Produced water treatment using forward osmosis membranes: Evaluation of extended-time performance and fouling

Author’s Accepted Manuscript Produced water treatment using forward osmosis membranes: evaluation of extended-time performance and fouling Elizabeth A. Bell, Taylor E. Poynor, Kathryn B. Newhart, Julia Regnery, Bryan Coday, Tzahi Y. Cath www.elsevier.com/locate/memsci

PII: DOI: Reference:

S0376-7388(16)32013-0 http://dx.doi.org/10.1016/j.memsci.2016.10.032 MEMSCI14818

To appear in: Journal of Membrane Science Received date: 29 December 2015 Revised date: 4 October 2016 Accepted date: 22 October 2016 Cite this article as: Elizabeth A. Bell, Taylor E. Poynor, Kathryn B. Newhart, Julia Regnery, Bryan Coday and Tzahi Y. Cath, Produced water treatment using forward osmosis membranes: evaluation of extended-time performance and f o u l i n g , Journal of Membrane Science, http://dx.doi.org/10.1016/j.memsci.2016.10.032 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting galley proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

Produced water treatment using forward osmosis membranes: evaluation of extended-time performance and fouling

Elizabeth A. Bell1, Taylor E. Poynor1, Kathryn B. Newhart1, Julia Regnery1, Bryan Coday1,2, Tzahi Y. Cath1* 1 Colorado School of Mines, Golden, CO 2 Carollo Engineers, Westminster, CO, USA * Corresponding author: Tel: (303) 273-3402; fax: (303) 273-3413, [email protected]

Abstract Forward osmosis (FO) membrane fouling and performance were systematically studied for extended-time during treatment of produced water using cellulose triacetate (CTA) and polyamide thin film composite (TFC) FO membranes. Performance was evaluated with integrity tests that measured water flux, reverse salt flux (RSF), and specific reverse salt flux (SRSF). The CTA membrane reached steady performance after one week and exhibited decreased water flux, RSF, and SRSF. The TFC membrane did not reach steady performance—water flux drastically decreased, and both RSF and SRSF increased. Streaming potential analyses was used to derive membrane zeta potential—the polyamide membrane zeta potential became increasingly negative over three weeks of continuous testing, while the CTA zeta potential was stable. The negative zeta potential reflected foulant deposition on the membrane surface and may have contributed to high RSF through the TFC membrane. The TFC membrane experienced a higher fouling propensity despite smoother, more hydrophilic, and more neutrally charged virgin membrane surfaces. Fouling layers on both membranes consisted of hydrocarbons, iron, and silica. Chemically enhanced osmotic backwashing was performed weekly, which removed calcium, sodium, and chloride from the membrane surface but only marginally improved water flux. Gas chromatographymass spectroscopy was used to measure hydrocarbon concentrations in the feed and draw solution. The results showed that while both membranes had over 90% rejection of neutral hydrophobic compounds, the TFC membrane exhibited a higher rejection of small organic molecules. Compounds with carbonyl functional groups were not well rejected compared to all other aliphatic and polycyclic aromatic hydrocarbons of the same molecular size, and CTA membrane had lower rejection of these compounds than the TFC membrane.

Keywords: forward osmosis; produced water; wastewater treatment; membrane fouling; organic rejection

1.

Introduction Forward osmosis (FO) is an emerging technology for separation and treatment of a broad

range of impaired waters, including produced water from oil and gas (O&G) exploration and production operations. Early studies have designated FO as an suitable technology for treatment of produced water [1-3], and a recent life cycle assessment demonstrated that FO was economically and environmentally favorable compared to the common practice of deep well injection [4]. FO membranes are semipermeable and use osmotic pressure differences to drive the diffusion of water from an impaired feed stream across the membrane to a higher salinity draw solution (DS) [5]. FO membranes readily reject dissolved constituents in produced water and provide high quality water for direct reuse in the upstream O&G industry or for feed into advanced separation processes such as reverse osmosis (RO) for reclamation and reuse in a broad range of applications [3, 6]. The technology is capable of recovering up to 85% of produced water and operating with feed total dissolved solid (TDS) concentrations as high as 150,000 mg/L [1, 3, 7]. The membranes are also expected to have a lower fouling propensity compared to pressure-driven membrane processes [8-10], which minimizes pretreatment requirements. Despite a lower fouling propensity, FO membrane fouling is still a major operational cost that decreases water production and can require frequent membrane cleaning [11]. Membrane fouling is typically characterized as biofouling, organic fouling, and inorganic fouling (i.e., scaling) [12]. Biofouling involves the attachment of planktonic microorganisms and subsequent growth of mature biofilms [13], and it has been identified in just one produced water fouling study [14]. Organic fouling is known to predominate and involves the sorption of organic compounds (e.g., hydrocarbons, chemical additives) onto the membrane surface [15]. Produced waters contain high levels of TDS that cause substantial inorganic fouling from the sorption or precipitation of inorganic solids onto or within the membrane [15]. Multivalent ions (e.g., calcium, barium, strontium, iron, silicon) are major ions found in produced water inorganic fouling [16, 17] and can also exacerbate organic fouling through ion-organic complexation [17]. Produced water treatment is a novel application for FO, and therefore, only few studies have systematically investigated FO membrane fouling and performance [16-24]. Many of these stud-

1

ies used real produced water found that organic and inorganic foulants were substantial. Inorganic fouling was determined through energy dispersive spectroscopy (EDS) and included ions such as calcium, magnesium, strontium, chlorine, iron, and barium [16, 17]. Coday et al. [17] used Fourier transform infrared (FTIR) spectroscopy to determine that the organic composition of the fouling layer consisted of calcium carbonate with residual oil. Biofouling was not identified in any of the FO produced water fouling studies. Some of the fouling studies compared cellulose triacetate (CTA) and polyamide thin film composite (TFC) membranes to determine important membrane properties that affected fouling propensity. TFC membranes had a higher fouling propensity than CTA membranes in three of the studies, which was attributed to high surface roughness [16, 18], high initial water flux [17], and strong hydrogen bonding capability [18]. Surface roughness is an often cited membrane property that causes high membrane fouling [16, 18, 25-27]; however, a recent produced water treatment study did not find a clear correlation between fouling propensity and surface roughness of FO membranes [17]. Instead, fouling propensity was attributed to high initial water flux, which convectively drags more foulants to the membrane surface and compacts the fouling layer [28-30]. An abundance of strong hydrogen bonding sites on TFC membrane surfaces has also been proposed as a contributor to membrane fouling by produced water [18]. TFC active layers have carboxyl groups (-COOH) that exhibit stronger adhesion forces than hydroxyl groups (-OH) [31], which are found in CTA active layers. Membrane cleaning is often used to mitigate fouling and restore water flux. Several produced water treatment studies have demonstrated effective cleaning strategies of FO membranes that include hydraulic cleans [3, 32], osmotic backwashing [3], and chemically enhanced osmotic backwashing [17]. Chun et al. [32] found that hydraulic cleaning effectively recovered water flux for CTA and TFC membranes after 48 hours of coal seam gas produced water treatment. Deionized water was recirculated on the feed (active layer) side of the membrane for one hour, which sufficiently removed foulants and fully restored water flux. Hickenbottom et al. [3] demonstrated that osmotic backwashing was more effective than hydraulic cleaning for CTA membranes treating drilling wastewater. In that study, drilling wastewater remained on the feed side of the membrane and deionized water replaced the DS, resulting in reverse diffusion of water across the membrane, removal of foulants, and full restoration of water flux at 30% of the hydraulic cleaning time. A recent study used both osmotic backwashing and chemically enhanced osmotic

2

backwashing (CEOB) for cleaning of CTA and TFC membranes fouled by produced water [17]. Osmotic backwashing was performed daily for 30 minutes over 48 hours of testing using produced water as feed solution (acting as DS) and recirculating deionized water in the DS flow channels. A separate set of tests investigated 30 minutes of daily CEOB over 96 hours using an 11 g/L chemical cleaning solution as the feed and recirculating deionized water in the DS channels. It was revealed that CEOB was more effective than osmotic backwashing, and that ethylenediamine–tetraacetic acid (EDTA) was an effective cleaning chemical. The study also found that cleaning efficacy was higher for smoother membrane surfaces. All three studies were conducted over less than one week with frequent membrane cleanings, which might not reflect field operations. This is exemplified in a one week FO pilot study that treated raw drilling wastewater and did not require cleaning [33]. The CTA membranes experienced only 18% decrease in water flux and recovered 85% of the water in drilling wastewater. Studies of FO membrane fouling by produced water have appropriately compared the widely used CTA membrane and emerging TFC membranes [16-18]. Two of the studies compared real and synthetic wastewater [16, 17], which is important for accurately characterizing membrane fouling and performance [16, 34]. However, the membranes were studied for less than four days, and fouling, performance, and cleaning efficacy may not reflect long-term field operation. Thus, the goals of this study were to expand on existing produced water fouling literature by comparing CTA and TFC membranes, operating membranes for longer periods of time, and using real produced water to elucidate the mechanisms of FO membrane fouling. The study compared the fouling and performance of CTA and TFC membranes treating produced water from the DenverJulesburg basin in Colorado continuously for three weeks. Water flux, reverse salt flux (RSF), specific reverse salt flux (SRSF), and organic compound rejection were compared for both FO membranes. Membrane autopsies were periodically performed throughout the study in order to characterize fouling layers and to investigate physiochemical changes to each membrane. Lastly, the study evaluated the efficacy of weekly CEOBs using EDTA solution.

2.

Materials and Methods

2.1.

Membranes A commercially available CTA membrane and a recently developed TFC membrane were

used in the study and were acquired from Hydration Technology Innovations (Albany, OR). It is

3

important to note that a surface-modified, antifouling TFC membrane was used in the current study as described in previous publications [17, 35]; the antifouling TFC membrane was used because a previous study found improved performance after produced water treatment with this membrane [35]. CTA and TFC membrane coupons were installed in custom-made flow cells with an effective membrane area of 190 cm2 each. The cells have asymmetric flow channels for the feed (25.4 x 7.62 x 0.24 cm) and DS (25.4 x 7.62 x 0.16 cm). The membranes were operated with the active layer facing the feed solution (no feed spacers) and tricot spacers in the DS channel to provide mixing and mechanical support for the membranes.

2.2.

Virgin membrane integrity tests Membrane integrity tests were conducted on new membrane coupons before each experiment

to ensure that the membranes were free of defects. During integrity tests water flux, RSF, and SRSF were measured using a bench scale system described in detail elsewhere [36-38]. The bench scale system incorporated a process control software (LabVIEW, National Instruments Corp., Austin, TX) and a data acquisition hardware (UE9-Pro, LabJack, Lockwood, CO) to collect data (i.e., feed solution volume, feed solution conductivity, DS conductivity) and to control experimental conditions (i.e., feed solution and DS dosing solution flow rates, 20 °C temperature). The feed reservoir contained 3 L of deionized water and was dosed with deionized water in order to maintain constant volume as water diffused through the FO membrane into the DS. The DS reservoir initially contained 1 L of 1 M NaCl solution (ACS grade, Fisher Scientific, Fair Lawn, NJ) and was intermittently dosed with a 300 g/L NaCl solution in order to maintain constant salinity and osmotic pressure. Membranes were operated for 90 minutes with co-current feed and DS flow rates of 2.1 and 1.3 L/min, respectively. Water flux (L/m2/hr) was defined as the forward diffusion of water from the feed solution to the DS (calculated from the volume of deionized water dosed into the feed reservoir) divided by membrane area and experimental time. A probe in the feed reservoir measured conductivity, and the concentration of NaCl was determined from the conductivity and used to calculate RSF (g/m2/hr). The division of water flux by RSF yields SRSF (mg/L), which is an important membrane parameter that directly relates to membrane efficiency [36, 39]. Lower SRSF values denote a more efficient and desirable membrane that preferentially allows for forward diffusion of water versus reverse diffusion of solutes.

4

2.3.

Membrane fouling system

2.3.1. System set-up The membrane fouling test system was operated with CTA and TFC membranes for up to three weeks, treating produced water. The system was comprised of 12 flow cells, a feed solution tank, two DS tanks, a DS dosing solution tank, a rotary vane feed pump (Procon, Smyrna, TN), two DS gear pumps (Micropump, Vancouver, WA), and two peristaltic dosing pumps (ColeParmer, Vernon Hills, IL) (Fig. 1). Virgin CTA and TFC membranes with a SRSF of less than 1,000 mg/L were installed in the system. The membranes operated in series within two groups based on membrane type (CTA and TFC). Each group contained six membrane cells in which three membranes were cleaned during the study (“CTA-C” and “TFC-C”) and three remained undisturbed, or “fouled” (“CTA-F” and “TFC-F”). CTA-C and TFC-C membranes were cleaned once per week for three weeks. Cleaned and fouled membranes operated for one, two, or three weeks and were taken offline for integrity testing and membrane characterization. A picture and a schematic diagram of the membrane fouling system are shown in Fig. 1.

CTA-C3

CTA-F3

CTA-C2

CTA-F2

CTA-C1

(b) CTA-F1

(a)

CTA DS

Dosing Sol.

TFC-C3

TFC-F3

TFC-C2

TFC-F2

TFC-C1

TFC-F1

Feed

TFC DS

Fig. 1. (a) a picture and (b) a schematic diagram of the membrane fouling test system. Cells were arranged in two groups based on membrane type (CTA and TFC), and were operated in series. The feed tank contained produced water from the Denver-Julesburg basin, and each membrane group had a dedicated DS tank. Membranes were labeled with ‘F’ to indicate no cleaning and ‘C’ to indicate cleaning. Numbers 1, 2, and 3 represent the number of weeks the membranes were online. Membrane cleaning was performed using CEBW according to methods from a previous FO membrane fouling study [17]. The cleaning procedure used an 11 g/L EDTA solution. Separate 5

EDTA solutions were pH adjusted with citric acid to 7.8 and 10.8 for CTA and TFC membranes, respectively. The EDTA solutions were recirculated on the feed side of the membranes and deionized water recirculated on the support layer side of the membranes. Water osmotically permeated from the deionized water to the EDTA solution and removed foulants from the membrane active layer. Membranes were cleaned on the bench scale system for two hours with feed and DS flow rates similar to those used in the membrane integrity tests. CTA membranes were cleaned separately from TFC membranes and membranes were cleaned in series within each group.

2.3.2. System operating procedure Produced water from the Denver-Julesburg basin in Colorado was used as feed solution and was operated in batch mode. Produced water feed solution batch was replaced twice per week; thus, six batches were used over the experimental period. On a daily basis, deionized water was added to the feed solution tank to replace permeated water and to maintain consistent osmotic pressure of the feed stream. The DS was 1 M NaCl solution, and each membrane group used a separate DS tank. Each DS tank operated continuously and was drained on a daily basis down to 10 L (~3 gallons). A 300 g/L NaCl solution was intermittently dosed to each DS tank in order to maintain constant DS concentration (osmotic pressure). The feed solution and DS tanks were measured twice per day for pH, temperature, and conductivity using a handheld probe (Oakton Instruments, Vernon Hills, IL). The feed solution and DS operated at co-current flow rates of 1.58 and 0.82 L/min, respectively.

2.3.3. Analytical procedures The feed solution and DS tanks were sampled twice per week in order to determine inorganic and organic chemistry. Feed and DS samples were collected at the beginning and end of each feed solution batch. Feed solution samples were filtered with a 0.45 μm filter for all analyses, and feed solution and DS samples were diluted with deionized water per individual analysis specifications. Anions were measured with an ion chromatograph (IC) (ICS-90, Dionex, Sunnyvale, CA), and cations were measured using inductively coupled plasma atomic emission spectroscopy (ICP-AES) (Optima 5400, PerkinElmer, Fremont, CA). Both analyses required a 30x dilution to prevent chloride interferences, and ICP-AES samples were acidified with nitric acid to a pH of 2. Dissolved organic carbon (DOC) and total nitrogen (TN) were measured using

6

a carbon analyzer (Shimadzu TOC-L, Columbia, MD), in which feed solution and DS samples were diluted 3x and 5x, respectively. Average concentrations of constituents in the feed solution are summarized in Table 1.

Table 1. Average concentrations of major constituents in the feed solution Analyte pH Total suspended solids Total dissolved carbon Total nitrogen Chemical oxygen demand Total dissolved solids Boron Barium Calcium Iron Potassium Lithium Magnesium Manganese Sodium Sulfur Silicon Strontium Fluoride Chloride Bromine Nitrate Sulfate ΣCations ΣAnions

Unit -mg/L mg/L mg/L mg/L mg/L mg/L mg/L mg/L mg/L mg/L mg/L mg/L mg/L mg/L mg/L mg/L mg/L mg/L mg/L mg/L mg/L mg/L meq/L meq/L

Average 6.61 212 582 38.2 2,320 22,610 32.9 19.3 682 72.2 52.7 10.6 79.9 1.32 11,064 15.7 58.7 102 22.8 19,912 259 171 258 548 -571

Three-dimensional excitation emission matrix (3D EEM) fluorescence spectroscopy was used to qualitatively analyze fluorescent organic species in the feed solution, CTA DS, and TFC DS using a spectrofluorometer (AquaLog, HORIBA Scientific, Edison, NJ). The detailed analytical procedure for 3D-fluorescence measurements using the AquaLog spectrofluorometer is described elsewhere [40]. Samples were diluted to 2 mg/L DOC to avoid quenching effects, and were analyzed with excitation wavelengths between 238.5 and 450 nm and emission wavelengths between 250 and 600 nm. Organic species were identified and quantified in the feed solution, CTA DS, and TFC DS using gas chromatography-mass spectrometry (GC-MS). GC-MS was used to quantify semi-

7

volatile linear aliphatic hydrocarbons in the n-C10 to n-C32 range and 16 polycyclic aromatic hydrocarbons (PAHs) according to the methods from Regnery et al. [41]. Samples were pretreated with solid-phase extraction (SPE) on C18 cartridges using an AutoTrace 280 SPE unit (Thermo Scientific, Waltham, MA) and analyzed with a HP 6890 gas chromatograph equipped with a HP 5973 mass spectrometer from Agilent Technologies (Palo Alto, CA) that used an Rtx-5Sil capillary column (Restek, Bellefonte, PA). Analytes were identified using mass spectra (Wiley AccessPak spectral library) and retention times of internal standards. Analyte concentrations were quantified by measuring peak areas of each compound relative to the respective internal standard peak area.

2.4.

Fouled membrane integrity tests After one, two, and three weeks, cleaned and fouled membranes were installed in the bench

scale system for integrity tests. Membranes were removed from the flow cells, and coupons were cut from the membranes and installed into a smaller, symmetric, custom-made flow cell (21.91 x 5.40 x 0.32 cm channel dimensions) with an effective membrane area of 120 cm2. The integrity tests were conducted under the same experimental conditions (i.e., temperature, flow rates, deionized water as feed solution, 1 M NaCl DS) as the virgin membrane integrity tests and measured water flux, RSF, and SRSF for all cleaned and fouled membranes.

2.5.

Membrane autopsy and characterization

2.5.1. Environmental scanning electron microscopy and energy dispersive spectroscopy An environmental scanning electron microscope (ESEM) (Quanta 600i, FEI, Tokyo, Japan) was used to examine fouling layers for cleaned and fouled membranes. Sample coupons were collected from cleaned and fouled membranes, dried in a desiccator, and gold sputtered prior to analysis (Technic Hummer VI, Cranston, RI). An energy dispersive x-ray spectrometer (EDS) (Genesis, EDAX, Mahwah, NJ) was used in conjunction with the ESEM in order to identify and compare the elemental content on the surface of cleaned and fouled membranes.

2.5.2. Fourier transform infrared spectroscopy Attenuated total reflection-Fourier transform infrared (ATR-FTIR) spectroscopy (Nicolet 4700 FTIR, Thermo Electron Corporation, Madison, WI) was used to determine the organic

8

composition of fouling layers on cleaned and fouled membranes. Membrane coupons were sampled from virgin, cleaned, and fouled membranes and dried in a desiccator for at least 24 hours before analysis.

2.5.3. Streaming potential analyses Streaming potential was measured for virgin, fouled, and cleaned membranes in order to determine changes in membrane zeta potential over the fouling period. The analysis used an electrokinetic analyzer equipped with an integrated titration unit and adjustable gap cell (SurPASS, Anton Paar GmbH, Austria) that measured streaming zeta potential at a pH of 7 in a 2 mM KCl solution. Membrane coupons were sampled from virgin, fouled, and cleaned membranes and dried in a desiccator for at least 24 hours before analysis.

2.5.4. Contact angle via captive bubble Membrane hydrophobicity was measured for virgin CTA and TFC membranes using contact angle via the captive bubble method. Membrane coupons were immersed in deionized water and a 5 μL air bubble was pipetted onto the active layer surface. Contact angles were measured using a digital goniometer (Model 200-00, Rame-Hart Instrument Company, Netcong, NJ). Contact angles measurements were averaged for three bubbles per coupon. Each bubble had three sets of ten measurements, and each measurement set was performed in 0.1 seconds.

3.

Results and Discussion

3.1.

Virgin membrane properties Water flux, RSF, SRSF, zeta potential, contact angle, and surface roughness are summa-

rized in Table 2. Virgin TFC membranes had higher water flux, lower RSF, and lower SRSF than virgin CTA membranes, which is in agreement with previous publications [17, 42]. Lower SRSF indicates that the TFC membranes were more selective and allowed less permeation of solutes. The TFC membranes are also smoother, more hydrophilic, and more neutrally charged; these three properties are regarded as important properties to reduce membrane fouling [12, 43, 44].

9

Table 2. Virgin CTA and TFC membrane properties that include water flux, reverse salt flux (RSF), specific reverse salt flux (SRSF), contact angle, and surface roughness Water Flux

RSF

SRSF

Zeta potential

Contact angle

Roughness (Ra/Rq)c nm 23.4/29.5b 10.4/13.6b

L/m2/hr g/m2/hr mg/L (mV)a º CTA 7.9±0.3 6.7±0.8 848±106 -33.5 45.1±1.4 TFC 11.0±1.2 2.5±0.4 233±44 -18.5 16.1±1.7 a Measured at pH 7 with 2 mM KCl solution. b Values from previous publication [17]. c Surface roughness was analyzed with atomic force microscopy and results are reported as average roughness (Ra) and root mean square roughness (Rq).

3.2.

Membrane performance: water flux, RSF, and SRSF Water flux, RSF, and SRSF as a function of time are shown in Fig. 2 for fouled and cleaned

CTA and TFC membranes. Water flux through both fouled and cleaned CTA membranes decreased by 50% and reached steady state flux after one week. RSF through the CTA membrane also reached steady state after one week, and decreased by 71% and 81% for fouled and cleaned CTA membranes, respectively. The water flux and RSF resulted in improved SRSF for the CTA membranes after three weeks. Water flux through the TFC membranes continually declined for three weeks and decreased by 91% and 85% for fouled and cleaned membranes, respectively. RSF increased by ~310% and ~325% for fouled and cleaned TFC membranes, respectively. Low water flux and high RSF resulted in drastically higher SRSF for the TFC membranes.

14

CTA-C TFC-C

(a)

(b) Reverse salt flux (g/m2/hr)

Water flux (L/m2/hr)

12

14

CTA-F TFC-F

10 8 6 4 2 0

12

CTA-C TFC-C

CTA-F TFC-F

10 8 6 4 2 0

0

7

Days

14

21

0

7

Days

14

21

10

Specific reverse salt flux (mg/L)

14000 (c) 12000

CTA-C TFC-C

CTA-F TFC-F

10000 8000 6000 4000 2000 0 0

7

Days

14

21

Fig. 2. (a) Water flux, (b) reverse salt flux (RSF), and (c) specific reverse salt flux (SRSF) results for cleaned CTA (“CTA-C”), fouled CTA (“CTA-F”), cleaned TFC (“TFC-C”), and fouled TFC (“TFC-F). Overall, the CTA membranes outperformed the TFC membranes with higher water flux, lower RSF, and lower SRSF after three weeks of testing. Visual inspection of the membranes revealed that the TFC membranes were more severely fouled than the CTA membranes (Fig. 3). High TFC fouling propensity was supported by the drastic water flux decline compared to the CTA membranes. Cleaned CTA and TFC membranes had only slightly higher water flux than fouled membranes, which demonstrated marginal foulant removal with CEOBs using EDTA. The finding contradicted a previous study that found significant improvements in water flux for CTA and TFC membranes cleaned with EDTA CEOB [17]. The discrepancy in cleaning efficacy may be because the present study had less frequent CEOBs, a longer fouling period and subsequent irreversible fouling, and lower cross-flow velocities during the fouling periods. The results also suggest that the CTA membranes are robust and can maintain steady water flux and RSF over three weeks. Long-term, sustained CTA water flux is in agreement with a previous pilot study that treated drilling wastewater for one week without CTA membrane cleaning [45].

11

Fig. 3. (a) CTA and (b) TFC membranes after three weeks of produced water treatment. The TFC membranes followed typical performance trends of fouled FO membranes, in which water flux decreased and RSF increased [46, 47]. Water flux decline was likely due to cake enhanced concentration polarization (CECP), which is cited as the major cause of water flux decline in osmotically driven membrane processes [8, 48-52]. CECP is a result of solute accumulation in the fouling layer and causes increased osmotic pressure at the membrane surface, decreased osmotic difference across the membrane, and decreased water flux. High RSF through the TFC membranes likely led to increased CECP, and high initial water flux may have compacted the fouling layer, which is known to worsen CECP [28]. It is also likely that the hydrogel coating layer on the surface-modified TFC membrane degraded over time, leaving the polyamide active layer more vulnerable to severe fouling. However, fouling layer effects on RSF were unclear and studies have yet to determine the cause of RSF changes associated with FO fouling. Several studies have proposed that negatively charged fouling layers induce a charge imbalance across the membrane [46, 47]. They hypothesized that the charge imbalance draws cations from the DS and increases RSF. In the present study, streaming potential was measured on virgin, fouled, and cleaned CTA and TFC membranes in order to explore this concept. Streaming potential measurements were performed at pH 7 and calculated zeta potentials are shown as a function of experimental elapsed time in Fig. 4.

12

Zeta Potential at pH 7 (mV)

0

CTA-C TFC-C

-5 -10

CTA-F TFC-F

-15 -20 -25 -30 -35 -40 -45 -50 0

7

14

21

Days

Fig. 4. Zeta potential values at pH 7 for CTA-C, CTA-F, TFC-C, and TFC-F membranes over 21 days of continuous experiments. Virgin CTA membranes had zeta potential values of –33.5 mV, and slightly decreased to – 36.2 and –38.4 mV for CTA-C and CTA-F membranes, respectively. Virgin TFC membranes had zeta potential values of –18.5 mV, and decreased to –39.0 and –43.0 mV for TFC-C and TFC-F membranes, respectively. The drastic decrease in zeta potential of the TFC membranes reflected their high fouling propensity; conversely, CTA membranes were less prone to fouling and were only slightly more negative after three weeks. Continually decreasing zeta potential for TFC membranes corresponds to increasing RSF and indicates that more negatively charged fouling layers possibly caused the drastic increase in RSF. However, systematic studies are still needed to elucidate the relationship between negatively charged fouling layers and RSF. The unique surface chemistry of the CTA and TFC active layers could also play an important role in the diffusion of solutes across the membrane. Traditional polyamide TFC membranes contain ionizable functional groups (i.e., amides (–NH–CO–), carboxylic acids (–COOH)) that have a fixed negative charge [53, 54], and CTA membranes contain non-ionizable functional groups (e.g., acetate (–COCH3), hydroxyl (–OH)) that do not possess a fixed charge [53, 55]. It is important to note that the TFC membrane used in this study was coated with a proprietary hydrogel by HTI in attempts to increase its anti-fouling properties and overall robustness; this new membrane has consistently exhibited minimal zeta potential (–10 to –20 mV) here and in previous studies where the membrane was first introduced. Also note that although CTA does not the-

13

oretically poses a fixed charge, it has been previously demonstrated that anions will preferentially aggregate near the surface of CTA membranes, resulting in a slightly negative zeta potential [56]. A recent study illustrated that TFC membranes had higher forward and reverse cation diffusion than CTA membranes, and the authors attributed the high cation diffusion to carboxyl groups and Donnan dialysis [56-58]. Additionally, forward and reverse diffusion of cations across TFC membranes were highly influenced by changes in the feed solution chemistry (i.e., electrolytic charge imbalances, pH), and solute fluxes across CTA membranes were not affected by changes in solution chemistry. The study provided evidence that TFC membrane solute flux was more sensitive to charge imbalances than CTA membrane solute flux. Fouled CTA and TFC membranes in the present study reached similar zeta potential, and TFC membrane RSF was possibly more influenced by fouling layer charge imbalances than CTA membrane RSF. Both water flux and RSF decreased for the CTA membranes. A previous study found slightly decreased water flux and RSF for CTA membranes after 48 hours of produced water treatment [35], and longer experimental periods would have likely yielded similar results to the present study. The cause for decline in water and solute diffusion was unclear, and the authors proposed a number of explanations. First, foulant adsorption and entrapment within the polymer matrix could have hindered water and solute diffusion [59-61], and the study suggested that the hydrophobic CTA membranes could have experienced higher organic adsorption compared to the more hydrophilic TFC membranes. Second, high ionic strength water may potentially deswell the active layer, decrease pore size, and subsequently lower water flux and RSF [62]. Ions may also neutralize pore charge, which would minimize electrostatic repulsion within the pores, decrease pore size, and decrease water and solute diffusion [63]. Overall, the interactions between produced water constituents and membranes are complex and any combination of them could have led to the observed changes in CTA and TFC membrane performance.

3.2.1. Membrane property effects on fouling Conventional fouling literature states that smooth, neutrally charged, and hydrophilic membranes are needed to optimize water flux and minimize fouling propensity [12, 43, 44]. According to these parameters, virgin TFC membranes had superior properties to CTA membranes (Table 1). However, TFC membranes had a higher fouling propensity than CTA membranes despite these virgin membrane properties.

14

Membrane heterogeneity provides a possible explanation for high TFC membrane fouling propensity. Conventional membrane characterization techniques only provide average surface values and do not reflect the heterogeneous distribution of properties [26, 31, 64, 65]. Several studies have shown a higher fouling potential correlation with heterogeneous property distribution versus average property values [26, 31, 65]. For example, Mi et al. [10] related fouling potential to heterogeneous adhesion forces for polyamide and cellulose acetate membranes. Polyamide membranes had lower average adhesion forces than cellulose acetate membranes, but had a heterogeneous distribution of very adhesive sites that resulted in a higher fouling propensity. Adhesion forces are a function of surface chemistry (e.g., functional groups, charge) and morphology (e.g., roughness) [31]; thus, virgin membrane properties may not accurately reflect the heterogeneous distribution of polyamide TFC membrane properties that led to higher fouling propensity.

3.2.2. Fouling layer characterization ESEM micrographs are shown in Fig. 5 for CTA-C, CTA-F, TFC-C, and TFC-F membranes after three weeks. All fouling layers were similar in appearance, but the foulant composition (e.g., organic, inorganic, biological) was not evident. Previous produced water membrane fouling studies showed that oil [17, 38], microorganisms [14], or inorganic scaling [16, 17, 38] predominated fouling layers, and these constituents likely comprised CTA and TFC fouling layers. The elemental compositions of the fouling layers were examined with EDS, and results for cleaned and fouled membranes are summarized in Table 3 as elemental weight percent (%). Iron and silicon were dominant elements in all the fouling layers, which was in agreement with previous produced water fouling studies [17, 66]. It is interesting to note that sodium and chloride were major elements only in the TFC-F fouling layers. The high weight percent of sodium and chloride may reflect the high RSF through the TFC membranes, and the lack of sodium and chloride in the TFC-C membranes indicate that EDTA may have removed the ions. Calcium was found only on CTA-F and TFC-F fouling layers and indicates that EDTA also removed calcium from the CTA-C and TFC-C fouling layers.

15

(a) CTA-F – Week 3

(b) CTA-C – Week 3

50 µm

(c) TFC-F – Week 3

50 µm

(d) TFC-C – Week 3

50 µm

50 µm

Fig. 5. ESEM micrographs at 1000x magnification of the (a) CTA-F, (b) CTA-C, (c) TFC-F, and the (d) TFC-C membranes after three weeks of continuous fouling experiments. Table 3. Energy dispersive spectroscopy (EDS) results CTA-C, CTA-F, TFC-C, and TFC-F fouling layers after three weeks of produced water treatment. Results are shown as % weight for each element. Element C O Na Al Si Au Cl Ca Fe Co

CTA-C 6.1 35.2 2.3 -9.2 4.0 --43.2 --

% Weight CTA-F TFC-C 6.0 3.3 34.1 48.8 -3.9 -0.5 10.0 12.6 3.4 5.3 --1.8 -44.8 25.4 -1.1

TFC-F 5.4 34.2 23.2 0.4 5.5 3.5 19.1 0.6 10.3 --

16

FTIR was used to determine the organic compositions of cleaned and fouled CTA and TFC fouling layers. Spectra for virgin, fouled, and cleaned CTA and TFC membranes are shown in Fig. 6 and summarized in Table 4. Virgin membrane peaks disappeared or were weakened in the presence of fouling layers. Foulant peaks for cleaned and fouled CTA and TFC membranes were similar and indicated that the fouling layers were the same. Peaks were representative of aliphatic hydrocarbons, aromatic hydrocarbons, and carbonates (Table 4) [67, 68]. The foulants were likely from organic matter in the formation and fracturing fluids [15]. A previous study used FTIR to examine membranes fouled with produced water and found evidence of biofouling [14]. Biofoulant peaks were not found in the present study, but it is possible that hydrocarbon peaks obscured biofoulant peaks.

(a)

(b)

Fig. 6. FTIR spectra comparison of virgin, cleaned, and fouled (a) CTA and (b) TFC membranes. Dotted lines indicate peaks associated with fouling layer. Table 4. Peak wavenumbers (cm–1), assignments, and interpretations for CTA and TFC membrane fouling layers. Peak assignments and interpretations were from Colthup et al. [67] and Matrajt et al. [68]. CTA fouling layer peaks (cm–1) 2954 2923 2853

TFC fouling layer peaks (cm–1) 2954 2923 2853

Peak assignment

Possible interpretations

Asymmetric -CH3 stretch Asymmetric -CH2 stretch Symmetric -CH2 stretch

Aliphatic hydrocarbon Aliphatic hydrocarbon Aliphatic hydrocarbon

17

CTA fouling layer peaks (cm–1) 1621 -1379 923

3.3.

TFC fouling layer peaks (cm–1) 1625 1458 1379 930

Peak assignment C=C stretch Asymmetric -CH2 bend Symmetric -CH3 bend Out-of-plane -CH2 bend

Possible interpretations Aromatic/carbonate Aliphatic/carbonate Aliphatic hydrocarbon Aromatics

Organic compound rejection

3.3.1. 3D EEM results The organic composition of the feed solution, CTA DS, and TFC DS were evaluated with 3D EEM in order to qualitatively describe organic rejection. The feed solution samples were taken from a fresh batch of produced water, and the DS samples were taken on day 21 of experiment. 3D EEM plots are shown for the feed solution, CTA DS, and TFC DS in Fig. 7. The feed solution had a peak located at excitation/emission wavelengths (Ex/Em) 270/300 nm and another region near Ex/Em 250/360 nm. CTA and TFC DS had similar 3D EEM plots with maximum intensities at Ex/Em 270/290 nm. The Ex/Em peak in the CTA and TFC DS was consistent with the peak in the feed solution, but at much lower intensity. The DS peak location and intensity indicated that the corresponding fluorescent compounds diffused through both membranes from the feed solution and were present at low concentrations in the DS. Fluorescent compounds in produced water were either aromatic compounds with two or more benzene rings [69], natural organic matter (e.g., humic and fulvic acids) [17, 25], or microbial proteins (e.g., tryptophan) [25]. Riecker et al. [69] identified polycyclic aromatic hydrocarbons (PAHs) and heterocyclic hydrocarbons as the major fluorescent compounds in crude oil throughout the Rocky Mountain region. A number of studies have determined fluorescent spectra for PAHs [25, 70, 71], and it is possible that PAHs contributed to the CTA and TFC DS peak. For example, fluorene (Ex/Em 260-280/305-311 [25, 70, 71]) matched closely with the peaks found in all three solutions; however, specific PAHs could not be determined because PAH standards were not analyzed in this study. Other studies have proposed that natural organic matter and microbial proteins were fluorescent compounds in produced water [17, 25]. Reported values for marine humic-like acids (Ex/Em 290-310/370-410), visible humic-like acids (Ex/Em 320360/420-460), soil fulvic-like acids (Ex/Em 521/455), and proteins (Ex/Em 240/324 and 280/324) were higher than the peaks found in the DS results but within fluorescent regions of the

18

feed solution [72, 73]. Thus, the membranes likely rejected natural organic matter and microbes, but allowed for the diffusion of PAHs.

Fig. 7. Three dimensional excitation-emission matrix (3D EEM) plots for (a) feed solution, (b) CTA DS, and (b) TFC DS. DS results represent samples taken on day 21, and feed solution results represent samples from a new produced water batch. 3.3.2. GC-MS results The organic compounds in the feed solution, CTA DS, and TFC DS were further analyzed with GC-MS in order to identify specific compounds, quantify concentrations, and calculate membrane rejection. Samples were taken from the feed solution, CTA DS, and TFC DS tanks. Feed solution samples were taken from new batches of produced water and DS was sampled on day 21. Table 5 summarizes compound names, classes (i.e., aliphatic hydrocarbons, PAHs), functional groups (i.e., alkyl, carbonyl), molecular weights (MW, g/mol), minimum projection areas (MPA, Å2), log D, pKa, and feed solution concentrations (μg/L). The aliphatic hydrocarbons are categorized as unsaturated or saturated and designated with a “U” or “S”, respectively. 19

Compound rejection is shown in Figure 8, in which compounds are organized by class and functional group and sorted by increasing molecular weight within each classification. The feed solution contained higher concentrations of aliphatic hydrocarbons compared to aromatic hydrocarbons (Table 5). Saturated aliphatics, unsaturated aliphatics with alkyl groups (pristane and phytane), 1-Octadecene, and PAHs had high rejection rates that ranged from 98100% for both CTA and TFC membranes. Unsaturated aliphatic compounds that contained carbonyl functional groups had lower rejection that ranged between 92-95% and 95-99% for CTA and TFC membranes, respectively.

Table 5. Organic compounds detected with GC-MS and associated class, functional group, molecular weight (MW, g/mol), minimum projection area (MPA, Å2), log-D, pKa, and feed solution concentration (μg/L). MW MPA Log Da,b pKaa (g/mol) (Å2)a Dodecanal Aliphatic (U) 184.3 23.3 4.40 17.8 Tridecanal Aliphatic (U) 198.3 30.5 4.85 17.8 Tetradecanal Aliphatic (U) 212.4 30.8 5.30 17.8 Pentadecanal Aliphatic (U) 226.4 30.5 5.76 17.8 Hexadecanal Aliphatic (U) 240.4 37.5 6.21 17.8 Octadecanal Aliphatic (U) 268.5 40.8 7.11 17.8 2-Octadecanone Aliphatic (U) 268.5 28.9 7.11 19.6 Pristane Aliphatic (U) 268.5 65.8 8.36 -Phytane Aliphatic (U) 282.6 57.0 8.81 -1-Octadecene Aliphatic (U) 252.5 31.7 8.27 -142.3- 28.35.43Sum C10-C33 n-alkanes Aliphatic (S) --464.9 62.7 15.34 Naphthalene PAH -128.2 24.0 2.98 -Fluorene PAH -166.2 24.6 3.75 17.8 Phenanthrene PAH -178.2 26.9 3.96 -Pyrene PAH -202.3 30.3 4.28 -Chrysene PAH -228.3 31.8 4.95 -Benz(a)anthracene PAH -228.3 27.9 4.95 -Benzo(a)pyrene PAH -252.3 32.0 5.27 -Benzo(b)fluoranthene PAH -252.3 34.3 5.27 -Benzo(k)fluoranthene PAH -252.3 33.6 5.27 -Benzo(ghi)perylene PAH -276.3 35.3 5.59 -Indeno(1,2,3-c,d)pyrene PAH -276.3 34.6 5.59 -Dibenz(a,h)anthracene PAH -278.4 32.1 5.93 -a Values were obtained from Marvin Beans 15.10.19 software (https://www.chemaxon.com). b Log-D values are log-Kow calculated at pH 8. c Feed solution concentration is the average of samples from four batches of feed solution. Compound

Class

Functional group Carbonyl Carbonyl Carbonyl Carbonyl Carbonyl Carbonyl Carbonyl Alkyl Alkyl --

Feed solutionc (μg/L) 28.8±50 56.3±378 61.8±55 45.3±37 165±208 94±36 215±185 95.2±17 89.6±15 46.3±11 2221±272 24.7±15 2.13±1.4 2.61±1.3 0.49±0.3 1.26±0.4 0.82±0.3 1.06±0.4 1.00±0.4 1.07±0.3 0.77±0.3 0.76±0.2 0.67±0.2

20

100%

Rejection (%)

98% 96% 94% 92%

TFC membrane CTA membrane

90%

Fig. 8. Rejection of organic compounds (%) detected with GC-MS. Results are shown for CTA and TFC membranes, and compounds are organized by class and functional group and sorted by increasing molecular weight within each classification: unsaturated aliphatic with carbonyl group (purple), unsaturated aliphatic with alkyl group (green), unsaturated aliphatic with no functional group (red), saturate aliphatic (orange), and polycyclic aromatic hydrocarbons (blue). Rejection values were calculated from day 21 DS and average feed solution concentrations (n = 4). Organic compound rejection is complex and involves compound properties, membrane properties, and compound-membrane interactions [63, 74-76]. Studies typically relate organic compound rejection to their hydrophobicity, charge (e.g., negative, positive, neutral), and molecular weight [63, 74, 77]. Compounds are considered hydrophobic when the octanol-water distribution coefficient (log-D) is greater than 2 and neutral when pKa values are higher than solution pH [63, 78, 79]. Therefore, all compounds were hydrophobic because log-D values were higher than 2 (Table 5) [78, 80], and all compounds were neutral because their pKa values were nonexistent or above the feed solution pH (Table 5). It is widely accepted that neutral hydrophobic compound rejection is dictated by steric hindrance and improves with increasing molecular weight [63, 74]. Fujioka et al. [78, 80] expanded on this concept and found that rejection correlated best with molecular size rather than molecular weight. The authors measured molecular size with MPA (Å2), which is the smallest twodimensional projection area for a three dimensional molecule. Thus, MPA was used in order to

21

explore the effects of molecular size on neutral hydrophobic compound rejection. Compound

100%

100%

98%

98% Rejection (%)

Rejection (%)

rejection is shown as a function of MPA in Fig. 9 for TFC and CTA membranes.

96%

94%

92% (a) CTA

90% 20

Aliphatic (U) - carbonyl Aliphatic (U) - alkyl Aliphatic (U) Aliphatic (S) PAH

40 60 80 Minimum projection area (Å2)

96%

94%

92% (b) TFC

90% 20

Aliphatic (U) - carbonyl Aliphatic (U) - alkyl Aliphatic (U) Aliphatic (S) PAH

40 60 80 2 Minimum projection area (Å )

Fig. 9. Compound rejection (%) as a function of minimum projection area (Å2) for (a) CTA and (b) TFC membranes. Compounds are organized by class and functional group: unsaturated aliphatics with carbonyl functional groups (purple), unsaturated aliphatics with alkyl functional groups (green), unsaturated aliphatics with no functional groups (red), saturated aliphatics (orange), and polycyclic aromatic hydrocarbons (PAHs) (blue). Rejection values were calculated from day 21 DS and average feed solution concentrations (n = 4). TFC membranes had higher rejection than CTA membranes for small neutral compounds, which is consistent with previous FO rejection studies [75, 76]. Xie et al. [76] attributed high TFC membrane rejection to high active layer thickness to porosity ratios (L/ε) and greater pore hydration. High L/ε values indicate that the membrane has more internal pore surface area, which leads to enhanced adsorption and compound removal. Hydrated pores effectively reduce pore size and improve steric hindrance [76]. High RSF may have also contributed to higher rejection because the reverse diffusion of salts may hinder the forward diffusion of organic compounds [74]. Compound functional groups seemed to affect neutral hydrophobic compound rejection. For both membrane types, compounds at the same MPA but with carbonyl functional groups had lower rejection than all other aliphatics and PAHs. Lower rejection may be because of the hy-

22

drogen bonding capability of carbonyl groups, which may allow for a greater ability to adsorb to and diffuse through the membranes [63, 78-81]. It is also interesting to note that CTA membranes had significantly lower rejection of these compounds than TFC membranes. The finding is in agreement with a previous study, in which low CTA rejection was attributed to neutral hydrophobic compound affinity to CTA membrane surfaces versus TFC membranes [77]. CTA membranes also have lower L/ε values [76] and may have become saturated with these compounds, which can lead to decreased rejection of neutral hydrophobic compounds [74].

4.

Conclusions CTA and TFC membrane performance and fouling was evaluated during three weeks of con-

tinuous produced water treatment. CTA membranes outperformed TFC membranes with higher water flux, lower RSF, and lower SRSF. CTA membrane efficiency improved over the threeweek period, with an overall decrease in SRSF; conversely, TFC membrane efficiency deteriorated with drastic increases in SRSF. Zeta potential measurements revealed that cleaned and fouled TFC membranes became increasingly negative, and cleaned and fouled CTA membrane charge remained stable throughout the study. The effects of fouling layer charge on the diffusion of cations could not be determined and systematic studies are recommended. TFC membranes had a much higher fouling propensity despite smoother, more hydrophilic, and more neutrally charged virgin membrane surfaces. ESEM and FTIR revealed that the foulant composition was similar for both membrane types. FTIR did not detect any biofoulant signatures, but found that organic fouling dominated and was likely comprised of aliphatic hydrocarbons, aromatic hydrocarbons, and carbonates. EDS results showed that inorganic fouling was dominated by iron and silicon; only TFC-F membrane had sodium and chloride, and only CTA-F and TFC-F membranes had calcium. High RSF likely caused sodium and chloride accumulation in TFC fouling layers, and sodium, chloride, and calcium ions were removed with EDTA CEOBs. Despite inorganic ion removal in the fouling layers, EDTA CEOBs only marginally improved water flux for fouled CTA and TFC membranes. 3D EEM plots indicated that PAHs potentially diffused through both membrane types, and naturally occurring organic matter and microorganisms were sufficiently rejected. GC-MS results showed that both membranes had over 90% rejection of neutral hydrophobic compounds, and TFC membranes achieved higher rejection than CTA membranes for molecules with low

23

MPA. Both membranes had lower rejection of compounds with carbonyl functional groups, and CTA membranes had lower rejection of these compounds than TFC membranes. Results from this study lead us to conclude that despite many claims, FO is not a low-fouling membrane separation processes, and that pretreatment of complex streams is needed to protect the FO membranes from excessive fouling and degradation of membrane material.

Acknowledgments Financial support for this work was provided by ConocoPhillips through their contribution to the Center for a Sustainable WE2ST (Water-Energy Education, Science, and Technology) at Colorado School of Mines. This Center was established to promote the joint sustainability of unconventional energy development and water resources in arid regions. The author also acknowledges the support provided by the NSF/SRN program under Cooperative Agreement No. CBET 1240584, and the support of colleagues from WE2ST for their support and guidance throughout this study. Findings, opinions, and conclusions in this work are those of the author and are not a statement or representation of the views, policies or opinions of ConocoPhillips or its employees or representatives. The authors would like to thank Hydration Technology Innovations for providing membranes for the study. The authors would also like to thank Estefani Bustos, Mike Veres, and Tani Cath for their assistance with laboratory analyses and technical support, and to John Veil and Dr. Wayne Buschmann for their technical guidance.

References [1]

[2]

[3]

[4]

D.L. Shaffer, L.H. Arias Chavez, M. Ben-Sasson, S. omero- argas Castrill n, . . Yip, M. Elimelech, Desalination and reuse of high-salinity shale gas produced water: drivers, technologies, and future directions, Environmental Science & Technology 47 (2013) 9569-9583. N.R. Hutchings, E.W. Appleton, R.A. McGinnis, Making high quality frac water out of oilfield waste, Proceedings of the the International Society of Petrolume Engineers 2010 Annual Technical Conference and Exhibition, Florence, Italy, September 19-22, 2010. K.L. Hickenbottom, N.T. Hancock, N.R. Hutchings, E.W. Appleton, E.G. Beaudry, P. Xu, T.Y. Cath, Forward osmosis treatment of drilling mud and fracturing wastewater from oil and gas operations, Desalination 312 (2013) 60-66. B.D. Coday, L. Miller-Robbie, E.G. Beaudry, J. Munakata-Marr, T.Y. Cath, Life cycle and economic assessments of engineered osmosis and osmotic dilution for desalination of Haynesville shale pit water, Desalination 369 (2015) 188-200.

24

[5] [6] [7]

[8] [9] [10]

[11] [12]

[13] [14]

[15]

[16]

[17]

[18]

[19]

[20]

[21]

T.Y. Cath, A.E. Childress, M. Elimelech, Forward osmosis: principles, applications, and recent developments, Journal of Membrane Science 281 (2006) 70-87. B.D. Coday, T.Y. Cath, Forward osmosis: novel desalination of produced water and fracturing flowback, J. AWWA 106 (2014) 37-38. R.L. McGinnis, N.T. Hancock, M.S. Nowosielski-Slepowron, G.D. McGurgan, Pilot demonstration of the NH3/CO2 forward osmosis desalination process on high salinity brines, Desalination 312 (2013) 67-74. S. Lee, C. Boo, M. Elimelech, S. Hong, Comparison of fouling behavior in forward osmosis (FO) and reverse osmosis (RO), Journal of Membrane Science 365 (2010) 34-39. R.W. Holloway, A.E. Childress, K.E. Dennett, T.Y. Cath, Forward osmosis for concentration of anaerobic digester centrate, Water Research 41 (2007) 4005-4014. B. Mi, M. Elimelech, Organic fouling of forward osmosis membranes: Fouling reversibility and cleaning without chemical reagents, Journal of Membrane Science 348 (2010) 337-345. R.W. Holloway, A. Achilli, T.Y. Cath, The osmotic membrane bioreactor: a critical review, Environmental Science: Water Research & Technology 1 (2015) 581-605. T. Nguyen, F. Roddick, L. Fan, Biofouling of water treatment membranes: A review of the underlying causes, monitoring techniques and control measures, Membranes 2 (2012) 804-840. M.T. Madigan, J.M. Martinko, K.S. Bender, D.H. Buckley, D.A. Stahl, Brock biology of microorganisms, 2014, 9780321897398 0321897390. S. Alzahrani, A.W. Mohammad, N. Hilal, P. Abdullah, O. Jaafar, Identification of foulants, fouling mechanisms and cleaning efficiency for NF and RO treatment of produced water, Separation and Purification Technology 118 (2013) 324-341. S. Alzahrani, A.W. Mohammad, Challenges and trends in membrane technology implementation for produced water treatment: a review, Journal of Water Process Engineering 4 (2014) 107-133. G. Chen, Z. Wang, L.D. Nghiem, X.-M. Li, M. Xie, B. Zhao, M. Zhang, J. Song, T. He, Treatment of shale gas drilling flowback fluids (SGDFs) by forward osmosis: Membrane fouling and mitigation, Desalination 366 (2015) 113-120. B.D. Coday, N. Almaraz, T.Y. Cath, Forward osmosis desalination of oil and gas wastewater: Impacts of membrane selection and operating conditions on process performance, Journal of Membrane Science 488 (2015) 40-55. P.H. Duong, T.-S. Chung, Application of thin film composite membranes with forward osmosis technology for the separation of emulsified oil–water, Journal of Membrane Science 452 (2014) 117-126. M.R. Chowdhury, J. Ren, K. Reimund, J.R. McCutcheon, A hybrid dead-end/cross-flow forward osmosis system for evaluating osmotic flux performance at high recovery of produced water, Desalination Y. Chun, S.-J. Kim, G.J. Millar, D. Mulcahy, I.S. Kim, L. Zou, Forward osmosis as a pretreatment for treating coal seam gas associated water: Flux and fouling behaviours, Desalination S. Lee, Y.C. Kim, Calcium carbonate scaling by reverse draw solute diffusion in a forward osmosis membrane for shale gas wastewater treatment, Journal of Membrane Science 522 (2017) 257-266.

25

[22]

[23] [24]

[25]

[26] [27]

[28]

[29]

[30]

[31]

[32]

[33]

[34]

[35]

[36] [37]

J. Minier-Matar, A. Hussain, A. Janson, R. Wang, A.G. Fane, S. Adham, Application of forward osmosis for reducing volume of produced/Process water from oil and gas operations, Desalination 376 (2015) 1-8. Q. She, R. Wang, A.G. Fane, C.Y. Tang, Membrane fouling in osmotically driven membrane processes: A review, Journal of Membrane Science 499 (2016) 201-233. M. Zhang, R. Liu, Z. Wang, B. Zhao, J. Song, M.J. Park, H.K. Shon, X.-M. Li, T. He, Dehydration of forward osmosis membranes in treating high salinity wastewaters: Performance and implications, Journal of Membrane Science 498 (2016) 365-373. Z. Zhou, L. Guo, A.M. Shiller, S.E. Lohrenz, V.L. Asper, C.L. Osburn, Characterization of oil components from the Deepwater Horizon oil spill in the Gulf of Mexico using fluorescence EEM and PARAFAC techniques, Marine Chemistry 148 (2013) 10-21. B. Mi, M. Elimelech, Chemical and physical aspects of organic fouling of forward osmosis membranes, Journal of Membrane Science 320 (2008) 292-302. M. Elimelech, Z. Xiaohua, A.E. Childress, H. Seungkwan, Role of membrane surface morphology in colloidal fouling of cellulose acetate and composite aromatic polyamide reverse osmosis membranes, Journal of Membrane Science 127 (1997) 101-109. M. Xie, L.D. Nghiem, W.E. Price, M. Elimelech, Impact of organic and colloidal fouling on trace organic contaminant rejection by forward osmosis: Role of initial permeate flux, Desalination 336 (2014) 146-152. C.Y. Tang, Q. She, W.C.L. Lay, R. Wang, A.G. Fane, Coupled effects of internal concentration polarization and fouling on flux behavior of forward osmosis membranes during humic acid filtration, Journal of Membrane Science 354 (2010) 123-133. S. Zou, Y. Gu, D. Xiao, C.Y. Tang, The role of physical and chemical parameters on forward osmosis membrane fouling during algae separation, Journal of Membrane Science 366 (2011) 356-362. J.A. Brant, K.M. Johnson, A.E. Childress, Characterizing NF and RO membrane surface heterogeneity using chemical force microscopy, Colloids and Surfaces A: Physicochemical and Engineering Aspects 280 (2006) 45-57. Y. Chun, S.-J. Kim, G.J. Millar, D. Mulcahy, I.S. Kim, L. Zou, Forward osmosis as a pretreatment for treating coal seam gas associated water: Flux and fouling behaviours, Desalination (2015) doi:10.1016/j.desal.2015.1009.1012 N.R. Hutchings, E.W. Appleton, R.A. McGinnis, Making high quality frac water out of oilfield waste, Proceedings of the the International Society of Petroleum Engineers 2010 Annual Technical Conference and Exhibition, Florence, Italy, September 19-22. M. Villain, I. Bourven, G. Guibaud, B. Marrot, Impact of synthetic or real urban wastewater on membrane bioreactor (MBR) performances and membrane fouling under stable conditions, Bioresource Technology 155 (2014) 235-244. B.D. Coday, C. Hoppe-Jones, D. Wandera, J. Shethji, J. Herron, K. Lampi, S.A. Snyder, T.Y. Cath, Evaluation of the transport parameters and physiochemical properties of forward osmosis Membranes after treatment of produced water, Journal of Membrane Science 499 (2016) 491-502. N.T. Hancock, T.Y. Cath, Solute coupled diffusion in osmotically driven membrane processes, Environmental Science & Technology 43 (2009) 6769-6775. N.T. Hancock, Engineered osmosis: Assessment of mass transport and sustainable hybrid system configurations for desalination and water reclamation, The Colorado School of Mines, Golden, CO (2011).

26

[38]

[39]

[40] [41]

[42]

[43]

[44]

[45]

[46]

[47]

[48] [49]

[50]

[51]

[52]

B.D. Coday, D.M. Heil, P. Xu, T.Y. Cath, Effects of transmembrane hydraulic pressure on performance of forward osmosis membranes, Environmental science & technology 47 (2013) 2386-2393. W.A. Phillip, J.S. Yong, M. Elimelech, Reverse draw solute permeation in forward osmosis: Modeling and experiments, Environmental Science & Technology 44 (2010) 5170-5176. A.M. Gilmore, Water quality measurements with HORIBA Jobin Yvon fluorescence instrumentation (Technical Report No. 38 Horiba Yobin Ivon), 2011, Edison, NJ. J. Regnery, B.D. Coday, S.M. Riley, T.Y. Cath, Solid-phase extraction followed by gas chromatography-mass spectrometry for the quanti- tative analysis of semi-volatile hydrocarbons in hydraulic fracturing wastewaters, Analytical Methods 8 (2016) 20582068. Y. Gu, Y.-N. Wang, J. Wei, C.Y. Tang, Organic fouling of thin-film composite polyamide and cellulose triacetate forward osmosis membranes by oppositely charged macromolecules, Water Research 47 (2013) 1867-1874. K. Boussu, A. Belpaire, A. Volodin, C. Van Haesendonck, P. Van Der Meeren, C. Vandecasteele, B. Van der Bruggen, Influence of membrane and colloid characteristics on fouling of nanofiltration membranes, Journal of Membrane Science 289 (2007) 220230. A. Tiraferri, Improving the Performance and Antifouling Properties of Thin-Film Composite Membranes for Water Separation Technologies, Yale University, 2012, 1267856769. B.D. Coday, P. Xu, E.G. Beaudry, J. Herron, K. Lampi, N.T. Hancock, T.Y. Cath, The sweet spot of forward osmosis: treatment of produced water, drilling wastewater, and other complex and difficult liquid streams, Desalination 333 (2014) 23-35. N.T. Hancock, P. Xu, M.J. Roby, J.D. Gomez, T.Y. Cath, Towards direct potable reuse with forward osmosis: Technical assessment of long-term process performance at the pilot scale, Journal of Membrane Science 445 (2013) 34-46. C.-M. Kim, S.-J. Kim, L.H. Kim, M.S. Shin, H.-W. Yu, I.S. Kim, Effects of phosphate limitation in feed water on biofouling in forward osmosis (FO) process, Desalination 349 (2014) 51-59. C. Boo, S. Lee, M. Elimelech, Z. Meng, S. Hong, Colloidal fouling in forward osmosis: role of reverse salt diffusion, Journal of Membrane Science 390 (2012) 277-284. E.M. Hoek, M. Elimelech, Cake-enhanced concentration polarization: a new fouling mechanism for salt-rejecting membranes, Environmental Science & Technology 37 (2003) 5581-5588. J.R. McCutcheon, M. Elimelech, Influence of concentrative and dilutive internal concentration polarization on flux behavior in forward osmosis, Journal of Membrane Science 284 (2006) 237-247. C.H. Tan, H.Y. Ng, Modified models to predict flux behavior in forward osmosis in consideration of external and internal concentration polarizations, Journal of Membrane Science 324 (2008) 209-219. S. Lee, J. Cho, M. Elimelech, Influence of colloidal fouling and feed water recovery on salt rejection of RO and NF membranes, Desalination 160 (2004) 1-12.

27

[53]

[54] [55]

[56]

[57]

[58] [59]

[60] [61]

[62] [63] [64] [65]

[66]

[67] [68]

[69]

M. Elimelech, W.H. Chen, J.J. Waypa, Measuring the zeta (electrokinetic) potential of reverse osmosis membranes by a streaming potential analyzer, Desalination 95 (1994) 269-286. R.J. Petersen, Composite reverse osmosis and nanofiltration membranes, Journal of Membrane Science 83 (1993) 81-150. A.E. Childress, M. Elimelech, Effect of solution chemistry on the surface charge of polymeric reverse osmosis and nanofiltration membranes, Journal of Membrane Science 119 (1996) 253-268. B.D. Coday, T. Luxbacher, A.E. Childress, N. Almaraz, P. Xu, T.Y. Cath, Indirect determination of zeta potential at high ionic strength: Specific application to semipermeable polymeric membranes, Journal of Membrane Science 478 (2015) 58-64. X. Lu, C. Boo, J. Ma, M. Elimelech, Bidirectional diffusion of ammonium and sodium cations in forward osmosis: Role of membrane active layer surface chemistry and charge, Environmental Science & Technology 48 (2014) 14369-14376. R.W. Holloway, R. Maltos, J. Vanneste, T.Y. Cath, Mixed draw solutions for improved forward osmosis performance, Journal of Membrane Science 491 (2015) 121-131. A.E. Childress, M. Elimelech, Relating nanofiltration membrane performance to membrane charge (electrokinetic) characteristics, Environmental Science & Technology 34 (2000) 3710-3716. H. Massaldi, C. Borzi, Non-ideal phenomena in osmotic flow through selective membranes, Journal of Membrane Science 12 (1982) 87-99. B. Van der Bruggen, C. Vandecasteele, Flux decline during nanofiltration of organic components in aqueous solution, Environmental Science & Technology 35 (2001) 35353540. V. Freger, T. Arnot, J. Howell, Separation of concentrated organic/inorganic salt mixtures by nanofiltration, Journal of Membrane Science 178 (2000) 185-193. C. Bellona, J.E. Drewes, P. Xu, G. Amy, Factors affecting the rejection of organic solutes during NF/RO treatment—a literature review, Water Research 38 (2004) 2795-2809. Y. Kim, S. Lee, S. Hong, Influence of solution chemistry on the surface heterogeneity of reverse osmosis membrane, Desalination and Water Treatment 43 (2012) 308-313. S. Lee, E. Lee, M. Elimelech, S. Hong, Membrane characterization by dynamic hysteresis: Measurements, mechanisms, and implications for membrane fouling, Journal of Membrane Science 366 (2011) 17-24. C.H. Webb, L. Nagghappan, G. Smart, J. Hoblitzell, R. Franks, Desalination of oilfieldproduced water at the San Ardo water reclamation facility, CA, Proceedings of the SPE Western Regional Meeting, San Jose, CA, March 24-26. N. Colthup, L. Daly, S. Wiberley, Introduction to infrared and Raman spectroscopy, 2, Academic Press, 1975, 032316160X. G. Matrajt, J. Borg, P. Raynal, Z. Djouadi, L. d'Hendecourt, G. Flynn, D. Deboffle, FTIR and Raman analyses of the Tagish Lake meteorite: Relationship with the aliphatic hydrocarbons observed in the Diffuse Interstellar Medium, Astronomy & Astrophysics 416 (2004) 983-990. R.E. Riecker, Hydrocarbon fluorescence and migration of petroleum, AAPG Bulletin 46 (1962) 60-75.

28

[70]

[71]

[72] [73]

[74]

[75] [76]

[77]

[78]

[79]

[80]

[81]

J. Beltran, R. Ferrer, J. Guiteras, Multivariate calibration of polycyclic aromatic hydrocarbon mixtures from excitation–emission fluorescence spectra, Analytica Chimica Acta 373 (1998) 311-319. N. Ferretto, M. Tedetti, C. Guigue, S. Mounier, R. Redon, M. Goutx, Identification and quantification of known polycyclic aromatic hydrocarbons and pesticides in complex mixtures using fluorescence excitation–emission matrices and parallel factor analysis, Chemosphere 107 (2014) 344-353. P.G. Coble, Characterization of marine and terrestrial DOM in seawater using excitationemission matrix spectroscopy, Marine Chemistry 51 (1996) 325-346. C.A. Stedmon, S. Markager, R. Bro, Tracing dissolved organic matter in aquatic environments using a new approach to fluorescence spectroscopy, Marine Chemistry 82 (2003) 239-254. B.D. Coday, B.G. Yaffe, P. Xu, T.Y. Cath, Rejection of trace organic compounds by forward osmosis membranes: a literature review, Environmental Science & Technology 48 (2014) 3612-3624. X. Jin, J. Shan, C. Wang, J. Wei, C.Y. Tang, Rejection of pharmaceuticals by forward osmosis membranes, Journal of Hazardous Materials 227 (2012) 55-61. M. Xie, L.D. Nghiem, W.E. Price, M. Elimelech, Relating rejection of trace organic contaminants to membrane properties in forward osmosis: measurements, modelling and implications, Water Research 49 (2014) 265-274. P. Xu, J.E. Drewes, T.-U. Kim, C. Bellona, G. Amy, Effect of membrane fouling on transport of organic contaminants in NF/RO membrane applications, Journal of Membrane Science 279 (2006) 165-175. T. Fujioka, S.J. Khan, J.A. McDonald, L.D. Nghiem, Rejection of trace organic chemicals by a nanofiltration membrane: the role of molecular properties and effects of caustic cleaning, Environmental Science: Water Research & Technology (2015) 846-854. N.T. Hancock, P. Xu, D.M. Heil, C. Bellona, T.Y. Cath, Comprehensive bench-and pilotscale investigation of trace organic compounds rejection by forward osmosis, Environmental Science & Technology 45 (2011) 8483-8490. T. Fujioka, S.J. Khan, J.A. McDonald, L.D. Nghiem, Nanofiltration of trace organic chemicals: A comparison between ceramic and polymeric membranes, Separation and Purification Technology 136 (2014) 258-264. T. Matsuura, S. Sourirajan, Physicochemical criteria for reverse osmosis separation of alcohols, phenols, and monocarboxylic acid in aqueous solutions using porous cellulose acetate membranes, Journal of Applied Polymer Science 15 (1971) 2905-2927.

Highlights  CTA and TFC FO membranes treated produced water for up to three weeks.  CTA outperformed TFC with higher water flux, lower RSF, and lower SRSF.  Foulants were similar for both membranes, and TFC had higher fouling propensity.  Weekly chemically enhanced osmotic backwashing minimally improve water flux.  Membranes rejected over 90% of detected compounds, and TFC had higher rejection.

29

30