w a t e r r e s e a r c h 6 9 ( 2 0 1 5 ) 1 2 0 e1 3 0

Available online at www.sciencedirect.com

ScienceDirect journal homepage: www.elsevier.com/locate/watres

Seawater-driven forward osmosis for enriching nitrogen and phosphorous in treated municipal wastewater: Effect of membrane properties and feed solution chemistry Wenchao Xue a, Tomohiro Tobino b, Fumiyuki Nakajima a, Kazuo Yamamoto b,* a Department of Urban Engineering, Graduate School of Engineering, The University of Tokyo, 7-3-1 Hongo, Bunkyo-ku, Tokyo 113-8656, Japan b Environmental Science Center, The University of Tokyo, 7-3-1 Hongo, Bunkyo-ku, Tokyo 113-0033, Japan

article info

abstract

Article history:

Seawater-driven forward osmosis (FO) is considered to be a novel strategy to concentrate

Received 6 July 2014

nutrients in treated municipal wastewater for further recovery as well as simultaneous

Received in revised form

discharge of highly purified wastewater into the sea with low cost. As a preliminary test,

19 October 2014

the performance of FO membranes in concentrating nutrients was investigated by both

Accepted 6 November 2014

batch experiments and model simulation approaches. With synthetic seawater as the draw

Available online 15 November 2014

solution, the dissolved organic carbon, phosphate, and ammonia in the effluent from a membrane bioreactor (MBR) treating municipal wastewater were 2.3-fold, 2.3-fold, and 2.1-

Keywords:

fold, respectively, concentrated by the FO process with approximately 57% of water

Forward osmosis

reduction. Most of the dissolved components, including trace metals in the MBR effluent,

Seawater driven

were highly retained (>80%) in the feed side, indicating high water quality of permeate to

Nutrient enrichment

be discharged. The effect of membrane properties on the nutrient enrichment performance

Membrane properties

was investigated by comparing three types of FO membranes. Interestingly, a polyamide

Feed solution pH

membrane possessing a high negative charge demonstrated a poor capability of retaining ammonia, which was hypothesized because of an ion exchange-like mechanism across the membrane prompted by the high ionic concentration of the draw solution. A feed solution pH of 7 was demonstrated to be an optimum condition for improving the overall retention of nutrients, especially for ammonia because of the pH-dependent speciation of ammonia/ ammonium forms. The modeling results showed that higher than 10-fold concentrations of ammonia and phosphate are achievable by seawater-driven FO with a draw solution to feed solution volume ratio of 2:1. The enriched municipal wastewater contains nitrogen and phosphorous concentrations comparable with typical animal wastewater and anaerobic digestion effluent, which are used for direct nutrient recovery. © 2014 Elsevier Ltd. All rights reserved.

* Corresponding author. Tel.: þ81 8036998063. E-mail address: [email protected] (K. Yamamoto). http://dx.doi.org/10.1016/j.watres.2014.11.007 0043-1354/© 2014 Elsevier Ltd. All rights reserved.

w a t e r r e s e a r c h 6 9 ( 2 0 1 5 ) 1 2 0 e1 3 0

1.

Introduction

Discharges of nitrogen and phosphorous from municipal wastewater into the environment are undesirable because these nutrients are one of the primary causes of eutrophication in the natural water environment. Therefore, strict rules have been established in many countries to reduce the release of nutrient chemicals into surface water. Alternatively, both nitrogen and phosphorous significantly contribute to agriculture and industry (Liu et al., 2011). Thus, the concept of nutrient recycling is used as a sustainable solution for developing modern wastewater treatment, and various technologies have been developed for recovering nutrients from wastewater, such as the magnesium ammonium phosphate crystallization process (Doyle and Parsons, 2002; Lind et al., 2000; Song et al., 2007). However, the applications of current technologies are usually limited to specific types of wastewater (such as animal wastewater or separated human urine) that contains a remarkably high concentration of nitrogen and phosphorous. Municipal wastewater that contains a sizable proportion of nutrient chemicals (Lind et al., 2001), the removal of which consumes considerable energy in treatment facilities, can be used as a resource for nutrient recycling. For effectively recovering nutrients from municipal wastewater, a means of obtaining suitable concentration level is required. Technologies using membranes have been widely applied as an approach to separate and/or concentrate nutrients from domestic wastewater (Bilstad, 1995; Lee and Lueptow, 2001). However, pressurized membrane processes, such as reverse osmosis (RO) and nanofiltration (NF), increase the cost of concentrating nutrients in wastewater due to the high hydraulic pressure required by these processes. Forward osmosis (FO) is a membrane process in which water molecules migrate by natural osmosis through a semipermeable membrane from a feed solution with lower osmotic pressure to a draw solution with higher osmotic pressure (Cath et al., 2006). FO provides several advantages over pressure-driven membrane technologies such as i) the operational cost and equipment investment for FO is reduced because it operates under low or no hydraulic pressure conditions, ii) FO achieves high water recovery due to a reduced impact of scaling and high rejection of a wide range of contaminants, and iii) FO is expected to have a lower membrane fouling propensity than pressure-driven membrane processes (Cornelissen et al., 2008; Mi and Elimelech, 2010; Lee et al., 2010). Consequently, laboratory- and pilot-scale studies of FO have been conducted for many applications. For example, seawater and brackish water desalination (McCutcheon et al., 2005, 2006; Phuntsho et al., 2012), commercial materials production (Jiao et al., 2004; Cath et al., 2005a, 2005b; Nayak and Rastogi, 2010), and wastewater treatment (Holloway et al., 2007; Achilli et al., 2009; Chen et al., 2014) by FO have attracted increasing attention over the past decade. From the viewpoint of retaining nutrients, the high potential of the FO process has been reported by Holloway et al. (2007) in a study of concentrating anaerobic digester centrate, and Zhang et al. (2014) in a study focusing on retaining nitrogen and phosphorous in urban source-separated urine using FO. However, limited information is available on the capability of FO for enriching

121

nitrogen and phosphorous for nutrient recovery from dilute wastewater, such as treated municipal wastewater. Meanwhile, studies on factors affecting FO performance on concentrating nutrients are desired to improve the efficiency of FO in commercial applications. In this study, a seawater-driven FO is proposed and applied as a low energy and low chemical expense process to enrich the concentration levels of nitrogen and phosphorous in treated municipal wastewater, to an extent suitable for direct recovery. The FO process considered in this study uses seawater as the draw solution. Seawater contains unexploited osmotic energy, and thus, it is a low-cost source of draw solution. Furthermore, the diluted seawater generated through the process is possible to be discharged into the sea as a mixture of highly treated wastewater and seawater, and hence, there is no need for regenerating the draw solution. This study aims to conduct a preliminary evaluation of the performance of the seawater-driven FO process for concentrating nitrogen and phosphorous in treated municipal wastewater by both experimental and modeling approaches. For improving the performance, the effects of membrane properties were investigated, including membrane charge characteristics and orientation as well as feed solution chemistry such as pH.

2.

Material and methods

2.1.

FO membranes

Three flat-sheet membranes (i.e., TFC, CTA-1, and CTA-2), specifically developed for FO by Hydration Technologies, Inc. (Albany, OR, U.S.), were used. The TFC membrane is prepared using polyamide on polysulfone with embedded support (Yip et al., 2010). Both the CTA-1 and CTA-2 membranes have an asymmetric structure with cellulose triacetate as their active layers. CTA-1 was supported by an embedded polyester screen mesh, and CTA-2 utilized a nonwoven support (McCutcheon et al., 2005; Yip et al., 2010). All membranes were tested in both orientations, i.e., with the active layer facing the feed solution (AL-FS) and the active layer facing the draw solution (AL-DS), to study the influence of membrane orientation and structure on FO performance. The membrane performance parameters, including the intrinsic water and solute permeabilities, A and B, respectively, and the resistance to solute diffusion within the membrane porous support layer, K, were measured in a RO filtration system. Each tested membrane coupon with the effective membrane area of 38.5 cm2 was installed into a dead-end filtration unit. Nitrogen (N2) gas was employed to provide the hydraulic pressure for RO filtration. The stirring speed was maintained at 300 rpm in the membrane unit during all experiments. A values were determined from the regression slope of water flux against hydraulic pressure over a range of applied pressures (i.e. 0.4e1.0 MPa). The values of B and K of CTA-1 membrane were determined following the description of Loeb et al. (1997) and Lee et al. (1981), and used for modeling simulation of FO performance. As the primary feed solutes that we concerned in this study, the permeability values, B, of the representative model nutrient solutes including ammonium chloride (NH4Cl),

122

w a t e r r e s e a r c h 6 9 ( 2 0 1 5 ) 1 2 0 e1 3 0

Fig. 1 e Schematic diagram of the laboratory-scale FO setup. sodium nitrite (NaNO2), sodium nitrate (NaNO3) and sodium dihydrogen phosphate (NaH2PO4) in CTA-1 membrane were determined based on feed and permeate concentration measurements with a feed solution initially containing 5 mmol L1 of each model solute. The zeta potential of the FO membrane was measured using an electrophoretic light scattering zeta potential analyzer (ELS-8000, Otsuka Electronics, Japan).

2.2.

Draw and feed solutions

As the draw solution, standard synthetic seawater (U.S.EPA, 1994) was prepared following the composition listed in Table S1. This synthetic seawater had an electric conductivity of 42.8 ± 0.3 mS cm1 and a pH of 8.12 ± 0.08 at ~25  C. Four primary solutes, NaCl, MgCl2, CaCl2, and Na2SO4, contributed more than 97% of the total osmotic pressure of the draw solution. In the experiments conducted to study the effect of feed solution pH on nutrients retention, 3.5 wt% NaCl solution was used as the draw solution to simplify the experiment (Ge et al., 2012). Two types of feed solutions were adopted: a synthetic feed solution and real effluent samples from a membrane bioreactor (MBR) treating municipal wastewater. The synthetic feed solution was prepared by dissolving four typical nutrients (NH4Cl, NaNO2, NaNO3 and NaH2PO4) into MilliQ water at a concentration of 10 mg L1 each for studying the retention performance of the nutrients by the three FO membranes. This feed solution had an electric conductivity of 0.29 ± 0.1 mS cm1 and a pH of 5.61 ± 0.05 at ~25  C. The MBR effluent samples were collected from a pilot inclined tube MBR (itMBR) set in a wastewater treatment plant in Japan. This MBR was specially designed for carbonaceous removal and biomass production from wastewater (Chen et al., 2012). A short hydraulic retention time (1e2 h) was adopted to suppress nitrification. Therefore, a high concentration of nutrients, including ammonia and phosphate, remained in the effluent. Other operation parameters and general performance of itMBR during the sampling period are presented in Table S2. The pH of the MBR effluent was 7.92 ± 0.16, and the electric conductivity was 1.57 ± 0.14 mS cm1 at ~25  C. The

effluent was used as a feed solution for FO after being filtered with a 0.45 mm membrane. In addition, MilliQ water was used as the feed solution in several batch experiments.

2.3.

Laboratory-scale FO setup and operation

A schematic diagram of the laboratory-scale FO setup is shown in Fig. 1. A cross flow, flat sheet membrane filtration unit with an effective filtration area of 60 cm2 (C10-T, Nitto Denko Co., Japan) was applied. This membrane module contained channels on both sides of the membrane for feed and draw solutions, respectively. The effective dimensions of each channel were measured as 167 mm (length), 36 mm (width), and 1.4 mm (height). Mesh spacers were installed at both channels to support the membrane as well as to increase turbulence in an effort to decrease the effect of the external concentration polarization (CP) during filtration. Countercurrent circulation of the draw and feed solutions was applied on each side of the membrane via peristaltic pumps (Masterflex, Cole-Parmer, USA). The cross-flow velocity was set at 8.3 cm s1 in all the experiments. The temperatures of the feed and draw solutions were controlled in both reservoirs at 25 ± 2  C by water bath. Water flux was determined using an electronic balance (GX-4000, AND Co., Japan) by measuring the rate at which the weight of the feed solution decreased. In experiments conducted with the synthetic feed solution, the draw solution tank contained 1 L of synthetic seawater, and the initial volume of the feed solution was 500 mL. To mitigate the impact of dilution of the draw solution, the duration of each filtration test was set to 2.5 h. Samples from the feed and draw solutions were taken at the beginning and at the end of the experiments, and the actual retention of nutrients, R, was calculated using Eq. (1), following the description of Jin et al. (2011). R ¼1

Js JW cs;f

(1)

where JS (mol$m2 s1) and JW (m s1) are the flux of the feed solute and water respectively, and cS,f (mol m3) is the average concentration of the feed solute.

123

w a t e r r e s e a r c h 6 9 ( 2 0 1 5 ) 1 2 0 e1 3 0

In experiments conducted with the MBR effluent as the feed solution, the draw solution tank contained 4 L of synthetic seawater, and the feed solution tank contained 2 L of filtered MBR effluent. It is noteworthy that a low volume ratio of draw solution to feed solution (i.e. 2:1) was adopted in this study to demonstrate the feasibility of seawater-driven FO process which was operated under unfavorable condition. To initialize the feed solution at a neutral pH range (6.6e7.0) and compare the retention performance with that using original MBR effluent as feed solution, pH adjustment was conducted prior to several bench testes using 0.1 M HCl/NaOH solutions. Solutions from the feed tank were sampled at specified time intervals for chemical measurements. At the end of the experiments, the water reduction ratio was calculated by dividing the overall volume of the permeate by the initial volume of the feed solution. Similarly, feed solutes recovery was determined by dividing the total amount remaining in the feed solution by the initial amount. It is worth noting that in FO filtration tests aiming at accumulating nutrients in MBR effluent, feed solutes recovery, which is different from the solute retentions described in Eq. (1), was adopted to provide more direct understanding on the performance of FO on enriching nitrogen and phosphorous in treated municipal wastewater. To ensure that the volume of the remaining feed solution was sufficient to fulfill the feed side pipe system, each test was conducted up to a point resulting in approximately 50% water reduction. Because the feed and draw solutions were circulated in nearly closed loops and the duration of each batch test was short, the effect of evaporation was neglected in calculating the water flux of FO.

2.4.

Analytical methods

Dissolved organic carbon (DOC) was determined using a total organic carbon (TOC) analyzer (TOC-V, Shimadzu, Japan). The concentration of ammonia was measured according to the salicylate method using a spectrophotometer (DR/2800, Hach). Anion concentrations in the feed solution were measured using ion chromatography (IC; 861 Contact IC, Metrohm Ion analysis). Concentrations of metals were determined using an inductively coupled plasma atomic emission spectrometer (ICP-AES; Optima 3000, Perkin Elmer) and an inductively coupled plasma mass spectrometer (ICPMS; 7500cx, Agilent).

2.5.

(3)

In the above equations, A is the membrane water permeability, Bd is the draw solute permeability, Kd is the resistance to draw solute diffusion within the membrane porous support layer, pHi and pLow are the osmotic pressures in the bulk of the draw and feed solutions, respectively, which were determined with considering the effect of reverse diffusion and forward permeation of draw and feed solutes. The reverse draw solute diffusion of JDS has been demonstrated to be determined by A and Bd, together with the osmotic proportionality coefficient of the draw solute, Hd, (Phillip et al., 2010; Tang et al., 2010), as expressed in the following equation. JDS Bd ¼ JW AHd

(4)

Conversely, the forward solute flux from the feed to the draw solution is driven by the concentration gradient across the active layer of the membrane, and is given with consideration for the influence of the internal CP as follows (Jin et al., 2011): JS ¼

Bf  Cf 1 þ Bf JW

JS ¼

   Bf exp JW Kf    Cf 1 þ Bf exp JW Kf JW

ðAL  FSÞ

(5)

ðAL  DSÞ

(6)

where Bf is the feed solute permeability of the membrane, Kf is the resistance to feed solute diffusion within the membrane porous support layer, and cf is the solute concentration in the bulk of the feed solution. By combining the above three flux models, iterative method was employed to predict the water flux and nutrients concentration in feed solution with MATLAB 7.1 as the executive tool. To simplify the calculation, Kd of NaCl was used instead of that of seawater. The root-mean-square deviation (RMSD) was adopted to evaluate the goodness of fit of the predicted values to the observed results, as shown in the following equation: sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi Pn 2 i¼1 ðAi  Fi Þ M¼ n1

(7)

where Ai is the experimental value and Fi the forecast value.

Flux models for the FO process

A series of flux models was applied to fit and predict the water flux and nutrient concentration in the feed solution. Principally, three types of mass transport were considered in the FO process: i) water permeation from feed to draw solution, ii) reverse solute diffusion from draw to feed solution, and iii) forward solute permeation from feed to draw solution. The water flux in FO, JW, has been described by considering the influence of internal CP with respect to each membrane orientation (Loeb et al., 1997; Lee et al., 1981).
Jw ¼


 1 Bd þ ApHi  Jw ln ðAL  DSÞ Kd Bd þ ApLow


Jw ¼


 1 Bd þ ApHi ln ðAL  FSÞ Kd Bd þ Jw þ ApLow

(2)

3.

Results and discussion

3.1.

Membrane characterization

TFC membrane had a value for A of 4.08  1012 m s1 Pa1, which was comparable to the values measured for the commercial TFCeRO membrane (Yip et al., 2010). CTA-1 membrane had a lower value for A than the TFC membrane (2.00  1012 m s1 Pa1), followed by CTA-2 membrane (1.28  1012 m s1 Pa1). Both the CTA membranes had A values in the same order of magnitude with those reported in literature (Tang et al., 2010; Jin et al., 2011; Zhang et al., 2014).

124

w a t e r r e s e a r c h 6 9 ( 2 0 1 5 ) 1 2 0 e1 3 0

The zeta potentials of the selected FO membranes were initially analyzed over a pH of 3e10, as shown in Fig. 2. For all membranes, the surface charge was negative in a pH range higher than 4. The zeta potential of the CTA-2 membrane was less sensitive to pH variation than the other two membranes, and demonstrated a lowest charge of 22 mV at a highest tested pH of 10. The CTA-1 membrane had a similar zeta potential distribution, as reported by Linares et al. (2011). The TFC membrane possessed a much higher negative charge than the CTA membranes, especially when the pH was higher than 6 (53 mV). This value was also lower than those of commercial RO and NF membranes reported in previous publications (Childress and Elimelech, 1996; Richards et al., 2010).

3.2. Experimental performance of FO for enriching nitrogen and phosphorous To evaluate the performance of the CTA-1 FO membrane for retaining nitrogen and phosphorous as well as other contaminants in the treated municipal wastewater, batch filtration tests were conducted using the effluent of MBR as the feed solution and synthetic seawater as the draw solution. The CTA-1 membrane was used because it is widely adopted in present FO applications (Cath et al., 2005a, 2005b; McCutcheon et al., 2006; Phuntsho et al., 2012; Chen et al., 2014). During FO filtration, an initial feed solution volume of 2000 mL was reduced to approximately 900 mL (water reduction of 57%). The solute recoveries of the nutrients and other pollutants in the feed solution are listed in Table 1. Ninety-four percent of the DOC remained in the feed solution, though a relatively large fluctuation was observed among the different batch tests. Ammonia was not effectively retained. One possible reason is that the relatively high pH of the MBR effluent (i.e., pH 8) enlarged the equilibrium ratio of the NH3/NHþ 4 pair in the feed solution, and hence impacted the retention of ammonia by the CTA-1 membrane (Qin et al., 2003), as will be discussed later. A low recovery of nitrite and nitrate (76% and 58%, respectively) was also obtained. Because nitrite and nitrate

Fig. 2 e Comparison of zeta potentials of three FO membranes over a pH of 3e10. Experiments were conducted with a background electrolyte of 10 mmol L¡1 NaCl.

Table 1 e Solute recovery of contaminants in itMBR effluent by CTA-1 membrane at a water recovery of 57%. Experiments were conducted without pH control of the feed solution. Concentration in itMBR effluent ± S.D. (mg L1 for major components, mh L-1 for trace metals) Major components TOC Ammonia-N Nitrite-N Nitrate-N Phosphate-P Sulfate-S Magnesium Calcium Barium Manganese Trace metals Lithium Boron Aluminum Iron Chromium Cobalt Gallium Strontium Cadmium

Solute recovery ± S.D. (%)

13.6 15.8 1.9 2.6 3.4 23.9 15.4 42.7 0.2 0.08

± 8.3 ± 7.2 ± 2.0 ± 4.2 ± 3.3 ± 6.7 ± 4.6 ± 9.8 ± 0.1 ± 0.03

94.0 ± 66.7 ± 75.6 ± 57.5 ± 92.1 ± 99.7 ± 91.6 ± 94.6 ± 92.3 ± 83.1 ±

1.5 26.3 2.7 5.6 0.21 0.05 0.3 18.3 0.6

± 1.1 ± 10.9 ± 1.7 ± 2.2 ± 0.04 ± 0.02 ± 0.1 ± 4.8 ± 0.9

87.2 ± 4.2 27.5 ± 7.5 88.9 ± 39.9 33.0 ± 8.7 57.8 ± 10.3 88.1 ± 19.1 82.8 ± 3.5 93.3 ± 12.8 N.D.

19.2 5.8 9.9 6.5 3.8 1.0 1.9 6.4 3.3 11.6

N.D.: not detected, S.D.: standard deviation.

were not the dominant components of the nitrogen species in the feed solution, their low retention did not impact the performance of the total nitrogen concentration by the FO process. As anticipated, phosphate was effectively retained by the CTA-1 membrane with a recovery ratio of 92%. This performance of FO on enriching nitrogen and phosphorous in treated municipal wastewater was consistent with that observed by Holloway et al. (2007) focusing on concentrating nutrients in digester centrate and that reported by Zhang et al. (2014) aiming at enriching nutrients in urine by FO. In general, for semipermeable membranes including CTA membranes, the separation performance is primarily determined by steric hindrance and electrostatic repulsion between the membrane and solute molecules. Hence, the high molecular weight and/ or negative charge of the DOC and phosphate resulted in retentions higher than that of ammonia, which were positively charged and/or existed as noncharged molecules. Therefore, enhancing the retention of nitrogen, especially ammonia, may become a critical issue for improving the performance of this FO process aimed at nutrients recovery. In addition to the nutrient chemicals, the retention of other pollutants such as trace metals in the effluent of MBR by the CTA-1 membrane were investigated (Table 1). Recoveries higher than 80% were observed for most analyzed compounds, except for boron, iron, and chromium, which demonstrated a solute recovery of 28%, 33%, and 58%, respectively. A low rejection of boron ranging from 29% to 62% has also been obtained using the same FO membrane by Jin et al. (2011). This was attributed to the high permeability of boron, which was 1e2 orders of magnitude higher than that of NaCl in the CTA membrane (Jin et al., 2011).

w a t e r r e s e a r c h 6 9 ( 2 0 1 5 ) 1 2 0 e1 3 0

125

Fig. 3 e Enhancement of nutrient concentration performances by feed solution pH control. An effort was made to enhance the retention of nitrogen by the CTA-1 FO membrane by adjusting the pH of the MBR effluent. Two effluent samples (S-1 and S-2; for composition, see Table S3) taken on different days were used for batch experiments to confirm the repeatability. The pH was adjusted from the original values of 8.0 and 7.7 to 7.0 and 6.6 for S-1 and S-2, respectively. The results given in Fig. 3 illustrate the variation of nutrient concentrations with and without pH adjustment to the feed solution. Because S-1 contained relatively low nitrite and nitrate concentrations and the results presented high fluctuation, only results for S-2 are shown for nitrite and nitrate. For the same reason, only the results from the test of S1 are shown for phosphate in the figure. A slightly lower DOC concentration in feed solution was obtained when the pH was adjusted to approximately 7. This is possibly because the decreased pH lessened the negative charge on both the CTA-1 membrane and DOC molecules, resulting in a weakened electrostatic repulsion between them. In contrast, the concentration of ammonia was noticeably enhanced by decreasing the pH of the feed solution from 8 to 6.6e7.0. As the pH of the feed solution reduced to approximately 7.0, ammonia was 2.1-fold concentrated compared with the 1.7-fold concentration at a pH of 8.0 for S-1 at a water reduction of 52%. Similar results were obtained in the tests for S-2. The different concentration factors obtained between S-1 and S-2 are probably due to the different compositions of the MBR effluent. Conversely, pH adjustment did not significantly influence the concentration performances of nitrite, nitrate, and phosphate. A pH between 7 and 8 showed comparable retentions of nitrite, nitrate, and phosphate by the CTA-1 membrane.

three commercial FO membranes (i.e., CTA-1, CTA-2, and TFC). FO water fluxes were initially measured using MilliQ water as the feed solution and synthetic seawater as the draw solution, as shown in Fig. 4. The AL-DS orientation usually yielded a higher flux than the AL-FS orientation, which can be attributed to the more significant dilution of the draw solution inside the membrane support layer due to water convection (Tang et al., 2010; Gray et al., 2006). The TFC membrane exhibited a highest water flux of 16.6 L m2 h1 in the AL-DS orientation, more than 1.5 times those achieved by the CTA membranes (10.5 L m2 h1 for the CTA-1 membrane and 9.2 L m2 h1 for the CTA-2 membrane) under the same

3.3. Factors affecting the performance of FO for concentrating nitrogen and phosphorous

Fig. 4 e Comparison of water flux of the three FO membranes for different membrane orientations. Experimental conditions for FO filtration were as follows: synthetic seawater as the draw solution, MilliQ as the feed solution, a feed and draw solution temperature of 25  C, and a cross-flow velocity of 8.3 cm s¡1.

3.3.1.

Membrane properties and orientations

The effect of membrane materials and properties on the enrichment of nutrients by FO was investigated by comparing

126

w a t e r r e s e a r c h 6 9 ( 2 0 1 5 ) 1 2 0 e1 3 0

membrane orientation and experimental conditions. Similarly, a higher water flux of 8.2 L m2 h1 was achieved by the TFC membrane relative to those of the CTA membranes (7.4 L m2 h1 for the CTA-1 membrane and 5.3 L m2 h1 for the CTA-2 membrane) with the AL-FS orientation, though the disparity was not as remarkable as when using the AL-DS orientation. In addition, the CTA-2 membrane with a nonwoven support layer showed poorer performance in water permeability compared with that of the CTA-1 membrane supported by a polyester screen mesh due to its higher internal CP (Cath et al., 2006; Ng et al., 2006). Nutrient retentions were then measured by batch FO filtrations using a synthetic feed solution and a synthetic seawater draw solution, as shown in Fig. 5. Retentions of ammonia by CTA membranes were approximately 90%, apparently higher than those by the TFC membrane in both orientations. In the case of the AL-FS orientation of the TFC membrane, even negative retention was obtained for ammonia, indicating that it was unable to concentrate ammonia on the feed side. Two hypotheses might explain these results: i) the greater ammonium permeability of the TFC membrane (4.60  107 m s1) contributed to a higher passage of ammonium compared with the CTA membranes that have ammonium permeabilities an order of magnitude lower (9.49  108 m s1 for CTA-1 and 8.71  108 m s1 for CTA-2), and ii) the high negative zeta potential of the TFC membrane at a similar level to that of a cation exchange membrane (Xie et al., 2011) would have prompted the ammonia transfer from the feed side to the draw side via a cation exchange-like mechanism across the membrane. In the FO process, the latter phenomenon would be further enhanced by a high ion supplement capacity of the draw

solution. In general, a desired FO membrane should have the characteristics of a high density active layer to develop solute rejection, and a small membrane thickness with a maximum support layer porosity to avoid the impact of internal CP and to achieve a high flux (Cath et al., 2006). Hence, significant efforts to improve the FO membrane have focused on tailoring the membrane structure. The phenomenon hypothesized in this study regarding the charge properties may provide another criterion for FO membrane selection in future applications of FO. Jin et al. (2011) developed a model for describing the retention of solutes in FO, attributable to the influence of internal CP in CTA membranes, which predicted that under the same experimental conditions, the AL-FS orientation always exhibited better retention than the AL-DS orientation. However, TFC presented a higher retention of ammonia with the AL-DS orientation than with the AL-FS orientation. As the impact of internal CP on ammonium was not as severe as other nutrients due to its lower retention by TFC membrane, meanwhile, the higher water flux in AL-DS orientation provided greater dilution effect in draw solution, the retention of ammonium is possibly improved with AL-DS orientation. On the other hand, this phenomenon may as well be associated with the high negative charge on the TFC membrane, which suggests that as the membrane charge accumulates at the active surface, the attraction of cations in the feed solution is further strengthened when the active layer is facing the feed solution. Consequently, the exchange of cations is accelerated; thus, retention is reduced. In contrast with ammonia, nitrite and nitrate were retained to a greater extent by the TFC membrane than by the CTA membranes, which is attributable to the stronger electrostatic repulsion due to the membrane negative charge. The low retention of nitrite and nitrate by the CTA membranes is consistent with observations obtained for a NF membrane (Paugam et al., 2004), possibly because the interactions between the polar structure of the solutes and the membrane enhance the partitioning of nitrate and nitrite molecules into the membrane (Hancock et al., 2011). All the membranes considered demonstrated a higher than 90% phosphate retention, probably due to its larger hydrated radius (van Voorthuizen et al., 2005) and higher negative charge compared with the other nutrient solutes. Note that the nutrient retentions observed here are lower than those reported by Holloway et al. (2007) because the calculation of retention in this study excluded the dilution effect of the draw solution.

3.3.2.

Fig. 5 e Retention of typical nutrients by the three FO membranes. Experimental conditions for FO filtration were as follows: synthetic seawater as the draw solution, MilliQ solution containing 10 mg L¡1 each of NH4Cl, NaNO2, NaNO3, and NaH2PO3 as the feed solution (pH ¼ 5.8), feed and draw solution temperature of 25  C, and a cross-flow velocity of 8.3 cm s¡1.

Feed solution pH

As mentioned above, the feed solution pH may be an essential factor affecting nutrient retentions, especially for ammonia retention by the FO membrane. To this end, the retentions of ammonia, nitrite, nitrate, and phosphate were investigated under varied pH conditions using CTA-1 membrane with the synthetic feed solution and the NaCl draw solution at a water flux of 7.5 L m2 h1, as shown in Fig. 6. As expected, ammonia retention was sensitive to pH, where retention was relatively high at a pH of 5 (94%) and gradually reduced with increasing pH. A substantial decline in ammonia retention occurred at a pH of 8. The following reversible reaction occurs between the conjugate acidebase pairs of ammonia (Qin et al., 2003):

w a t e r r e s e a r c h 6 9 ( 2 0 1 5 ) 1 2 0 e1 3 0

Fig. 6 e pH effects on nitrogen and phosphorous retention. FO water flux was approximately 7.5 L m¡2 h¡1.

þ NHþ 4 #H þ NH3 ðaqÞ ðpKa ¼ 9:3Þ

Ammonia predominantly exists in an aqueous solution as NHþ 4 below a pH of 9.3 and as uncharged NH3 (aq) above this value. Because of the Donnan exclusion by a charged membrane, NH3 (aq) is much easier to transfer through the membrane than NHþ 4 . In our results, ammonia retention was closely correspondent with the variation of pH-dependent ammonia speciation, indicating that the relatively low retention of ammonia was mainly the result of ammonia speciation. In contrast to ammonia, the retentions of nitrite, nitrate, and phosphate were enhanced by increasing pH from 5 to 8. This can be attributed to the increase of negative charge on the membrane surface (Fig. 2), which strengthened the electrostatic repulsion between the negatively charged ions and the membrane. The speciation transformation from H2PO 4 to HPO2 4 (pKa ¼ 7.2) is likely to further improve the retention of phosphate. Conversely, a reduction of nitrite and nitrate retention was observed at a pH of 9, which was possibly because the enhancement of phosphate retention and/or reduction of ammonia retention at a pH of 9 intensified the passage of nitrite and nitrate to fulfill the condition of electroneutrality in the feed solution. With all the above taken together, a neutral pH (approximately 7) is considered to be the optimal trade-off for improving the overall retention of nutrients.

3.4. Modeling approach evaluation of the feasibility of the seawater-driven FO process for concentrating nutrients To further evaluate the possibility of using the seawaterdriven FO to achieve nutrient concentration in treated municipal wastewater, a modeling approach was applied to simulate the behavior of FO together with experimental validation. Regarding the former discussion of membrane properties and feed solution chemistry, the simulation was conducted using the AL-FS orientation of a low charged FO

127

membrane (CTA-1) under a feed solution pH of 7. Parameters used for simulation are summarized in Table 2. As discussed above, similar water permeability coefficient was observed with that reported in previous studies. The B values for ammonium and phosphate obtained in this study was approximately one order of magnitude lower than those obtained by Zhang et al. (2014) in the study with urine as feed solution. This variation was possibly due to the impact of multi-solutes in feed solution used in their experiment. On the other hand, the calculated resistance to solute diffusion within membrane porous support layer, Kd was in the same order of magnitude with that reported by Zhang et al. (2014). In general, the membrane performance properties showed good consistency within these two studies. The modeling results of water reduction and nutrient concentration in the feed solution in batch filtration tests are illustrated in Fig. 7. As the batch filtration continued, the increasing rate of permeate volume gradually reduced due to the impacts of the concentration of the feed solution and the dilution of the draw solution. A good agreement between the experimental results and the predicted values were revealed by the RMDS values (0.04e0.11) for ammonia, nitrite, and phosphate. However, the concentration performance of nitrate was overestimated by the simulation. In an earlier study, overestimation of nitrate retention in FO by a phenomenological model for electrolyte permeation was also reported by Hancock et al. (2011). The authors attributed the discrepancy between the model predictions of nitrate retention and the experimental results to the interactions between the polar structure of nitrate and the membrane, which increases nitrate partitioning to the membrane. According to the modeling results, water reduction as high as 93% is theoretically achievable with a volume ratio of draw solution to feed solution of 2:1. At this extent of water reduction, ammonia and phosphate in the feed solution can theoretically be concentrated at a higher than 10-fold concentration. This concentration is comparable with those of animal wastewater and anaerobic digestion effluents (Cai et al., 2013), which have been widely tested as the feed for nutrient recovery from wastewater (Liu et al., 2011; Song et al., 2011). The simulation results indicate that a seawater-driven FO provides suitable performance for enriching nitrogen and phosphorous in treated municipal wastewater for further nutrient recovery. Nevertheless, as the feed solution is concentrated, water flux of FO continuously decreases, which significantly impacts the efficiency of FO process. Although increasing the volume ratio of draw solution to feed solution would attenuate the water flux decline by mitigating the impact of draw solution dilution, further improvement of membrane materials and module design is desired to produce higher efficiency of this FO process in future application.

4.

Conclusions

The performance of nutrients concentrated in treated municipal wastewater using seawater-driven FO was evaluated by both batch experiments and modeling approaches. At a water reduction of approximately 50%, DOC and phosphate were 2.3-fold concentrated, ammonia was 2.1-fold

128

w a t e r r e s e a r c h 6 9 ( 2 0 1 5 ) 1 2 0 e1 3 0

Table 2 e Parameters used for modeling FO in batch tests. Water permeability A, (m$s1 Pa1) Feed solution

Draw solution

a

Solute permeability Bf, (m s1)

NH4Cl NaNO2 NaNO3 NaH2PO4

Osmotic proportionality coefficient Hf, (Pa m3 mol1) Draw solute permeability Bd, (m s1) Resistance to solute diffusion within membrane porous support layer, Kd, (s m1) Osmotic proportionality coefficient Hd, (Pa m3 mol1)

2.00 9.49 1.60 1.75 6.54 4.25 2.72 7.13

       

1012 108 107 107 109 103 107a 105

4.25  103

Value for NaCl.

Fig. 7 e Model fitting of water permeation and nutrient concentration in the feed solution in batch FO filtration.

w a t e r r e s e a r c h 6 9 ( 2 0 1 5 ) 1 2 0 e1 3 0

concentrated, while nitrite and nitrate were not well concentrated, and were as low as 1.9-fold and 1.3-fold, respectively. In addition, retentions higher than 80% were usually obtained for trace metals. The simulation results demonstrated the feasibility of FO for concentrating municipal wastewater nutrients: the major forms of nitrogen and phosphorous (i.e., ammonia and phosphate) could theoretically be obtained at concentrations higher than 10-fold using a volume ratio of draw solution to feed solution of 2:1. Based on the results, further studies of the feasibility and process design of seawater-driven FO aimed at nutrients enrichment are worth conducting. In addition, the membrane charge property was found to be a potential factor dominating the behavior of electrolyte transfer in the FO process. It is possible for a highly charged FO membrane to increase the permeation and exchange of counter-ions from both sides of the membrane due to the special double solution system of the FO process. Feed solution pH played an essential role in nutrient retention, especially ammonia retention, by the FO membrane due to speciation variation with respect to the pH. A neutral pH of approximately 7 is optimal for improving the overall retention of nutrients by the FO membrane.

Acknowledgments This study was conducted under the support of the New Energy and Industrial Technology Development Organization (NEDO) and Japan Society for the Promotion of Science (JSPS) through the Funding Program for World-Leading Innovative R&D on Science and Technology (FIRST) of the Council for Science and Technology Policy, Cabinet Office, Government of Japan, and a collaboration research with Bureau of Sewerage, Tokyo Metropolitan Government.

Appendix A. Supplementary data Supplementary data related to this article can be found at http://dx.doi.org/10.1016/j.watres.2014.11.007.

references

Achilli, A., Cath, T.Y., Marchand, E.A., Childress, A.E., 2009. The forward osmosis membrane bioreactor: a low fouling alternative to MBR processes. Desalination 239 (1), 10e21. Bilstad, T., 1995. Nitrogen separation from domestic wastewater by reverse osmosis. J. Membr. Sci. 102, 93e102. Cai, T., Park, S.Y., Li, Y., 2013. Nutrient recovery from wastewater streams by microalgae: status and prospects. Renew. Sustain. Energy Rev. 19, 360e369. Cath, T.Y., Gormly, S., Beaudry, E.G., Flynn, M.T., Adams, V.D., Childress, A.E., 2005a. Membrane contactor processes for wastewater reclamation in space: Part I. Direct osmotic concentration as pretreatment for reverse osmosis. J. Membr. Sci. 257 (1), 85e98. Cath, T.Y., Adams, D., Childress, A.E., 2005b. Membrane contactor processes for wastewater reclamation in space: II. Combined

129

direct osmosis, osmotic distillation, and membrane distillation for treatment of metabolic wastewater. J. Membr. Sci. 257 (1), 111e119. Cath, T.Y., Childress, A.E., Elimelech, M., 2006. Forward osmosis: principles, applications, and recent developments. J. Membr. Sci. 281 (1), 70e87. Chen, J., Yamamoto, K., Tobino, T., 2012. Concentrated Biomass Production from an In-line Sludge Thickener Followed by a Membrane Bioreactor (Buenos Aires, Argentina). Chen, L., Gu, Y., Cao, C., Zhang, J., Ng, J.W., Tang, C., 2014. Performance of a submerged anaerobic membrane bioreactor with forward osmosis membrane for low-strength wastewater treatment. Water Res. 50, 114e123. Childress, A.E., Elimelech, M., 1996. Effect of solution chemistry on the surface charge of polymeric reverse osmosis and nanofiltration membranes. J. Membr. Sci. 119 (2), 253e268. Cornelissen, E.R., Harmsen, D., de Korte, K.F., Ruiken, C.J., Qin, J.J., Oo, H., Wessels, L.P., 2008. Membrane fouling and process performance of forward osmosis membranes on activated sludge. J. Membr. Sci. 319 (1), 158e168. Doyle, J.D., Parsons, S.A., 2002. Struvite formation, control and recovery. Water Res. 36 (16), 3925e3940. Gray, G.T., McCutcheon, J.R., Elimelech, M., 2006. Internal concentration polarization in forward osmosis: role of membrane orientation. Desalination 197 (1), 1e8. Ge, Q., Su, J., Amy, G.L., Chung, T.S., 2012. Exploration of polyelectrolytes as draw solutes in forward osmosis processes. Water Res. 46 (4), 1318e1326. Hancock, N.T., Phillip, W.A., Elimelech, M., Cath, T.Y., 2011. Bidirectional permeation of electrolytes in osmotically driven membrane processes. Environ. Sci. Technol. 45 (24), 10642e10651. Holloway, R.W., Childress, A.E., Dennett, K.E., Cath, T.Y., 2007. Forward osmosis for concentration of anaerobic digester centrate. Water Res. 41 (17), 4005e4014. Jiao, B., Cassano, A., Drioli, E., 2004. Recent advances on membrane processes for the concentration of fruit juices: a review. J. Food Eng. 63 (3), 303e324. Jin, X., Tang, C.Y., Gu, Y.S., She, Q.H., Qi, S.R., 2011. Boric acid permeation in forward osmosis membrane processes: modeling, experiments, and implications. Environ. Sci. Technol. 45 (6), 2323e2330. Lee, K.L., Baker, R.W., Lonsdale, H.K., 1981. Membranes for power generation by pressure-retarded osmosis. J. Membr. Sci. 8 (2), 141e171. Lee, S., Lueptow, R.M., 2001. Membrane rejection of nitrogen compounds. Environ. Sci. Technol. 35 (14), 3008e3018. Lee, S., Boo, C., Elimelech, M., Hong, S., 2010. Comparison of fouling behavior in forward osmosis (FO) and reverse osmosis (RO). J. Membr. Sci. 365 (1), 34e39. Linares, R.V., Yangali-Quintanilla, V., Li, Z., Amy, G., 2011. Rejection of micropollutants by clean and fouled forward osmosis membrane. Water Res. 45 (20), 6737e6744. Lind, B.B., Ban, Z., Byden, S., 2000. Nutrient recovery from human urine by struvite crystallization with ammonia adsorption on zeolite and wollastonite. Bioresour. Technol. 73 (2), 169e174. Lind, B.B., Ban, Z., Byden, S., 2001. Volume reduction and concentration of nutrients in human urine. Ecol. Eng. 16 (4), 561e566. Liu, Y., Kwag, J.-H., Kim, J.-H., Ra, C., 2011. Recovery of nitrogen and phosphorus by struvite crystallization from swine wastewater. Desalination 277 (1), 364e369. Loeb, S., Titelman, L., Korngold, E., Freiman, J., 1997. Effect of porous support fabric on osmosis through a Loeb-Sourirajan type asymmetric membrane. J. Membr. Sci. 129 (2), 243e249. McCutcheon, J.R., McGinnis, R.L., Elimelech, M., 2005. A novel ammonia-carbon dioxide forward (direct) osmosis desalination process. Desalination 174 (1), 1e11.

130

w a t e r r e s e a r c h 6 9 ( 2 0 1 5 ) 1 2 0 e1 3 0

McCutcheon, J.R., McGinnis, R.L., Elimelech, M., 2006. Desalination by ammonia-carbon dioxide forward osmosis: Influence of draw and feed solution concentrations on process performance. J. Membr. Sci. 278 (1), 114e123. Mi, B.X., Elimelech, M., 2010. Organic fouling of forward osmosis membranes: fouling reversibility and cleaning without chemical reagents. J. Membr. Sci. 348 (1), 337e345. Nayak, C.A., Rastogi, N.K., 2010. Forward osmosis for the concentration of anthocyanin from Garcinia indica Choisy. Sep. Purif. Technol. 71 (2), 144e151. Ng, H.Y., Tang, W.L., Wong, W.S., 2006. Performance of forward (direct) osmosis process: membrane structure and transport phenomenon. Environ. Sci. Technol. 40 (7), 2408e2413. Paugam, L., Taha, S., Dorange, G., Jaouen, P., Quemeneur, F., 2004. Mechanism of nitrate ions transfer in nanofiltration depending on pressure, pH, concentration and medium composition. J. Membr. Sci. 231 (1), 37e46. Phillip, W.A., Yong, J.S., Elimelech, M., 2010. Reverse draw solute permeation in forward osmosis: modeling and experiments. Environ. Sci. Technol. 44 (13), 5170e5176. Phuntsho, S., Shon, H.K., Majeed, T., El Saliby, I., Vigneswaran, S., Kandasamy, J., Hong, S., Lee, S., 2012. Blended fertilizers as draw solutions for fertilizer-drawn forward osmosis desalination. Environ. Sci. Technol. 46 (8), 4567e4575. Qin, J.J., Oo, M.H., Wai, M.N., Wong, F.S., 2003. Effect of feed pH on an integrated membrane process for the reclamation of a combined rinse water from electroless nickel plating. J. Membr. Sci. 217 (1), 261e268. Richards, L.A., Vuachere, M., Schafer, A.I., 2010. Impact of pH on the removal of fluoride, nitrate and boron by nanofiltration/ reverse osmosis. Desalination 261 (3), 331e337.

Song, Y., Yuan, P., Zheng, B., Peng, H., Yuan, F., Gao, Y., 2007. Nutrients removal and recovery by crystallization of magnesium ammonium phosphate from synthetic swine wastewater. Chemosphere 69 (2), 319e324. Song, Y.-H., Qiu, G.-L., Yuan, P., Cui, X.-Y., Peng, J.-F., Zeng, P., Duan, L., Xiang, L.-C., Qian, F., 2011. Nutrients removal and recovery from anaerobically digested swine wastewater by struvite crystallization without chemical additions. J. Hazard. Mater. 190 (1), 140e149. Tang, C.Y.Y., She, Q.H., Lay, W.C.L., Wang, R., Fane, A.G., 2010. Coupled effects of internal concentration polarization and fouling on flux behavior of forward osmosis membranes during humic acid filtration. J. Membr. Sci. 354 (1), 123e133. U.S.EPA, 1994. Short-term Methods for Estimating the Chronic Toxicity of Effluents and Receiving Waters to Marine and Estuarine Organisms, second ed. van Voorthuizen, E.M., Zwijnenburg, A., Wessling, M., 2005. Nutrient removal by NF and RO membranes in a decentralized sanitation system. Water Res. 39 (15), 3657e3667. Xie, H., Saito, T., Hickner, M.A., 2011. Zeta potential of ionconductive membranes by streaming current measurements. Langmuir 27 (8), 4721e4727. Yip, N.Y., Tiraferri, A., Phillip, W.A., Schiffman, J.D., Elimelech, M., 2010. High performance thin-film composite forward osmosis membrane. Environ. Sci. Technol. 44 (10), 3812e3818. Zhang, J., She, Q., Chang, V.W., Tang, C.Y., Webster, R.D., 2014. Mining nutrients (N, K, P) from urban source-separated urine by forward osmosis Dewatering. Environ. Sci. Technol. 48 (6), 3386e3394.