ARTICLE pubs.acs.org/est
Comprehensive Bench- and Pilot-Scale Investigation of Trace Organic Compounds Rejection by Forward Osmosis Nathan T. Hancock,† Pei Xu,† Dean M. Heil,† Christopher Bellona,‡ and Tzahi Y. Cath*,† †
Department of Civil and Environmental Engineering, Colorado School of Mines, 1500 Illinois Street, Golden, Colorado 80401, United States ‡ Department of Civil and Environmental Engineering, Clarkson University, 8 Clarkson Avenue, Potsdam, New York 13699, United States
bS Supporting Information ABSTRACT: Forward osmosis (FO) is a membrane separation technology that has been studied in recent years for application in water treatment and desalination. It can best be utilized as an advanced pretreatment for desalination processes such as reverse osmosis (RO) and nanofiltration (NF) to protect the membranes from scaling and fouling. In the current study the rejection of trace organic compounds (TOrCs) such as pharmaceuticals, personal care products, plasticizers, and flame-retardants by FO and a hybrid FORO system was investigated at both the bench- and pilot-scales. More than 30 compounds were analyzed, of which 23 nonionic and ionic TOrCs were identified and quantified in the studied wastewater effluent. Results revealed that almost all TOrCs were highly rejected by the FO membrane at the pilot scale while rejection at the bench scale was generally lower. Membrane fouling, especially under field conditions when wastewater effluent is the FO feed solution, plays a substantial role in increasing the rejection of TOrCs in FO. The hybrid FO-RO process demonstrated that the dual barrier treatment of impaired water could lead to more than 99% rejection of almost all TOrCs that were identified in reclaimed water.
1. INTRODUCTION 1.1. Reclamation and Water Reuse and the Presence of Trace Organic Compounds (TOrCs) in Water and Wastewater. A recent report to Congress highlighted the conse-
quences of climate change on water resources in the Western and Southwestern US.1 With expected drier climate, new water resources will be required to fulfill the needs of growing population centers and economies. Coastal utilities already use seawater as an unconventional source of drinking water, and inland communities will have to increasingly rely on reclaimed water and desalinated brackish groundwater for nonpotable and indirect potable reuse.2 In the US and around the world water providers are already treating impaired water to supplement their fresh water supplies. Examples include the Water District Groundwater Replenishment System in Orange County, California,3 the Prairie Waters Project in Aurora, Colorado,4 and the NeWater treatment clusters in Singapore.5 These facilities rely on processes such as microfiltration (MF), reverse osmosis (RO), and ultraviolet irradiation with advanced oxidation processes (UV/AOP) to treat impaired water to a high quality. Yet, one of the major concerns related to water reuse is currently the presence of trace organic compounds (TOrCs) in effluents from conventional wastewater treatment facilities and consequently their current prevalence in many water resources.6,7 While RO, NF, UV/AOP, carbon adsorption, and natural system (i.e., riverbank filtration and aquifer recharge) processes demonstrated efficient removal r 2011 American Chemical Society
of TOrCs from impaired and treated water,7 12 new treatment processes might be required to provide multibarrier protection of potable waters. 1.2. Forward Osmosis As a New Process for Reclamation of Impaired Water. Recent advancements in research and development demonstrated that osmosis can be utilized in engineered systems to separate contaminants from water using the forward osmosis (FO) process.13 In FO, clean water can be extracted from contaminated streams utilizing special semipermeable membranes that have contaminant rejection capabilities similar to those of RO membranes. In the process, water diffuses from the feed stream, through a dense, semipermeable FO membrane, and into a draxw solution that has a high osmotic pressure.13 Past studies demonstrated that domestic wastewater, anaerobic digester sludge, food and beverage products, brackish groundwater, and even seawater could be effectively treated with FO.14 16 A recent study also demonstrated the effective dual barrier characteristics of the FO process when hybridized with seawater RO for treatment of secondary and tertiary treated domestic effluents.16 In the hybrid process, impaired water is first treated by an FO membrane and water is drawn into a seawater draw solution, and the diluted draw solution is desalinated by RO to produce a stream of purified water (see Supporting Received: May 16, 2011 Accepted: August 12, 2011 Revised: August 2, 2011 Published: August 12, 2011 8483
dx.doi.org/10.1021/es201654k | Environ. Sci. Technol. 2011, 45, 8483–8490
Environmental Science & Technology Information, section S1). The rejection of wastewater constituents such as organic carbon, nutrients (i.e., ammonia, nitrate, phosphate), and suspended solids by the two tight barriers was very high (95 99.9%), the FO membranes protected the sensitive RO membranes from fouling and scaling, and energy saving could be achieved.16 1.3. Previous Studies That Employed FO To Remove TOrCs from Impaired Water. An early bench-scale study by Cartinella et al.17 demonstrated the rejection by FO of a limited number of TOrCs from wastewater generated in life support systems. The study investigated the removal of three hormones that were spiked into batches of synthetic wastewater. Results revealed that membrane fouling and the presence of surfactants in the feed solution substantially improved rejection of TOrCs at higher water recoveries. A more recent study by Cath et al.16 compared TOrCs rejection by FO at the pilot- and bench-scale. A pilotscale FO-RO hybrid system was utilized, and a continuous stream of secondary effluent was used as a feed stream to an FO plate and frame membrane cell. Only a limited number of TOrCs was quantified in the effluent. Bench-scale tests were performed with water spiked with the same TOrCs. Results from short-term experiments (7 14 days) demonstrated that rejection of TOrCs by FO membranes and by the dual barrier FO-RO process are very high (>95%) and can lead to potentially successful incorporation of the FO process in future water reclamation systems for potable reuse. Although the previous studies demonstrated the high efficiency of FO in removal of TOrCs, the experiments were too short, with limited number of compounds, and with a custom-made membrane cell. Fundamental NF and RO studies8,18 20 have demonstrated that the rejection of TOrCs is primarily governed by steric interactions (i.e., increased rejection with increased molecular size), electrostatic exclusion, and adsorption (often related to solute hydrophobicity); however, other factors such as a solute’s functional groups (e.g., hydroxyl groups, carbonyl groups, amide groups) may play a role in TOrCs rejection.21 A similar systematic investigation is needed to better understand the rejection mechanisms of TOrCs during FO treatment. 1.4. Objectives. The main objective of the current study was to investigate the rejection of TOrCs by FO and by the dual barrier FO-RO processes using mainly seawater as a renewable source of draw solution. This includes testing for extended time (i.e., months) of newer generation FO membranes in a novel spiral wound commercial packaging configuration, with wastewater of different qualities (i.e., tertiary and secondary effluents), and with a larger number of TOrCs having a broader range of physical and chemical properties. Pilot-scale experiments were conducted to assess the performance of commercial FO membrane modules under field conditions, and bench-scale experiments were conducted to further elucidate the mechanisms and conditions controlling rejection of TOrCs by FO membranes. Bench-scale tests were conducted with a flat sheet FO membrane from the same cast, using synthetic feed solutions spiked with similar types and concentrations of TOrCs that were observed in the effluent of the treated wastewater, and with various draw solutions compositions.
2. MATERIAL AND METHODS 2.1. Membranes. FO membranes made from a hydrophilic cellulose-based polymer were acquired from Hydration Technology Innovations (HTI) (Albany, OR). The membranes incorporate a
ARTICLE
cellulose triacetate (CTA) active layer, and a woven polyester mesh is embedded in the CTA to create a flat sheet membrane. For the bench-scale study, flat sheet FO membrane coupons were used. For the pilot study the CTA membrane was employed in a spiral wound packaging configuration that improves membrane-packing density (i.e., membrane active surface area per module volume) for commercial applications (additional details presented in Supporting Information, section S2). Seawater RO spiral wound membranes (SW30 2540, Dow Filmtec, Edina, MN) were used for the pilot RO system. These membranes were chosen for their high salt rejection (>99.4%22) that enables production of concentrated draw solution for the FO process. Water permeance and NaCl rejection for the fresh FO membrane were measured during experiments conducted in RO mode and found to be 0.78 ( 0.02 L/m2-h-bar and 93.2 ( 0.9%, respectively. Feed at a constant temperature of 25 °C was either deionized water for permeance tests or a 2,000 mg/L NaCl solution for rejection tests. Tests were conducted with feed pressures of 100 and 150 psig (0.69 and 1.03 MPa). Water permeance and NaCl rejection for the SW30 membrane (1.31 L/m2-h-bar and 99.4%, respectively) were obtained from the manufacturer.22 2.2. Test Systems. 2.2.1. Bench-Scale FO Apparatus. The FO bench-scale apparatus utilized a membrane cell constructed with symmetric flow channels on both sides of the membrane that facilitated parallel, cocurrent flow along the 455-cm2 membrane. A programmable logic control (PLC) system was developed to maintain constant experimental conditions (i.e., feed volume of 3 L, draw solution concentration of 30 g/L, and system temperature of 20 °C) and to collect data during the experiments. Feed and draw solution were continuously circulated between their respective tanks and the membrane cell at a flow rate of 1.6 L/min, and feed and draw solution samples were intermittently drawn for analysis. Details about the design and operating conditions and control of the system are provided in our previous publication.23 2.2.2. Pilot-Scale FO-RO System. A pilot-scale FO-RO hybrid system was utilized for the long-term testing of the spiral wound commercial FO membrane. Details about the design and operating conditions and control of the system are provided in our previous publication16 and in the Supporting Information, section S3. In the current study the system was deployed at a wastewater treatment research facility on the campus of the Colorado School of Mines (Golden, CO). Approximately 7000 gallons per day (26.5 m3) of domestic wastewater are treated at the facility by a demonstration scale sequencing batch membrane bioreactor (SBMBR) system (Aqua-Aerobic Systems, Inc., Rockford, IL). Feed to the FO membrane was a continuous side stream from the SBMBR operated with an ultrafiltration (UF) membrane (Puron, KMS, Wilmington, MA). To simulate secondary treated effluent water quality, activated sludge from the SBMBR was dosed into the SBMBR permeate at different rates. 2.3. Water Chemistry. 2.3.1. Draw Solutions. Synthetic sea salt (Instant Ocean, Mentor, OH) was used for preparation of draw solution for all pilot- and two bench-scale experiments. During most of the pilot-scale experiment (approximately three continuous months) the draw solution was maintained at a concentration of approximately 30 g/L sea salt by the RO subsystem. For a limited time, the RO subsystem produced concentrated draw solution of 60 g/L sea salt. In one bench-scale experiments, ACS grade sodium chloride (Fisher Scientific, Atlanta, 8484
dx.doi.org/10.1021/es201654k |Environ. Sci. Technol. 2011, 45, 8483–8490
Environmental Science & Technology
ARTICLE
Table 1. Average Concentrations of Major Constituents and Water Quality Parameters Measured in SBMBR Permeate and Composition of the Feed Solution for the FO Bench-Scale Tests pilot-scale FO feed
bench-scale FO feed
constituent
mg/L
mg/L
bicarbonate boron
54.3 ( 1.7 0.08 ( 0.04
54 -
O-phosphate potassium
pilot-scale FO feed
bench-scale FO feed
mg/L
mg/L
12.3 ( 2.5 11.7 ( 1.7
barium
0.06 ( 0.02
-
pH
7.1 ( 0.1
bromide
0.06 ( 0.01
-
silica
10.6 ( 1.1
5 7.0 -
calcium
36.1 ( 3.6
35
sodium
54.5 ( 3.6
56
chloride
72.0 ( 5.6
97
strontium
0.20 ( 0.02
-
3.9 ( 0.7
4a
sulfate
79.8 ( 5.4
magnesium
10.3 ( 1.0
10
nitrate
37.3 ( 15.4
SUVA, L/mg 3 m
-
TDS
DOC
a
constituent
2.4 ( 0.1 325.0 ( 22.7
74 332
5 mg/L humic acid sodium salt (Aldrich, St. Louis, MO) was added to the feed.
GA) was used as draw solution at 30 g/L to elucidate draw solution chemistry effects on TOrCs rejection. 2.3.2. Feed Solutions. SBMBR permeate was the feed to the pilot-scale FO membrane and was continuously pumped into the feed side of the FO spiral wound module at a rate of 12 L/min. The quality of the feedwater (SBMBR effluent with or without activated sludge dosing) to the pilot FO fluctuated throughout the testing period because of the decentralized nature of the SBMBR facility. Fluctuation in the number of users, precipitation events, and constantly changing conditions in the SBMBR system may contribute to varying feedwater quality. Average SBMBR permeate water quality based on 23 samples taken over a 38 day period is summarized in Table 1. Data indicate that the dominant ions in the SBMBR permeate are sulfate, chloride, sodium, calcium, and magnesium. Variable nitrate concentration observed in samples may result from a variety of factors associated with the SBMBR system operation (e.g., aeration cycles and duration) and influent water quality. The concentrations of major constituents that were added to the feedwater during the bench-scale tests are also summarized in Table 1. 2.4. Analytical Methods for TOrCs. TOrCs were quantified using a liquid chromatography tandem mass spectrometry (LCMS/MS) system assembled with an HPLC system (1200 Series, Agilent Technologies Inc., Santa Clara, CA) coupled to a tandem mass spectrometer (3200 Q Trap, Applied Biosystems, Carlsbad, CA) according to the matrix suppression abating method described in Vandeford and Snyder.24 To alleviate issues induced by matrix suppression of environmental samples, an isotope of each TOrC was spiked into each sample at a known concentration. The absolute recovery of the isotope was used to adjust the concentration of the native TOrC. Solid phase extraction was used to preconcentrate TOrCs from the samples prior to LC-MS/MS analysis. The volume of sample used for solid phase extraction varied depending on the dissolved organic carbon (DOC) concentration in each sample. Volumes processed for RO product water, draw solution, and feed samples were 500, 250, and 100 mL, respectively; however, draw solution samples with total dissolved solids (TDS) exceeding 30 g/L were diluted prior to solid phase extraction to abate low isotope recovery. Samples were analyzed using both ESI+ and ESI modes24 to quantify 32 ionic and nonionic TOrCs. Of the 32 TOrCs, 23 were detected in samples and are summarized in Table 2. The concentrations of TOrCs in the synthetic feed during benchscale experiments are also summarized in Table 2. Previous
research8,12 designated hydrophobic nonionic TOrCs as those with log Kow values above 2. Ten of the 14 nonionic TOrCs in this study were hydrophobic with log D values (log Kow calculated at pH 6) above 2. Classification, structure, and charge of each TOrC were determined using ChemIDplus Advanced (United States National Library of Medicine, Bethesda, MD), and the octanol water distribution coefficient (log Kow) at solution pH of 6 (log D) was obtained from ACD/Laboratories software version 8.14 (ACD/LABORATORIES, Toronto, Canada). 2.5. Experimental Procedures. 2.5.1. Bench-Scale Experimental Procedures. Three bench-scale experiments were performed. For each experiment, 3 L of feed solution was prepared with deionized water dosed with the major ions and TOrCs identified in the SBMBR permeate (Tables 1 and 2, respectively). In the second and third experiments approximately 5 mg/L of humic acid (Aldrich, St. Louis, MO) was added to the feed for a targeted DOC concentration of 4 mg/L to examine the effect of organic matter on solute rejection. One liter of draw solution was prepared and transferred to the draw solution tank at the beginning of each experiment. In the first and second experiments the draw solution was seawater (Instant Ocean) at 30 g/L TDS, and in the third experiment the draw solution was sodium chloride at 30 g/L to compare the impact of salt composition on TOrCs permeation. Draw solution and feed concentrations were maintained constant throughout the experiments using a PLC system.23 Feed and draw solution samples were drawn for analysis at the beginning of each experiment, after 1 L of water permeated the membrane from the feed into the draw solution, and after a second liter of water permeated into the draw solution (end of the experiment). Water flux and reverse salt flux from the draw solution to the feed were monitored and recorded throughout each experiment. TOrCs concentrations and water flux were used to conduct mass balance and calculate TOrCs rejection by the FO membranes (Section S5 in the SI). Membrane integrity tests were conducted before and after the set of experiments to confirm that the rejection of ions by the membrane did not change during the experiments. Before each experiment, the feed solution was recirculated on the feed side of the membrane for 12 h to ensure that adsorption of TOrCs to the membrane did not affect mass transport and mass balance calculations. Following the 12 h stabilization period, each experiment lasted 4 5 h. 2.5.2. Pilot System Experimental Procedure. The pilot-scale system was initially tested with SBMBR permeate as FO feed continuously for 650 h (27 days). Subsequently, the FO feed was 8485
dx.doi.org/10.1021/es201654k |Environ. Sci. Technol. 2011, 45, 8483–8490
Environmental Science & Technology
ARTICLE
Table 2. TOrCs Detected by LC-MS/MS in ESI+ and ESI Modes and Select Physicochemical Properties TOrC
classification
charge
log D @ pH 6a
pilot FO feed,b ng/L
MW
bench FO feed,c ng/L
ESI+ amitriptyline
antidepressant
positive
1.45
277.40
3.0 ( 1.1
atenolol
beta-blocker
positive
2.73
266.34
26.7 ( 10.2
30
benzophenone
preservative
neutral
3.18
182.21
90.1 ( 22.0
100
caffeine
stimulant
neutral
0.13
194.19
32.2 ( 18.0
50
carbamazepine
anticonvulsant
neutral
2.67
236.27
388 ( 301
500
DEET
insect repellent
neutral
1.96
191.27
38.8 ( 22.3
50
diazepam
benodiazepine tranquilizer
neutral
2.96
284.70
0.63 ( 0.1
-
dilantin diphenhydramine
antiepileptic antihistamine
neutral positive
2.52 1.07
252.27 255.36
133 ( 28 132 ( 45
150 150
fluoexetine
antidepressant
positive
1.03
309.33
18.3 ( 6.0
20
hydrocodone
analgesic
neutral
0.61
299.37
14.3 ( 12.4
20
oxybenzone
ingredient in sunscreens
neutral
3.63
228.24
25.9 ( 8.3
30
-
primidone
anticonvulsant
neutral
0.40
218.25
1.6 ( 0.4
-
sulfamethoxazole
anti-infective
negative
0.49
253.28
185 ( 175
250
TCEP
flame retardant
neutral
2.67
250.19
366 ( 152
400
trimethoprim
anti-infective
positive
0.42
290.32
21.6 ( 17.6
40
bisphenol A
plasticizer
neutral
3.43
228.29
89.6 ( 26.8
100
diclofenac
anti-inflammatory agent
negative
2.23
296.15
38.4 ( 33.1
50
ibuprofen
anti-inflammatory agent
negative
2.12
206.29
34.1 ( 11.3
40
methylparaben
personal care product
neutral
1.86
152.15
23.7 ( 9.0
30
naproxen
anti-inflammatory agent
negative
1.81
230.26
74.1 ( 33.9
100
triclosan
antimicrobial
neutral
5.17
289.54
90.5 ( 23.7
100
triclocarban
antimicrobial
neutral
5.74
315.58
267 ( 41
300
ESI
a
Obtained from ACD/Laboratories software version 8.14 (ACD/LABORATORIES, Toronto, Canada). b Feed concentration of TOrCs measured in grab samples drawn prior to FO membrane during intervals B through E. c Concentration of TOrCs in the feed solution for the FO bench-scale tests.
Table 3. Summary of Operating Conditions during the Pilot Studyc days of
DS concentration
DS flow rate
TSS concentration
operation
mg/L
L/min
mg/L
0 14 (A)
30,000 ( 600 (est.)
1.6 ( 0.3
0
14 21 (B)a 22 27 (C)a
54,500 ( 1300 27,600 ( 300
2.0 ( 0.3 2.3 ( 0.2
0 0
27 31 (D)
29,500 ( 350
2.2 ( 0.1
5.3 ( 1.6
31 37 (D)b
28,700 ( 1200
2.4 ( 0.3
15.9 ( 2.0
37 39 (E)b
29,600 ( 1300
2.3 ( 0.1
50.2 ( 6.3
a
Chemical cleaning of the RO subsystem performed between intervals B and C. b Feed flow reversal performed on FO spiral wound membrane for one hour prior to increasing TSS concentration during interval E. c Letter indices enumerated in the days of operation column are used to correlate operation intervals identified in Figure 2.
constantly dosed with activated sludge from the SBMBR for an additional 300 h (12 days) to study the effects of membrane fouling on water productivity and on rejection of nutrients and TOrCs. Activated sludge was dosed into the FO feed stream before the feed pump and a 50-mesh strainer. Operating conditions, including draw solution concentration, draw solution flow rate, and feed total suspended solids (TSS) concentration for specific time intervals during the pilot study are summarized in Table 3. Samples for TOrCs analysis were collected during intervals B through E. During 960 h (40 days) of continuous operation for
TOrCs rejection investigation the hybrid process produced approximately 7,700 L of RO product water and the FO membrane processed more than 650,000 L of wastewater effluent.
3. RESULTS AND DISCUSSION 3.1. Bench-Scale Performance: TOrCs Rejection. Benchscale FO tests were performed to investigate the effect of draw solution composition and organic matter on the rejection of TOrCs in a relatively simple matrix for ease of analysis. TOrC rejection during the three bench-scale experiments is summarized in Figure 1. Because several compounds (i.e., bisphenol A, diclofenac, benzophenone, and triclosan) were not quantified in one or more draw solution samples, rejection could not be accurately determined and were not included in Figure 1. Bench-scale FO rejection of TOrCs was between 40 and 98% and depended primarily on molecular size and charge. Rejection of positively and negatively charged TOrCs was greater than 80%, while the rejection of the nonionic TOrCs was more variable, between approximately 40 and 90%. With the exception of TCEP, the rejection of the nonionic TOrCs tended to increase with increasing molecular weight, which is to be expected based on hindered diffusion. The chlorinated flame retardant TCEP, however, exhibited lower rejection than expected based on size. Although TCEP was demonstrated to be highly rejected by RO and NF membranes,25 rejection of TCEP in our past FO bench scale study was consistently low.26 While ionic TOrC rejection was greater than 80%, rejection of these compounds was lower than expected 8486
dx.doi.org/10.1021/es201654k |Environ. Sci. Technol. 2011, 45, 8483–8490
Environmental Science & Technology
Figure 1. Rejection of select TOrCs by the FO membrane during the three bench-scale experiments. In the first experiment feed contained major ions and TOrCs and draw solution was seawater at 30 g/L. In the second experiment feed contained major ions, TOrCs, and humic acid, and draw solution was seawater at 30 g/L. In the third experiment feed contained major ions, TOrCs, and humic acid, and draw solution was NaCl at 30 g/L. Numbers in parentheses represent molecular weight of the compound. ** designates questionable results and rejection could not be calculated. Error bars represent standard deviation derived from three sampling events in each of the experiments.
Figure 2. Compiled water flux data from the pilot-scale system. Letters marked on the graph indicate intervals of interest during the pilot study and are defined in Table 3.
based on previous experiments demonstrating sodium chloride rejection of approximately 94%. This less than optimal rejection of ionic compounds could be explained by the relatively low water flux across the membrane (approximately 5 L/m2-h) and the relatively low surface charge of the cellulose acetate FO membrane (zeta potential lower than 10 mV at pH range of 5 to 9).
ARTICLE
The addition of humic acid to the feedwater when operating with seawater draw solution had minimal effect on the rejection of TOrCs by the FO membrane. Generally, the type of draw solution employed did not substantially affect rejection of TOrCs during bench-scale experiments. Methylparaben was the only compound that exhibited a substantial increase in rejection with the NaCl draw solution compared to the simulated seawater draw solution (oxybenzone exhibited a statistically insignificant increase as well); however, there is no apparent explanation for this behavior. DEET was detected at a higher concentration in the draw solution than in the feed in samples from the NaCl draw solution experiment, indicating that the sample may have been contaminated during processing for LC/MS-MS analysis. 3.2. Pilot-Scale Performance. 3.2.1. Water Flux. Water flux was continually monitored during the pilot study. Water flux as a function of operation time is shown in Figure 2. Letter indices represent operation intervals enumerated in Table 3. During interval A the pilot system was operated for approximately 250 h with SBMBR permeate as FO feed before reaching a steady-state water flux. This was likely due to interaction of the membrane with various constituents in the feed that changed the surface properties of the virgin FO membrane. Also, a film of organic and biological foulants accumulated on the FO membrane surface during this time interval and further contributed to the flux decline. Membrane cleaning in similar experiments was not able to restore the initial water flux of a new FO membrane.27 During the last 100 h of interval A the water flux was nearly constant. During interval B the draw solution concentration was doubled, and although the driving force for the process (i.e., the osmotic pressure of the draw solution) was doubled during this interval, water flux did not increase proportionally. This is attributed to internal concentration polarization phenomenon in FO.28 Water flux during this interval was almost constant, and additional flux decline from membrane fouling was not observed. Chemical cleaning of the RO subsystem and draw solution channels in the FO spiral-wound membrane was performed between interval B and C to remove mineral scale within the RO subsystem. During interval C, draw solution concentration was reduced to approximately 30 g/L sea salt. An equivalent water flux to the end of interval A was observed during this time period. During interval D activated sludge was dosed into the FO feed stream to achieve 5 and later 16 mg/L TSS, and a decrease in water flux was observed. During interval E the dosing rate of activated sludge was increased to achieve approximately 50 mg/L TSS in the FO feed, and additional water flux decline was observed. 3.2.2. Pilot-Scale Performance: TOrCs Rejection. Samples were drawn from the feed, draw solution, and product water of the RO pilot system during interval B through E and analyzed by the isotope dilution LC-MS/MS method. Isotope recoveries for TOrCs in each stream are summarized in the Supporting Information, section S4. All isotope recoveries exceeded minimum thresholds for accurate TOrC concentration quantification based on previous research by Vanderford and Snyder.24 Of the 23 compounds detected in the SBMBR permeate (FO feed), 14 are nonionic, four are negatively charged, and five are positively charged at solution pH between 6 and 7 (representative of the draw solution and feed pH, respectively). Nine TOrCs were analyzed for but were not detected in the samples. These include acetaminophen, atrazine, cimetidine, gemfibrizol, ketoprofen, propylparaben, meprobamate, 4-n-nonylphenol, and norfluoxetine. 8487
dx.doi.org/10.1021/es201654k |Environ. Sci. Technol. 2011, 45, 8483–8490
Environmental Science & Technology
ARTICLE
Figure 3. Rejection of TOrCs by the hybrid process. TOrCs are divided into classifications based on charge and hydrophobicity. Nonionic hydrophobic compounds are those with log D values greater than two. TOrC molecular weight is indicated in parentheses after the name of each compound. Operation intervals defined in Table 3 are shown above each column of results. (†) corresponds to TOrC concentrations measured below the detection limit for the instrument in the draw solution (red) or RO product water (blue) and rejection cannot be calculated. (X) corresponds to instances where TOrCs or their isotopes produced poor chromatographs and a nondetect of that TOrC is recorded in the draw solution (red) or RO product water (blue). (*) designates assumption of 100% rejection because the compound was detected only in the feed and not in the draw solution or RO product water. (cluster of four black diamonds) designates samples where TOrC concentration was measured in the feed and the RO product water but the draw solution sample registered a nondetect. Error bars represent standard deviation derived from four sampling campaigns during interval B, seven sampling campaigns during interval C, nine sampling campaigns during interval D, and three sampling campaigns during interval E.
Average feed concentration of TOrCs with their associated standard deviation is summarized in Table 2. Concentration of TOrCs detected in the feed varied over the 25 day sampling campaign following fluctuations in usage patterns (e.g., diurnal variation29) or wastewater treatment.25 TOrC concentrations varied substantially between different constituents. Dilantin, diphenhydramine, sulfamethoxazole, TCEP, and triclocarban were routinely detected at concentrations exceeding 100 ng/L. Recent studies demonstrated that these compounds are routinely detected in wastewater treatment plant effluents at concentrations significantly greater than LC-MS/MS quantification limits.30 Other TOrCs, such as amitriptyline, diazepam, and primidone, were detected at very low concentrations (i.e., below 10 ng/L) likely due to their low usage by the residents. Rejection of several TOrCs that were poorly quantified during bench-scale experiments (i.e., bisphenol A, diclofenac, benzophenone, and triclosan) was well characterized during the pilot-scale investigation. There was negligible difference in TOrC concentrations between SBMBR permeate and SBMBR permeate dosed with activated sludge. Rejection of TOrCs by the hybrid FO-RO process was calculated for all 23 compounds detected in the SBMBR permeate. Rejection of nonionic, nonionic hydrophobic, negatively
charged, and positively charged TOrCs by the hybrid process is summarized in Figure 3. TOrCs are listed in the order of increasing molecular weight. Three rejection values are calculated for each TOrC; these include rejection by the FO membrane, rejection by the RO membrane, and total rejection of the hybrid FO-RO process. Rejection measurements for each group of TOrCs are also divided into three discrete operation intervals, including intervals B, C, and combined D and E (Table 3). Across all TOrC groups, the RO membrane generally provided a similar rejection of contaminants to the FO membrane. However, a few exceptions were observed: the RO system provided significantly better rejection than the FO system for bisphenol A during intervals B and C and the FO system provided better rejection for methylparaben (all intervals), oxybenzone (intervals C, D, and E), amitriptyline (all intervals), and triclosan (Interval B). It is worth noting that because low feed concentrations were quantified for many compounds, and often near the detection limit of the LC-MS/ MS, rejection can be a poor metric for comparison. Also, reduced apparent rejection of constituents may be expected under conditions of low water flux;31 in the current study the RO flux was adjusted to produce permeate at the same flow rate of the FO module. 8488
dx.doi.org/10.1021/es201654k |Environ. Sci. Technol. 2011, 45, 8483–8490
Environmental Science & Technology During intervals C, D, and E, the rejection of nonionic compounds (Figure 3) by the RO membrane generally increased with increasing molecular weight in agreement with findings in previous research.8 Compounds exhibiting average RO rejection less than 90% during one or more operation intervals included methylparaben, caffeine, benzophenone, oxybenzone, bisphenol A, triclosan, and amitriptyline. Several of these compounds including methylparaben, bisphenol A, and triclosan have phenolic functional groups, which were demonstrated to have lower than expected rejection based on molecular size due to adsorption to and partitioning through membrane materials.32 34 Benzophenone has been quantified in RO permeate samples collected from water reuse facilities, which is likely due to hydrophobic interactions with membrane materials.35 Rejection of TOrCs during pilot-scale experiments was significantly greater than observed for bench-scale experiments under all conditions evaluated (Figures 1 and 3). While the exact reason for this disparity is unclear, membrane compaction, the establishment of a fouling layer, and optimized hydrodynamic conditions in the pilot-scale system may result in more efficient rejection. Furthermore, the FO membrane achieved substantial removal of all constituents except for caffeine during interval B and C. It is possible that the relative hydrophilicity of caffeine (log D equal to 0.13) and its relatively small molecular weight contribute to its ability to partition into the relatively hydrophilic FO membrane and to diffuse through the active layer. Hydrophobic nonionic compounds demonstrated lower correlation with molecular weight (e.g., benzophenone and triclosan rejection by RO, and bisphenol A rejection by FO); yet, with the exception of oxybenzone during interval B, high rejection of these constituents by the hybrid process was maintained (Figure 3). Bisphenol-A rejection by the FO membrane was consistently the lowest of all TOrCs among all groups. A relatively high hydrophobicity (log D equal to 3.43) and small molecular weight may partially explain the ability of bisphenol A to diffuse through the membrane. It is also possible for the phenolic structure of bisphenol A to more easily associate with the acetylated hydroxyl groups of the CTA membrane, thus enhancing its transport through the membrane by increasing its concentration at the membrane surface.33 Enhanced rejection of bisphenol A was observed during intervals D and E, when small doses of activated sludge were injected into the FO feed stream. Previous research19 suggests that a fouling layer on the membrane surface could separate and inhibit the interaction of the bisphenol A molecule with the membrane surface, thus increasing rejection. Rejection of charged TOrCs was generally high (Figure 3) and is in agreement with previous research.8,36,37 Negatively charged constituents may be rejected by electrostatic exclusion arising from the negative surface charge of the CTA and SW30 membranes. The mechanism for rejection of positively charged compounds is not completely understood; yet, results from previous studies36 39 indicate that constituent rejection in excess of 90% is possible. The dual barrier FO-RO process was able to achieve consistently high removal of TOrCs from the wastewater feed stream. Accumulation of TOrCs in the draw solution closed loop16 (Supporting Information, section S3) was observed; however, a mass balance of the water and TOrCs diffusing through the membrane between sampling intervals revealed that a high rejection of these constituents was maintained. TOrC accumulation in the draw solution also challenged the SWRO membrane with greater concentration of these constituents than would be
ARTICLE
expected in an open loop system (Supporting Information, section S1). Despite the shortcomings associated with operating the system with the draw solution in a closed loop, total system rejection of TOrCs generally exceeded 99%.
’ ASSOCIATED CONTENT
bS
Supporting Information. Details on the FO-RO dual barrier system for high efficiency water purification, novel spiral wound FO membrane modules, pilot-scale FO-RO system, and TOrCs analysis and rejection calculations. This material is available free of charge via the Internet at http://pubs.acs.org.
’ AUTHOR INFORMATION Corresponding Author
*Phone: (303) 273-3402. Fax: (303) 273-3413. E-mail: tcath@ mines.edu.
’ ACKNOWLEDGMENT The authors acknowledge the support of California Department of Water Resources (Grant 46-7446-R-08) and the AWWA Abel Wolman Fellowship. Special thanks to Aqua-Aerobic Systems, Inc. and to Hydration Technologies Innovations for providing membranes, systems, and Supporting Information for this study and to Monte Dipalma for his analytical support. ’ REFERENCES (1) Alexander, P.; Brekke, L.; Davis, G.; Gangopadhyay, S.; Grantz, K.; Hennig, C.; Jerla, C.; Llewellyn, D.; Miller, P.; Pruitt, T.; Raff, D.; Scott, T.; Tansey, M.; Turner, T. SECURE Water Act Section 9503(c) Reclamation Climate Change and Water 2011; US Bureau of Reclamation: Denver, CO, 2011. (2) Miller, G. W. Integrated concepts in water reuse: managing global water needs. Desalination 2006, 187 (1 3), 65–75. (3) Weikel, D., Sewage in O.C. goes full circle. Los Angeles Times January 2, 2008, January 2, 2008. (4) Meyer, J. P., Tapping used water. Denver Post January 23, 2007. (5) BUB NEWater Technology. http://www.pub.gov.sg/products/ NEWater/Pages/default.aspx (accessed month day, year). (6) Kolpin, D. W.; Furlong, E. T.; Meyer, M. T.; Thurman, E. M.; Zaugg, S. D.; Barber, L. B.; Buxton, H. T. Pharmaceuticals, hormones, other organic wastewater contaminants in U.S. streams, 1999 2000: a national reconnaissance. Environ. Sci. Technol. 2002, 36 (6), 1202–1221. (7) Snyder, S. A.; Leising, J.; Westerhoff, P.; Yoon, Y.; Mash, H.; Vanderford, B. Biological and physical attenuation of endocrine pisruptors and pharmaceuticals: Implications for water reuse. Ground Water Monit. Rem. 2004, 24 (2), 108–118. (8) Bellona, C.; Drewes, J. E.; Xu, P.; Amy, G. Factors affecting the rejection of organic solutes during NF/RO treatment - a literature review. Water Res. 2004, 38 (12), 2795–2809. (9) Drewes, J. E.; Heberer, T.; Rauch, T.; Reddersen, K. Fate of pharmaceuticals during ground water recharge. Ground Water Monit. Rem. 2003, 23 (3), 64–72. (10) Snyder, S. A.; Westerhoff, P.; Yoon, Y.; Sedlak, D. L. Pharmaceuticals, personal care products, and endocrine disruptors in water: Implications for the water industry. Environ. Eng. Sci. 2003, 20 (5), 449–469. (11) Westerhoff, P.; Yoon, Y.; Snyder, S.; Wert, E. Fate of endocrinedisruptor, pharmaceutical, and personal care product chemicals during simulated drinking water treatment processes. Environ. Sci. Technol. 2005, 39 (17), 6649–6663. (12) Xu, P.; Drewes, J. E.; Bellona, C.; Amy, G.; Kim, T.-U.; Adam, M.; Heberer, T. Rejection of emerging organic micropollutants in 8489
dx.doi.org/10.1021/es201654k |Environ. Sci. Technol. 2011, 45, 8483–8490
Environmental Science & Technology nanofiltration/reverse osmosis membrane applications. Water Environ. Res. 2005, 77 (1), 40–48. (13) Cath, T. Y.; Childress, A. E.; Elimelech, M. Forward osmosis: Principles, applications, and recent developments. J. Membr. Sci. 2006, 281 (1 2), 70–87. (14) Achilli, A.; Cath, T. Y.; Marchand, E. A.; Childress, A. E. The forward osmosis membrane bioreactor: A low fouling alternative to MBR processes. Desalination 2009, 239 (1 3), 10–21. (15) Cath, T. Y.; Gormly, S.; Beaudry, E. G.; Flynn, M. T.; Adams, V. D.; Childress, A. E. Membrane contactor processes for wastewater reclamation in space: Part I. Direct osmotic concentration as pretreatment for reverse osmosis. J. Membr. Sci. 2005, 257 (1 2), 85–98. (16) Cath, T. Y.; Hancock, N. T.; Lundin, C. D.; Hoppe-Jones, C.; Drewes, J. E. A multi barrier hybrid osmotic dilution process for simultaneous desalination and purification of impaired water. J. Membr. Sci. 2010, 362, 417–426. (17) Cartinella, J. L.; Cath, T. Y.; Flynn, M. T.; Miller, G. C.; Hunter, K. W.; Childress, A. E. Removal of natural steroid hormones from wastewater using membrane contactor processes. Environ. Sci. Technol. 2006, 40 (23), 7381–7386. (18) Nghiem, L. D.; Coleman, P. J. NF/RO filtration of the hydrophobic ionogenic compound triclosan: Transport mechanisms and the influence of membrane fouling. Sep. Purif. Technol. 2008, 62 (3), 709–716. (19) Nghiem, L. D.; Vogel, D.; Khan, S. Characterising humic acid fouling of nanofiltration membranes using bisphenol A as a molecular indicator. Water Res. 2008, 42 (15), 4049–4058. (20) Xu, P.; Drewes, J. E.; Kim, T.-U.; Bellona, C.; Amy, G. Effect of membrane fouling on transport of organic contaminants in NF/RO membrane applications. J. Membr. Sci. 2006, 279 (1 2), 165–175. (21) Bellona, C.; Marts, M.; Drewes, J. E. The effect of organic membrane fouling on the properties and rejection characteristics of nanofiltration membranes. Sep. Purif. Technol. 2010, 74 (1), 44–54. (22) DOW Water and Process Solutions, DOW FILMTEC SW302540.http://www.dowwaterandprocess.com/products/membranes/ sw30_2540.htm (accessed month day, year). (23) Hancock, N. T.; Cath, T. Y. Solute coupled diffusion in osmotically driven membrane processes. Environ. Sci. Technol. 2009, 43, 6769–6775. (24) Vanderford, B. J.; Snyder, S. A. Analysis of Pharmaceuticals in Water by Isotope Dilution Liquid Chromatography/Tandem Mass Spectrometry. Environ. Sci. Technol. 2006, 40 (23), 7312–7320. (25) Bellona, C.; Drewes, J. E.; Oelker, G.; Luna, J.; Filteau, G.; Amy, G. Comparing nanofiltration and reverse osmosis for drining water augmentation. J. Am. Water Works Assoc. 2008, 100 (9), 102–116. (26) Cath, T. Y.; Drewes, J. E.; Lundin, C. D. A novel hybrid forward osmosis process for drinking water augmentation using impaired water and saline water sources - final report (Project #4150); Water Research Foundation: Denver, CO, 2009. (27) Hancock, N. T. Engineered Osmosis: Assessment of Mass Transport and Sustainable Hybrid System Configurations for Desalination and Water Reclamation. Colordo School of Mines, Golden, CO, 2011. (28) McCutcheon, J. R.; Elimelech, M. Influence of concentrative and dilutive internal concentration polarization on flux behavior in forward osmosis. J. Membr. Sci. 2006, 284 (1 2), 237–247. (29) Crites, R.; Tchobanoglous, G. Small and Decentralized Wastewater Management Systems; McGraw-Hill: Columbus, 1998. (30) Dickenson, E. R. V.; Snyder, S. A.; Sedlak, D. L.; Drewes, J. E. Indicator compounds for assessment of wastewater effluent contributions to flow and water quality. Water Res. 2011, 45 (3), 1199–1212. (31) Mulder, M. Basic principles of membrane technology, 2nd ed.; Kluwer Academic Publishers: Dordrecht, The Netherlands, 1997; p 564. (32) Kimura, K.; Amy, G.; Drewes, J.; Watanabe, Y. Adsorption of hydrophobic compounds onto NF/RO membranes: an artifact leading to overestimation of rejection. J. Membr. Sci. 2003, 221 (1 2), 89–101. (33) Matsuura, T.; Sourirajan, S. Physicochemical criteria for reverse osmosis separation of alcohols, phenols, and monocarboxylic acid in aqueous solutions using porous cellulose acetate membranes. J. Appl. Polym. Sci. 1971, 15 (12), 2905–2927.
ARTICLE
(34) Williams, M. E.; Hestekin, J. A.; Smothers, C. N.; Bhattacharyya, D. Separation of organic pollutants by reverse osmosis and nanofiltration membranes: A mathematical models and experimental verification. Ind. Eng. Chem. Res. 1999, 38 (10), 3683–3695. (35) Drewes, J. E.; Sedlak, D. L.; Snyder, S.; Dickenson, E. R. V. Development of indicators and surrogates for chemical contaminant removal during wastewater treatment and reclamation (04-HHE-1CO); Water Environment Research Foundation: Alexandria, VA, 2009. (36) Verliefde, A. R. D.; Heijman, S. G. J.; Cornelissen, E. R.; Amy, G.; Van der Bruggen, B.; van Dijk, J. C. Influence of electrostatic interactions on the rejection with NF and assessment of the removal efficiency during NF/GAC treatment of pharmaceutically active compounds in surface water. Water Res. 2007, 41 (15), 3227–3240. (37) Van der Bruggen, B.; Schaep, J.; Wilms, D.; Vandecasteele, C. Influence of molecular size, polarity and charge on the retention of organic molecules by nanofiltration. J. Membr. Sci. 1999, 156 (1), 29–41. (38) Nghiem, L. D.; Sch€afer, A. I.; Elimelech, M. Role of electrostatic interactions in the retention of pharmaceutically active contaminants by a loose nanofiltration membrane. J. Membr. Sci. 2006, 286 (1-2), 52–59. (39) Nghiem, L. D.; Sch€afer, A. I.; Elimelech, M. Pharmaceutical retention mechanisms by nanofiltration membranes. Environ. Sci. Technol. 2005, 39 (19), 7698–7705.
8490
dx.doi.org/10.1021/es201654k |Environ. Sci. Technol. 2011, 45, 8483–8490
