmosis (FO) membranes. J Membr Sci 383:214–223 World Health Organization (WHO), Guidelines for drinking water quality, 1, World Health Organization, Geneva, 1993, pp. 45– 46 Zhao S, Zou L (2011) Effects of working temperature on separation performance, membrane scaling and cleaning in forward osmosis desalination. Desalination 278:157–164
