Ion Selective Permeation Through Cellulose Acetate Membranes in

Oct 24, 2013 - Andriy Yaroshchuk , Merlin L. Bruening. Journal of Membrane ... Wang , Yuefeng F. Xie. Journal of Membrane Science 2015 476, 410-420 ...
0 downloads 0 Views 401KB Size
Download PDF
Article pubs.acs.org/est

Ion Selective Permeation Through Cellulose Acetate Membranes in Forward Osmosis Gavin J. Irvine, Sahadevan Rajesh, Michael Georgiadis, and William A. Phillip* Department of Chemical and Biomolecular Engineering, University of Notre Dame, 182 Fitzpatrick Hall, Notre Dame, Indiana 46656, United States S Supporting Information *

ABSTRACT: Solute−solute interactions can have a dramatic impact on the permeation of solutes through dense polymeric membranes. In particular, understanding how solute−solute interactions can affect the design of osmotically driven membrane processes (ODMPs) is critical to the successful development of these emerging water treatment and energy generation processes. In this work, we investigate the influence that solute−solute interactions have on nitrate permeation through an asymmetric cellulose acetate forward osmosis membrane. A series of experiments that included systematic modifications to the cation paired with nitrate, the identity of the draw solute, and the solution pH were conducted. These experiments reveal that in the unique operating geometry of ODMPs, where solute containing solutions are present on both sides of the membrane, nitrate fluxes are significantly higher (>15 times in some cases) than predicted by existing models for solute permeation in ODMPs. The identity of the cation paired with nitrate influences the flux of nitrate; the identity of the cation in the draw solution does not affect the flux of nitrate; however, the identity of the anion in the draw solution has the most significant impact on the flux of nitrate. These results suggest that an ion exchange mechanism, which allows nitrate to switch rapidly with anions from the draw solution, is present when cellulose acetate based membranes are used in ODMPs.

■

more concentrated draw solution.3 Unfortunately, existing membranes are not perfect barriers, and solutes can also permeate across the membrane. Because of the unique operating geometry of ODMPs, solutes can permeate in the same direction as the water flux, forward permeation, or in the opposite direction of the water flux, reverse permeation. A number of recent papers have explored the topic of solute permeation in ODMPs through a variety of lenses. A common goal of these papers was to investigate the potential negative impacts that forward and reverse solute permeation could have on an overall osmotic process, such as increased downstream processing, enhanced fouling, and loss of driving force. The reverse flux of solute and reverse flux selectivity (i.e., the ratio of the forward water flux to the reverse flux of solute) are important metrics when analyzing the cost and effectiveness of a draw solute for a given application.18,19 It has also been noted that the reverse permeation of draw solutes influences fouling rates in ODMPs,10,12,20,21 as well as the operation of semibatch processes such as osmotic membrane bioreactors.22,23 The forward flux of solutes from feed solution to draw solution is also an active area of study. In particular, the ability of forward

INTRODUCTION Membrane-based water purification and desalination technologies already play an important role in the water supply portfolio. For example, reverse osmosis (RO) is the state-of-the-art technology for large-scale seawater desalination;1 ultrafiltration (UF) is used to ensure microbial and viral clearance from drinking water, and membrane bioreactors (MBRs) are increasingly being used for wastewater treatment.2 As the demand for fresh water continues to grow, new methods for extracting useable water from nontraditional sources will need to be developed. Osmotically driven membrane processes (ODMPs) are an emerging set of membrane technologies,3 which have shown promise in treating highly impaired water sources,4,5 enhancing seawater desalination,6,7 and generating energy.8 Several attractive features, such as their lower irreversible fouling propensity9−12 and operation at low applied pressures, have driven the interest in ODMPs. However, there are still obstacles that must be overcome for these emerging technologies to be developed into accepted processes. Among the areas that require further studies, a fundamental understanding of solute permeation in ODMPs is necessary to exploit the full potential of these processes for water treatment.13−17 In ODMPs, two solutions of differing concentration are placed on opposite sides of a semipermeable membrane, which induces water permeation from the less concentrated feed solution to the © 2013 American Chemical Society

Received: Revised: Accepted: Published: 13745

August 12, 2013 October 20, 2013 October 24, 2013 October 24, 2013 dx.doi.org/10.1021/es403581t | Environ. Sci. Technol. 2013, 47, 13745−13753

Environmental Science & Technology

■

osmosis (FO) systems to remove contaminants such as boron (in the form of boric acid)24,25 and trace organic contaminants26−28 has been compared to the rejection of these solutes by RO systems. In all of these studies, it was concluded that the two techniques were comparable in their ability to remove these dissolved solutes. Predictive models for solute permeation in ODMPs have been developed.14,15,29 When the appropriate phenomena (e.g., concentration polarization,15,30 solute−solvent interactions,16,31 and chemical equilibrium17) are included in the model, the agreement between the model and experiment is strong. Among these existing studies, a limited number have examined the influence of solute−solute interactions on the bidirectional permeation of solutes in ODMPs. Specifically, as solutes from the draw solution move past solutes from the feed solution within the membrane, they may interact with each other and alter their permeation rates. One prior study combined experimental and modeling efforts to explore the bidirectional permeation of strong electrolytes.14 In that effort, it was found that the reverse fluxes of ions could be predicted accurately using solute transport coefficients measured for single salt systems (i.e., when the salt was diffusing across the membrane in the absence of other solutes), which indicated that the effects of solute−solute interactions were minimal for most of the systems studied. Evidence of electrostatic interactions between ions from the feed solution and ions from the draw solution was observed, but the effect was minor. In the study described above and the initial study on the bidirectional permeation of solutes, it was noticed that for nitrate containing systems, the experimentally measured fluxes of nitrate ions were significantly higher than those predicted by models that ignored solute−solute interactions.13,14 These deviations could have significant implications on the design of ODMP unit operations. For example, several researchers have proposed using innovative FO systems to treat impaired water sources.3,32 Ultimately, this water may be used for human consumption, in which case, the systems will need to be designed to remove nitrate effectively due to the link between nitrate consumption and human health concerns (e.g., methemoglobinemia). Therefore, understanding these deviations, and their impact on ODMPs, is an important step in the development of these emerging processes, especially as evidence for solute−solute interactions is being observed in other systems.33 In this study, we seek to develop a better understanding of bidirectional permeation of solutes in ODMPs. Specifically, a series of experiments are performed in an attempt to elucidate why existing models for solute permeation in ODMPs do not accurately predict experimentally measured nitrate fluxes, even though these models have demonstrated good success in predicting fluxes for many other systems. In the series of experiments, solution properties such as pH, the identity of the salt used as a draw solute, and the cation paired with nitrate in the feed solution are modified in order to observe their effect on the flux of nitrate through a cellulose acetate membrane. Taken together these experiments suggest that cellulose acetate membranes have an ion selective nature in the presence of nitrate ions that causes anions to permeate more rapidly than predicted by models that ignore solute−solute interactions. The implications of the results for the operation of ODMPs along with the implications on the current understanding of solute permeation in water treatment membranes are discussed.

Article

MATERIALS AND METHODS

Feed and Draw Solutions. All of the salts used in this study were at least 99% pure and were obtained from Sigma-Aldrich. The pH of the feed and draw solutions was adjusted using buffers to determine the effect of pH. Citric acid, potassium hydrogen phthalate (KHP), and tris(hydroxymethyl)aminomethane (TRIS; Sigma-Aldrich, St. Louis, MO) were used to produce solutions at pH 3, 4, and 8, respectively. A Fisher Scientific AP 115 pH/ORP meter with an Accumet Gel-Filled polymer body pH/ATC double junction combination electrode was used to monitor the pH of the solutions over the course of the experiments. A Millipore water purification system (Milli Q Advantage A10, Millipore Corporation, Billerica, MA) provided deionized water (DI water), which was used for preparing the draw and feed solutions, rinsing the test cell at conclusion of an experiment, and diluting samples for ion chromatography analysis. Forward Osmosis Membrane and Bench Scale System. An asymmetric cellulose triacetate (CTA) membrane (Hydration Technology Innovations, Albany, OR), which has been used extensively in prior research efforts, was used in all the experiments reported here.13−15,34 Experiments where the active layer of the membrane faces the draw solution are referred to as experiments run in the PRO mode. Experiments where the active layer faces the feed solution are referred to as experiments run in the FO mode. A detailed description of the bench scale system that was used for the forward osmosis experiments is provided in several references.34,35 The water flux, Jw, from the feed solution to the draw solution was calculated from the mass vs. time data. Solution Sampling, Ion Analysis, and Flux Calculations. Ion fluxes were quantified in several different types of experiments. Single salt tests were performed with a draw solution that contained a single salt and a feed solution of DI water. These tests were used to determine the characteristic membrane parameters for the electrolyte systems used in this study (e.g., the solute permeability coefficients, B). Mixed salt tests are defined as tests where two distinct salts were dissolved in the draw solution and DI water was used as the feed solution. Last, in bidirectional permeation experiments, both the draw solution and the feed solution contained dissolved salts. These experiments are designed to investigate the influence of draw solutes on the permeation of feed solutes. The solute (i.e., ion) flux, Js, was calculated by measuring the ion concentration at regular intervals over the course of a 2 h experiment. A simple mass balance was used to relate the solute flux to the time evolution of the solute concentration c(V0 ± JW A m t )Js A m t

(1)

where c is the solute concentration in solution, V0 is the initial volume of the solution, JW is the measured water flux, Am is the membrane area, and t is the time. Addition is used in the parentheses for the draw solution because its volume is increasing. Subtraction is used for the feed solution. For single salt experiments, the increasing ion concentration in the feed solution was determined via a conductivity probe (Oakton CON 6 Acorn Series). For mixed salt and bidirectional permeation experiments, 2.5 mL samples of the feed and draw solution were withdrawn every half hour. The collected samples were diluted (5:1 v/v DI water to sample) and analyzed using ion chromatography (IC; A64 Thermo Dionex ICS-5000 using AS23 and AS16 columns for anion and cation analysis, respectively). 13746

dx.doi.org/10.1021/es403581t | Environ. Sci. Technol. 2013, 47, 13745−13753

Environmental Science & Technology

Article

Inductively coupled plasma-optical emission spectroscopy was used to determine the amount of potassium in solution.

■

RESULTS AND DISCUSSION Membrane Characteristics. Single salt experiments were conducted in PRO and FO modes to determine the solute permeability coefficients, B, and structural parameter, S, of the membrane, respectively. The PRO mode (i.e., draw solution adjacent to the active layer but with no applied pressure) experiments were executed at several different concentrations to determine if the solute permeability coefficients were a function of concentration; they were not (Figure S2). Recent reports have demonstrated that the B value of some membranes designed for ODMPs may vary when pressure is applied.33 No hydraulic pressure is applied to the membrane in the experiments reported here, so the solute permeability coefficients are taken as constant. The average B value for each salt is tabulated in Table 1. These Table 1. Characteristic Transport Coefficients for Dissolved Saltsa salt

B (10−8 m/s)

D (10−9 m2/s)

k (10−5 m/s)

NaClO4 KNO3 NaNO3 NaClO3 NaBr NH4Cl KCl NaCl MgCl2 Ca(NO3)2 Na2SO4

21.7 18.1 12.8 10.6 8.53 7.68 5.92 4.72 1.61 1.42 0.65

1.53 1.91 1.56 1.67 1.62 2.00 1.99 1.61 1.25 1.28 1.23

1.4 1.6 1.4 1.5 1.5 1.7 1.7 1.5 1.2 1.3 1.2

a

The permeability coefficients were determined using the method described in ref 15. The diffusion coefficients were calculated using the ionic diffusion coefficients from ref 36, and the mass transfer coefficients were calculated using the correlation developed in ref 37 for the crossflow cell geometry used in this study.

values were compared to previously published values and showed good agreement with the literature.13,14 The membrane structural parameter found from FO experiments was equal to 498 μm. This compares well with previously published results.15,34,36,37 These transport parameters were used in concert with an existing model for solute permeation in ODMPs, which ignores the influence of solute−solute interactions and assumes a solution-diffusion transport mechanism through the membrane active layer, to predict the flux of ions (details and schematic provided in the Support Information).14,15 Experimentally measured ion fluxes are compared to these predictions below. Influence of Draw Solute Concentration on Feed Solute Flux. In this set of bidirectional permeation experiments, three different types of feed solutions were used: a sodium nitrate solution, a sodium bromide solution, and a sodium perchlorate solution. The concentration of salt in the feed solution was kept constant at 0.05 M, and draw solutions with increasing concentrations of sodium chloride were used (Figure 1). Increasing the concentration of the draw solution increases the water flux generated (Figure S3), which results in increased concentration polarization on the feed side of the membrane. However, beyond this increased concentration polarization, which is accounted for in the existing model for solute

Figure 1. The fluxes of anions in bidirectional permeation experiments plotted vs the draw solution concentration. Feed solutions containing (a) sodium bromide, (b) sodium nitrate, and (c) sodium perchlorate were used. The concentration of salt in the feed solution was held constant at 0.05 M. The concentration of sodium chloride in the draw solution was varied between 0.25 and 2.0 M. Discrete points indicate experimentally measured fluxes; the error bars represent one standard deviation. The solid lines are the fluxes predicted by an existing model that assumes no interactions between the feed solute and draw solute.14 The solute permeability coefficients reported in Table 1 and the experimentally measured water fluxes were used to calculate the predicted fluxes. In panel b, the dotted lines represent the range of fluxes predicted when the variability in the solute permeability coefficient for NaNO3, B (12.8 × 10−8 ± 2.0 × 10−8 m/s), structural parameter, S (498 ± 61 μm), and mass transfer coefficient, k (1.4 × 10−5 ± 0.7 × 10−5 m/ s), which is observed between membrane samples, are taken into account.

permeation in ODMPs, the concentration of sodium chloride in the draw solution should not affect the flux of the anions from the feed solution. The anion fluxes predicted by the existing model for solute permeation in ODMPs are plotted as solid lines in Figure 1. The curve is nearly flat for anions from the feed solution because the effects of concentration polarization are miniscule. For the 13747

dx.doi.org/10.1021/es403581t | Environ. Sci. Technol. 2013, 47, 13745−13753

Environmental Science & Technology

Article

sodium bromide and sodium perchlorate feed solutions, good agreement between the experimentally measured anion fluxes, which are plotted as discrete points in Figure 1, and the model predictions is observed. Interestingly, the experimentally measured nitrate fluxes increase noticeably as the concentration of the sodium chloride draw solution increases. Error propagation calculations were completed to determine if the observed deviations between the experimental and predicted fluxes of nitrate were the result of the variation between membrane samples and/or experimental uncertainty in the solute permeability coefficients.38 For this calculation, one standard deviation of the solute permeability coefficient, B, was used as the error, and an error of 50% was assumed for the external mass transfer coefficient, k. The dashed lines in Figure 1 represent the results of this calculation and provide an upper and lower limit for the predicted fluxes. The experimentally measured nitrate fluxes fall well outside of the spread in predicted fluxes, which indicates that it is not variability in membrane samples or experimental uncertainty that is causing the differences we observe. Instead, it is likely that the fundamental assumptions used to develop the existing model for solute permeation in ODMPs are not valid for nitrate salts permeating through the CTA membrane. The subsequent sets of experiments were conducted to explore this deviation in more detail. Anion Fluxes in Mixed Solute Systems. In the solution diffusion mechanism of solute transport, the solute first dissolves in the membrane and then diffuses down its concentration gradient. As a result, the solute permeability coefficients depend on the product of the ion partition coefficient and diffusion coefficient.39 Ion partitioning experiments were conducted to measure the partition coefficients of nitrate and chloride for single salt and mixed salt solutions to explore the possibility that increased anion partition coefficients were responsible for the large deviation between predicted and measured anion fluxes observed in Figure 1 (details in Supporting Information Table S2). The sodium nitrate and sodium chloride concentrations were selected to mimic the concentrations used in the bidirectional tests. The partition coefficient of chloride did not change between the single salt and mixed salt experiments (Table S2). A slight increase in the partition coefficient of nitrate was observed, 0.14 to 0.19. However, the increase was small and not enough to explain the fundamentally different behavior in Figure 1. An increased membrane potential is likely the driving force for the increase of the nitrate partition coefficient in mixed salt systems.40 In the mixed salt systems, there is a higher concentration of sodium cations in the solution surrounding the membrane. This drives more sodium ions into the membrane, and a higher concentration of anions is needed to maintain electroneutrality. Because the nitrate ions are more soluble in the membrane phase than the chloride ions, they partition more readily, and an increased partition coefficient of nitrate is observed in mixed salt systems. A set of mixed solute experiments was conducted to complement the ion partitioning experiments. In these experiments, 0.05 M NaNO3 and a predetermined concentration of NaCl were dissolved in the draw solution. DI water was used as the feed solution. The trends of the chloride and nitrate fluxes measured during these experiments, which are presented in Figure 2, are notably different than the anion fluxes measured when the NaNO3 and NaCl solutions are present on opposite sides of the membrane. First, the existing model predicts accurately the experimentally measured chloride fluxes in the

Figure 2. The fluxes of anions in mixed salt draw experiments plotted vs the concentration of sodium chloride in the draw solution. The concentration of sodium nitrate in the draw solution was held constant at 0.05 M. The concentration of sodium chloride in the draw solution was varied between 0.25 and 1.0 M. Discrete points indicate experimentally measured fluxes. The solid lines are the fluxes predicted by an existing model that assumes no interactions between the feed solute and draw solute.14

mixed draw solute experiment. The model did not accurately predict the chloride flux in the bidirectional permeation experiment. Second, the slight deviation between the experimentally measured nitrate fluxes and the model predictions do not depend on the NaCl concentration, like they did in the bidirectional experiments. Instead the deviation remains constant, and its magnitude is consistent with the increased nitrate partitioning noted above. The nitrate partition coefficient increases 1.4 times, and the nitrate flux increases 1.8 times. Therefore, the unique operating geometry of ODMPs is necessary to produce the increased anion fluxes observed in Figure 1. Influence of Cation Identity on the Flux of Nitrate. In order to begin exploring the anomalous nitrate fluxes in bidirectional permeation experiments further, the identity of the cation in solution was varied, and the nitrate and chloride fluxes were quantified. Along with the sodium chloride system, potassium chloride and ammonium chloride systems (0.25 to 1.0 M) were used as draw solutions. The feed solution was maintained at a sodium nitrate concentration of 0.05 M. The experimental results in Figure 3a demonstrate that the identity of the draw cation paired with the chloride ions in the draw solution has little to no effect on the experimentally measured flux of nitrate. However, the experimentally measured nitrate fluxes for all three salt systems are greater than the nitrate fluxes predicted by the existing model. The identity of the cation in the draw solution does affect the measured flux of chloride, as shown in Figure 3b. For each draw solute, the measured chloride fluxes exceed the fluxes predicted by theory. It is interesting to note that not all the measured chloride fluxes are equal. Instead, the chloride fluxes measured for the KCl and NH4Cl systems are larger than the chloride flux measured for the NaCl draw solution. This may be attributed to the fact that the solute permeability coefficients for KCl and NH4Cl are higher than the solute permeability coefficient for NaCl. Subsequent experiments involved varying the cation paired with nitrate in the feed solution. In addition to the sodium nitrate system, potassium nitrate and calcium nitrate systems were studied. A sodium chloride draw solution of varying concentration was used in all of these experiments. The results plotted in Figure 4 again demonstrate that the measured fluxes of anions are 13748

dx.doi.org/10.1021/es403581t | Environ. Sci. Technol. 2013, 47, 13745−13753

Environmental Science & Technology

Article

Figure 3. The fluxes of chloride and nitrate plotted vs. the draw solute concentration for draw solutes containing various cations. In the legend, the first solute listed refers to the draw solute and the second to the feed solute. The concentration of sodium nitrate in the feed solution was held constant at 0.05 M. The concentration of the draw salt was varied between 0.25 and 1.0 M. Discrete points indicate experimentally measured fluxes; the error bars represent one standard deviation. The dashed lines are the fluxes predicted by an existing model using the solute permeability coefficients reported in Table 1.14

Figure 4. The fluxes of chloride and nitrate plotted vs. the draw solute concentration for feed solutes containing various cations. The concentration of nitrate in the feed solution was held constant at 0.05 M. For calcium nitrate, this corresponded to a molar concentration of the salt equal to 0.025M. Sodium chloride was used as the draw solute in all of the experiments; its concentration was varied between 0.25 and 1.0 M. Discrete points indicate experimentally measured fluxes; the error bars represent one standard deviation. The dashed lines indicate the fluxes predicted by an existing model using the solute permeability coefficients reported in Table 1.14

greater than the predicted fluxes of anions. In Figure 4a, the nitrate fluxes measured for the potassium nitrate and sodium nitrate experiments are comparable, but the nitrate flux measured in the experiments using calcium nitrate is measurably smaller when compared to the other two systems. These experimental results begin to suggest that there are two transport mechanisms for nitrate permeation: the solutiondiffusion mechanism and an ion exchange mechanism. In the solution-diffusion mechanism, an anion from the feed solution permeates across the membrane with a cation from the feed solution in order to maintain electroneutrality. In the ion exchange mechanism, an anion from the feed solution exchanges with an anion from the draw solution to maintain electroneutrality. The presence of two transport mechanisms would explain the differences in the nitrate fluxes between the NaNO3 and Ca(NO3)2 systems. The solute permeability coefficient of Ca(NO3)2 is small, so the nitrate flux generated by the solution diffusion mechanism will be small. The extremely low flux predicted for the Ca(NO3)2 feed solution provides evidence of

this. However, because there is a second mechanism for nitrate transport, even when the permeability of the parent salt is drastically reduced, the flux of nitrate is still appreciable. The same reasoning can be used to explain the difference in the chloride fluxes observed between the KCl and NaCl systems presented in Figure 3. That is, there is a larger contribution to the chloride flux from the solute diffusion mechanism for KCl because of its higher solute permeability coefficient. Therefore, the chloride flux was larger in this system than the NaCl system, even though the measured nitrate flux was the same in both experiments. Influence of Draw Anion Identity on the Flux of Nitrate. Experiments conducted using draw solutions that contained sodium bromide, sodium chlorate, and sodium sulfate provide further evidence for the presence of an ion exchange mechanism. The results of these experiments, which are plotted in Figure 5, demonstrate that the experimentally measured nitrate fluxes show a strong dependence on the identity of the anion in the draw solution. This is in stark contrast to the predicted nitrate 13749

dx.doi.org/10.1021/es403581t | Environ. Sci. Technol. 2013, 47, 13745−13753

Environmental Science & Technology

Article

Figure 5. The fluxes of the draw anion and nitrate plotted vs the draw solute concentration for draw solutes containing various anions. The concentration of sodium nitrate in the feed solution was held constant at 0.05 M. The experiment with a sodium sulfate draw solution was run with 0.05 M potassium nitrate feed solution. The concentration of draw salt was varied between 0.10 and 1.0 M. Discrete points indicate experimentally measured fluxes; the error bars represent one standard deviation. The dashed lines represent the fluxes predicted by an existing model using the solute permeability coefficients reported in Table 1.14

Figure 6. The fluxes of the draw and feed ions plotted vs the draw solute concentration for quaternary ion experiments. In the legend, the first solute listed refers to the draw solute and the second to the feed solute. The concentration of the feed salt was held constant at 0.05 M. The concentration of draw salt was varied between 0.25 and 1.0 M. Discrete points indicate experimentally measured fluxes; the error bars represent one standard deviation. Discrete points marked with a dot in the middle correspond to the anion flux, and discrete points marked with a cross in the middle correspond to the cation flux. The dashed lines represent the fluxes predicted by an existing model using the solute permeability coefficients reported in Table 1.14

fluxes, which do not depend on the identity of the anion in the draw solution. The magnitudes of the nitrate flux at a constant draw solution concentration suggest that the flux of nitrate increases with increasing draw solute permeability coefficients. This dependence on the identity of the anion in the draw solution is consistent with the Hofmeister series.41,42 Less hydrated ions have higher permeabilities and enhance the flux of nitrate relative to the model predictions to a larger extent than more hydrated ions, which have lower permeabilities. This observation seems reasonable because a more permeable draw anion would be able to exchange with a nitrate ion more rapidly. The results of the experiments using a sodium sulfate draw solution are interesting because the experimental nitrate fluxes match well with the predictions of the model. This suggests that when a draw anion with a sufficiently low permeability, i.e., one that cannot exchange with nitrate, is used, the flux of nitrate returns to the value predicted by existing models that only account for a solution-diffusion transport mechanism. This complements the results from the calcium nitrate system

discussed above to provide further evidence for the presence of two transport mechanisms. In the calcium nitrate experiment, the solution-diffusion mode of transport is limited due to the low solute permeability coefficient of this salt. However, because a relatively mobile chloride ion was present in the draw solution, the ion exchange mode was still active, resulting in an appreciable nitrate flux. Because the existing model for solute transport assumes a solution-diffusion mechanism, which is negligible in the case of Ca(NO3)2, large deviations between theory and experiment are observed. When sodium sulfate is used as the draw solute, the ion exchange mechanism is impeded because the large, divalent sulfate anion cannot effectively switch with the nitrate. Therefore, the solution-diffusion mechanism is the dominant mode for nitrate transport. Furthermore, strong agreement between theory and experiment is observed because the existing model for solute permeation assumes a solution-diffusion mechanism. 13750

dx.doi.org/10.1021/es403581t | Environ. Sci. Technol. 2013, 47, 13745−13753

Environmental Science & Technology

Article

structure of water.41,42 Therefore, it is possible that the structure of water within the membrane is responsible for the anion exchange behavior. However, recent studies are challenging the paradigm that the structure of water is responsible for the ordering of the Hofmeister series.49−51 These studies demonstrate that anion-solute interactions are driving the Hofmeister effects observed in some systems. Therefore, it is possible that these specific anion−solute interactions are driving a facilitated transport mechanism. Future work will focus on identifying if a specific chemical interaction or the state of water within the cellulose acetate membrane is responsible for the anion selective transport through cellulose acetate membranes. Regardless of its origins, accounting for the ion exchange that is a result of these solute− solute interactions is essential for the accurate design of ODMPs. Additionally, understanding these interactions at the molecular level may lend deeper insights in the permeation of solvents and solutes through polymeric membranes used for more well established membrane processes, such as reverse osmosis.

Ion Permeation in Quaternary Systems. Experiments conducted with quaternary ion systems provide further evidence for ion selective transport across cellulose acetate membranes in the presence of nitrate. In these experiments, the fluxes of all of the ions present were quantified. In Figure 6, the discrete data points represent the experimentally measured fluxes. The data points with a dot in the center correspond to anion fluxes, and data points with crosses in the center correspond to cation fluxes. For both the ions from the feed solution (Figure 6a) and the ions from the draw solution (Figure 6b), the measured fluxes of anions are larger than the measured fluxes of cations, which is consistent with an anion selective process. Similar to the ternary experiments reported above, the identity of the cation in the draw solution does not affect the flux of nitrate (e.g., the flux of nitrate is similar regardless of whether a divalent magnesium or monovalent sodium cation is used). Influence of pH on the Flux of Nitrate. A set of experiments was conducted to determine if pH affected the observed results. Buffers were used to control the pH of the sodium chloride draw solutions and sodium nitrate feed solutions at values of pH 3, 4 and, 8. The pH of the DI water experiment was 5.7. Over this pH range, the fluxes of the anions were not affected by the pH of the solutions surrounding the membrane, as shown by Figure S5. Experiments were also run where gradients in pH were maintained across the membrane (e.g., the draw solution was kept at pH 3 while the feed solution was held at pH 8), but no differences in the measured anion fluxes were observed (see SI for more details). On the Ion Selective Transport of Nitrate Through Cellulose Acetate Membranes. The experimental results presented above provide conclusive evidence for ion selective transport through cellulose acetate membranes in the presence of nitrate. Donnan dialysis and facilitated transport are two known mechanisms that are capable of producing anion exchange across a membrane. Donnan dialysis relies on the presence of positive charges within the membrane to produce anion exchange.43,44 This seems to be an unlikely explanation for the system studied here. First, Donnan dialysis is a nonselective process implying that all anions would exchange indiscriminately. We only observe the reported phenomenon for nitrate containing systems (Figure 1). Furthermore, bromide and perchlorate were selected as feed anions specifically because the permeabilities of these anions are lower and higher than the permeability of nitrate, respectively. If an electrically coupled transport process were responsible for the exchange of nitrate, deviations between model predictions and experiments would likely be observed for the more permeable perchlorate as well. Last, Donnan dialysis requires an abundance of positive charges within the membrane to produce anion selective transport. The cellulose acetate membranes used in this study have been characterized extensively, and these studies demonstrate that cellulose acetate membranes carry no charge or a slightly negative charge over the pH range examined.33,45 Therefore, Donnan dialysis is an unlikely explanation for the observed results. Facilitated transport is the term frequently used to refer to membrane transport mechanisms that rely on a selective chemical reaction or interaction to speed the permeation of a solute.43 Given the specificity of the system studied to nitrate, it seems reasonable that a selective interaction could underlie the reported results.46−48 Interestingly, the dependence of the flux of nitrate on the identity of the various draw anions did follow the Hofmeister series. This trend is frequently attributed to the

■

ASSOCIATED CONTENT

S Supporting Information *

Experimental details. Description of bidirectional solute permeation model. Solute permeability coefficients measured at different concentrations. Water fluxes generated during bidirectional permeation experiments. Mixed draw solution results. Molar fluxes of ions measured as a function of pH. This material is available free of charge via the Internet at http://pubs. acs.org.

■

AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS We acknowledge the support of the Center for Sustainable Energy at Notre Dame. We thank Jon Loftus and the Center for Environmental Science and Technology at the University of Notre Dame for access to and assistance with the ion chromatograph used in this study. Additionally, we thank Trenton Jackson for assistance in measuring some of the single salt permeability coefficients. Special thanks to Hydration Technology Innovations for providing the membrane utilized in this study.

■

REFERENCES

(1) Elimelech, M.; Phillip, W. A. The Future of Seawater Desalination: Energy, Technology, and the Environment. Science 2011, 333 (6043), 712−717. (2) Shannon, M. A.; Bohn, P. W.; Elimelech, M.; Georgiadis, J. G.; Marinas, B. J.; Mayes, A. M. Science and technology for water purification in the coming decades. Nature 2008, 452 (7185), 301−310. (3) Cath, T. Y.; Childress, A. E.; Elimelech, M. Forward osmosis: Principles, applications, and recent developments. J. Membr. Sci. 2006, 281 (1−2), 70−87. (4) McGinnis, R. L.; Hancock, N. T.; Nowosielski-Slepowron, M. S.; McGurgan, G. D. Pilot demonstration of the NH3/CO2 forward osmosis desalination process on high salinity brines. Desalination 2013, 312, 67−74. (5) Hickenbottom, K. L.; Hancock, N. T.; Hutchings, N. R.; Appleton, E. W.; Beaudry, E. G.; Xu, P.; Cath, T. Y. Forward osmosis treatment of

13751

dx.doi.org/10.1021/es403581t | Environ. Sci. Technol. 2013, 47, 13745−13753

Environmental Science & Technology

Article

drilling mud and fracturing wastewater from oil and gas operations. Desalination 2013, 312, 60−66. (6) Martinetti, C. R.; Childress, A. E.; Cath, T. Y. High recovery of concentrated RO brines using forward osmosis and membrane distillation. J. Membr. Sci. 2009, 331 (1−2), 31−39. (7) Cath, T. Y.; Hancock, N. T.; Lundin, C. D.; Hoppe-Jones, C.; Drewes, J. E. A multi-barrier osmotic dilution process for simultaneous desalination and purification of impaired water. J. Membr. Sci. 2010, 362 (1−2), 417−426. (8) Logan, B. E.; Elimelech, M. Membrane-based processes for sustainable power generation using water. Nature 2012, 488 (7411), 313−319. (9) Lay, W. C.; Chong, T. H.; Tang, C. Y.; Fane, A. G.; Zhang, J.; Liu, Y. Fouling propensity of forward osmosis: investigation of the slower flux decline phenomenon. Water Sci. Technol. 2010, 61 (4), 927−936. (10) Mi, B.; Elimelech, M. Chemical and physical aspects of organic fouling of forward osmosis membranes. J. Membr. Sci. 2008, 320 (1−2), 292−302. (11) She, Q. H.; Wong, Y. K. W.; Zhao, S. F.; Tang, C. Y. Y. Organic fouling in pressure retarded osmosis: Experiments, mechanisms and implications. J. Membr. Sci. 2013, 428, 181−189. (12) Zou, S.; Gu, Y. S.; Xiao, D. Z.; Tang, C. Y. Y. The role of physical and chemical parameters on forward osmosis membrane fouling during algae separation. J. Membr. Sci. 2011, 366 (1−2), 356−362. (13) Hancock, N. T.; Cath, T. Y. Solute coupled diffusion in osmotically driven membrane processes. Environ. Sci. Technol. 2009, 43 (17), 6769−6775. (14) Hancock, N. T.; Phillip, W. A.; Elimelech, M.; Cath, T. Y. Bidirectional permeation of electrolytes in osmotically driven membrane processes. Environ. Sci. Technol. 2011, 45 (24), 10642− 10651. (15) Phillip, W. A.; Yong, J. S.; Elimelech, M. Reverse Draw Solute Permeation in Forward Osmosis: Modeling and Experiments. Environ. Sci. Technol. 2010, 44 (13), 5170−5176. (16) Yong, J. S.; Phillip, W. A.; Elimelech, M. Coupled reverse draw solute permeation and water flux in forward osmosis with neutral draw solutes. J. Membr. Sci. 2012, 392, 9−17. (17) Yong, J. S.; Phillip, W. A.; Elimelech, M. Reverse Permeation of Weak Electrolyte Draw Solutes in Forward Osmosis. Ind. Eng. Chem. Res. 2012, 51 (41), 13463−13472. (18) Achilli, A.; Cath, T. Y.; Childress, A. E. Selection of inorganicbased draw solutions for forward osmosis applications. J. Membr. Sci. 2010, 364 (1−2), 233−241. (19) Ge, Q. C.; Ling, M. M.; Chung, T. S. Draw solutions for forward osmosis processes: Developments, challenges, and prospects for the future. J. Membr. Sci. 2013, 442, 225−237. (20) Boo, C.; Lee, S.; Elimelech, M.; Meng, Z. Y.; Hong, S. Colloidal fouling in forward osmosis: Role of reverse salt diffusion. J. Membr. Sci. 2012, 390, 277−284. (21) She, Q. H.; Jin, X.; Li, Q. H.; Tang, C. Y. Y. Relating reverse and forward solute diffusion to membrane fouling in osmotically driven membrane processes. Water Res. 2012, 46 (7), 2478−2486. (22) Nawaz, M. S.; Gadelha, G.; Khan, S. J.; Hankins, N. Microbial toxicity effects of reverse transported draw solute in the forward osmosis membrane bioreactor (FO-MBR). J. Membr. Sci. 2013, 429, 323−329. (23) Xiao, D. Z.; Tang, C. Y. Y.; Zhang, J. S.; Lay, W. C. L.; Wang, R.; Fane, A. G. Modeling salt accumulation in osmotic membrane bioreactors: Implications for FO membrane selection and system operation. J. Membr. Sci. 2011, 366 (1−2), 314−324. (24) Jin, X.; Tang, C. Y.; Gu, Y. S.; She, Q. H.; Qi, S. R. Boric Acid Permeation in Forward Osmosis Membrane Processes: Modeling, Experiments, and Implications. Environ. Sci. Technol. 2011, 45 (6), 2323−2330. (25) Kim, C.; Lee, S.; Shon, H. K.; Elimelech, M.; Hong, S. Boron transport in forward osmosis: Measurements, mechanisms, and comparison with reverse osmosis. J. Membr. Sci. 2012, 419, 42−48. (26) Alturki, A. A.; McDonald, J. A.; Khan, S. J.; Price, W. E.; Nghiem, L. D.; Elimelech, M. Removal of trace organic contaminants by the forward osmosis process. Sep. Purif. Technol. 2013, 103, 258−266.

(27) Hancock, N. T.; Xu, P.; Heil, D. M.; Bellona, C.; Cath, T. Y. Comprehensive Bench- and Pilot-Scale Investigation of Trace Organic Compounds Rejection by Forward Osmosis. Environ. Sci. Technol. 2011, 45 (19), 8483−8490. (28) Xie, M.; Price, W. E.; Nghiem, L. D.; Elimelech, M. Effects of feed and draw solution temperature and transmembrane temperature difference on the rejection of trace organic contaminants by forward osmosis. J. Membr. Sci. 2013, 438, 57−64. (29) Field, R. W.; Wu, J. J. Mass transfer limitations in forward osmosis: Are some potential applications overhyped? Desalination 2013, 318, 118−124. (30) Yaroshchuk, A. Influence of osmosis on the diffusion from concentrated solutions through composite/asymmetric membranes: Theoretical analysis. J. Membr. Sci. 2010, 355 (1−2), 98−103. (31) Su, J. C.; Chung, T. S. Sublayer structure and reflection coefficient and their effects on concentration polarization and membrane performance in FO processes. J. Membr. Sci. 2011, 376 (1−2), 214−224. (32) Hoover, L. A.; Phillip, W. A.; Tiraferri, A.; Yip, N. Y.; Elimelech, M. Forward with Osmosis: Emerging Applications for Greater Sustainability. Environ. Sci. Technol. 2011, 45 (23), 9824−9830. (33) Coday, B. D.; Heil, D. M.; Xu, P.; Cath, T. Y. Effects of Transmembrane Hydraulic Pressure on Performance of Forward Osmosis Membranes. Environ. Sci. Technol. 2013, 47 (5), 2386−2393. (34) Yip, N. Y.; Tiraferri, A.; Phillip, W. A.; Schiffman, J. D.; Elimelech, M. High Performance Thin-Film Composite Forward Osmosis Membrane. Environ. Sci. Technol. 2010, 44 (10), 3812−3818. (35) McCutcheon, J. R.; McGinnis, R. L.; Elimelech, M. A novel ammonia-carbon dioxide forward (direct) osmosis desalination process. Desalination 2005, 174 (1), 1−11. (36) Cussler, E. L. Diffusion: mass transfer in fluid systems, 3rd ed.; Cambridge University Press: Cambridge, U. K., 2009; pp xvii, 631. (37) Hoek, E. M. V.; Kim, A. S.; Elimelech, M. Influence of crossflow membrane filter geometry and shear rate on colloidal fouling in reverse osmosis and nanofiltration separations. Environ. Eng. Sci. 2002, 19 (6), 357−372. (38) Bevington, P. R.; Robinson, D. K. Data Reduction and Error Analysis for the Physical Sciences, 3rd ed.; McGraw-Hill: Boston, 2003; pp xi, 320. (39) Geise, G. M.; Lee, H. S.; Miller, D. J.; Freeman, B. D.; Mcgrath, J. E.; Paul, D. R. Water Purification by Membranes: The Role of Polymer Science. J. Polym. Sci., Part B: Polym. Phys. 2010, 48 (15), 1685−1718. (40) Cheng, C.; Yaroshchuk, A.; Bruening, M. L. Fundamentals of Selective Ion Transport through Multilayer Polyelectrolyte Membranes. Langmuir 2013, 29 (6), 1885−1892. (41) Kunz, W.; Lo Nostro, P.; Ninham, B. W. The present state of affairs with Hofffieister effects. Curr. Opin. Colloid Interface Sci. 2004, 9 (1−2), 1−18. (42) Zhang, Y. J.; Cremer, P. S. Interactions between macromolecules and ions: the Hofmeister series. Curr. Opin. Chem. Biol. 2006, 10 (6), 658−663. (43) Baker, R. W. Membrane Technology and Applications, 2nd ed.; J. Wiley: Chichester, U. K., 2004; pp x, 538. (44) Sarkar, S.; Sengupta, A. K.; Prakash, P. The Donnan Membrane Principle: Opportunities for Sustainable Engineered Processes and Materials. Environ. Sci. Technol. 2010, 44 (4), 1161−1166. (45) Nguyen, T. P. N.; Yun, E. T.; Kim, I. C.; Kwon, Y. N. Preparation of cellulose triacetate/cellulose acetate (CTA/CA)-based membranes for forward osmosis. J. Membr. Sci. 2013, 433, 49−59. (46) Qi, Z.; Cussler, E. L. Membrane separation of potassium nitrate from mixed brines. J. Membr. Sci. 1984, 19 (3), 259−272. (47) Kreevoy, M. M.; Kotchevar, A. T.; Aften, C. W. Decontamination of Nitrate Polluted Water. Sep. Sci. Technol. 1987, 22 (2−3), 361−372. (48) Kreevoy, M. M.; Nitsche, C. I. Progress toward a Solution to the Nitrate Problem. Environ. Sci. Technol. 1982, 16 (9), 635−637. (49) Freire, M. G.; Neves, C. M. S. S.; Silva, A. M. S.; Santos, L. M. N. B. F.; Marrucho, I. M.; Rebelo, L. P. N.; Shah, J. K.; Maginn, E. J.; Coutinho, J. A. P. H-1 NMR and Molecular Dynamics Evidence for an Unexpected Interaction on the Origin of Salting-In/Salting-Out Phenomena. J. Phys. Chem. B 2010, 114 (5), 2004−2014. 13752

dx.doi.org/10.1021/es403581t | Environ. Sci. Technol. 2013, 47, 13745−13753

Environmental Science & Technology

Article

(50) Kalra, A.; Tugcu, N.; Cramer, S. M.; Garde, S. Salting-in and salting-out of hydrophobic solutes in aqueous salt solutions. J. Phys. Chem. B 2001, 105 (27), 6380−6386. (51) Holz, M. Nuclear magnetic relaxation as a selective probe of solute solvent and solute solute interactions in multi-component mixtures. J. Mol. Liq. 1995, 67, 175−191.

13753

dx.doi.org/10.1021/es403581t | Environ. Sci. Technol. 2013, 47, 13745−13753