ABSTRACT
This research tested the treatment efficacy of an Energy Savings Nanofiltration 1 Low Fouling (ESNA 1-LF) nanofiltration (NF) and an Energy Saving Polyamide 2 (ESPA2) reverse osmosis (RO) membrane for removing malathion from water. Both membranes are of composite polyamide construction. The study included measuring malathion rejection using both pristine membranes and membranes exposed to a simulated secondary wastewater effluent foulant before and after two types of clean-in-place procedures. Across all conditions studied, malathion rejection ranged from 84 to 95% for the ESNA1-LF NF membrane and 77 to 94% for the ESPA2 RO membrane. Contact angle measurements were also collected for each membrane exposure condition. While the contact angle measurements indicated changes to the hydrophobicity of the selective layer of the membranes, they did not correlate to changes in the performance of malathion rejection. As expected, it was observed that malathion rejection improved with the introduction of foulant. Also, the clean-in-place procedures helped restore flux while maintaining malathion rejection.
HIGHLIGHTS
ENSA-1LF and ESPA2 membranes removed 77–95% of malathion from the waters studied.
Malathion rejection improved with the introduction of simulated secondary wastewater effluent foulant.
Clean-in-place procedures helped restore flux while maintaining malathion rejection.
INTRODUCTION
Molecular structure of (a) malathion and (b) chemical warfare agent VX.
While malathion concentrations in most waters are low to nonexistent, malathion has been detected at high concentrations in a variety of different water sources internationally. In an assessment of 1,098 journal articles (selected through a targeted keyword search in Scopus, Embase, and PubMed databases with inclusion criteria including the reporting of malathion in water resources, articles published in English, and that they were cross-section studies), the mean values of malathion in highly contaminated drinking waters were 26,684.64 μg/L, ground waters 397.75 μg/L, and surface waters 9,310.78 μg/L (Vasseghian et al. 2022). The meta-analysis also found that in these highly contaminated waters, exposure to malathion residues in drinking water sources was above recommended levels of carcinogenic risks in some locations. The United States Environmental Protection Agency currently does not regulate malathion in drinking water, but it is listed on the drinking water contaminant candidate list (2022) and has a health advisory level of 0.2 mg/L (U.S. EPA 2018). The U.S. guidelines for malathion limits in surface waters that range from 0.1 to 140 μg/L depend on the state (Agency for Toxic Substances & Disease Registry 2003).
Several different water treatment processes have been found effective at removing malathion. Advanced oxidation processes involving hydrogen peroxide in combination with iron, ozone, or iron with ultraviolet irradiation have been found to remove 60–99, >99, and 94.3% of malathion, respectively (Roche & Prados 1995; Roe & Lemley 1997; Huston & Pignatello 1999). Ozone at a dose of 5 mg/L and 10 min of contact time also removed 99% of malathion from water (Roche & Prados 1995). Additionally, adsorption has proved to be an effective method for treating malathion-contaminated waters (Gupta et al. 2002; Chatterjee et al. 2010; Jusoh et al. 2011; Marican & Durán-Lara 2018; Sabbagh et al. 2021). While there are proven treatment technologies, not all water treatment scenarios will have the equipment and funds available to implement additional advanced oxidation processes or sorbents, and therefore, it is necessary to consider additional treatment technologies as well.
RO membranes were developed for water treatment applications in the 1970s (Warsinger et al. 2018). They are composed of support and a top polymer layer that permeates water. The polymer layer has sub-nanometer pores and is most often made from polyamide or cellulose acetate materials (Lee et al. 2011). Pressure is applied to drive water through the membrane leaving rejected contaminants on the feed side. Due to their high salt rejection, RO is the most used desalination technology in the water treatment industry and has also been successfully used to separate low-molecular-weight organic compounds such as pesticides (Chian et al. 1975). NF membranes were introduced approximately 20 years later as an alternative to RO with higher water permeability and therefore lower operating pressures and energy costs. This stems from NF membranes having a higher molecular weight cutoff (MWCO) of approximately 100–5,000 Da compared to RO's MWCO of <100 Da (Peter-Varbanets et al. 2009). Malathion has a MWCO of 330.4 Da, a molecular width of 0.510 nm, and a log n-octanol/water partition coefficient of 2.36 (Kiso et al. 2000), so rejection by sieving is expected to be successful for RO membranes and most, but not all, NF membranes. NF membranes are also more commonly designed to remove specific organic contaminates than RO (Bellona et al. 2004). Several membrane-based studies have indicated effective treatment of malathion using RO or NF membranes (Chian et al. 1975; Kiso et al. 2000; Zhang & Pagilla 2010; Sorour & Shaalan 2013). However, there are gaps in these studies because only a subset of pristine membrane material types and feed solutions were tested. At a conventional water treatment plant, membrane processes follow multiple pre-treatment unit treatment processes such as screening, coagulation, flocculation, and settling basins. In emergency mobile water treatment applications, minimal pre-treatment occurs prior to treatment of the malathion. Both RO and NF membrane processes become fouled when feed streams contain dissolved inorganic and organic matter. This can either increase or decrease separation efficiency but causes a reduction of membrane flux that requires cleaning to restore normal performance (Agenson & Urase 2007). This is particularly applicable in potable reuse and mobile treatment applications. The research presented in this paper uniquely contributes to the field by expanding the membranes tested and by including foulants before and after cleaning.
EXPERIMENTAL
Membranes and crossflow test unit
Test sequence
Accelerated fouling experiments
Accelerated fouling experiments were conducted to condition the membranes to a state that would trigger clean-in-place procedures during typical operations. According to the manufacturer, routine cleaning is advised when the normalized permeate flow has decreased by 10–20%, the normalized permeate quality decreased by 10–20%, or the normalized pressure has dropped by 15–30% (Hydranautics 2020). Fouling was performed through a multi-step process. Membranes were compacted and equilibrated under a target pressure of 1.72 MPa by running DI water through the system until flux plateaued (approximately 20–45 min). After the DI water step, a 2,000 mg/L solution of NaCl (Product Number S271-3, Fischer Scientific, Waltham, Massachusetts) was fed through the system at a target pressure of 1.72 MPa until flux plateaued (approximately 20–45 min). Salt rejection was quantified by using conductivity (Model 19820-10 TDS/Conductivity Meter, Cole-Parmer, Vernon Hills, IL) with percentage rejection calculated by subtracting the permeate concentration from the feed concentration and dividing by the feed concentration. Finally, a simulated secondary wastewater effluent foulant solution comprised of 20 mg/L alginate (Product Number 180947-100G, Sigma Aldrich, St. Louis, Missouri) and 20 mg/L total organic carbon (Suwannee River NOM, Catalog Number 2R101N, International Humic Substances Society, Denver, CO) was recirculated at a target pressure of 250 psi until flux dropped by at least 20% (approximately 2 h). The conditioned membrane coupons were stored moist (in approximately 1 mL of DI water) until used in cleaning experiments.
Malathion treatability experiments
During Phase II experiments, new and fouled membranes were tested for their ability to remove malathion. A 10 mg/L solution of malathion was prepared using Spectracide Malathion Insect Spray Concentrate (Product Number 071121109002, Spectrum Brands, Inc., Madison, WI). This was selected to represent a highly contaminated drinking water source. Spectracide contains 44% of the solvent naphtha (petroleum). No attempt to boil off the naphtha was attempted. DI water was recirculated for at least 10 min followed by the recirculation of 2,000 mg/L NaCl for at least 10 min (note that one experimental run, ESNA1 with cleaning solution A, did not include the 2,000 mg/L NaCl step, but the DI water step was recirculated for 20 min instead of 10). Following the equilibration with water/salt, the malathion feed solution was recirculated through the membranes for 60 min with paired samples collected from the feed tank and the permeate tanks for both membranes at 20, 40, and 60 min. The crossflow velocity during sample collection ranged from 0.04 to 0.12 m/s and averaged 0.08 m/s across all experiments pre- and post-cleaning. Rejection was calculated as a percentage difference at each time point of the feed minus the permeate concentration divided by the feed concentration and averaged for reporting the overall rejection in this paper. This experimental sequence (DI, salt, malathion) was then repeated post-cleaning. Malathion concentrations were quantified using EPA method 622 (EPA 1992), which is a gas chromatographic analytical method.
Cleaning experiments
Two cleaning solutions were investigated in this research. Solution A was a gentle cleaning solution from Nitto Hydranautics Technical Service Bulletin 107.72 recommended for light to moderate levels of organic foulants. It contained 2% (w) sodium tripolyphosphate (STPP) (Product Number 238503-500G, Sigma Aldrich, St. Louis, MO), 0.8% (w) the sodium salt of ethylenediaminetetraacetic acid (Na-EDTA) (Product Number E9884-500G, Sigma Aldrich, St. Louis, MO), and pH adjusted to 10 using hydrochloric acid (Product Number 258148-2.5L, Sigma Aldrich, St. Louis, MO) (Hydranautics 2020). Solution B was a high pH cleaning solution that used NaOH (Product Number S318-1, Fischer Scientific, Waltham, MA) to adjust pH to 11.5. No detergent was added to solution B (0.03% [w] sodium dodecyl sulfate is recommended in TSB 107.72) to test a simpler option for use during emergency response field operations. The cleaning procedure was a six-step process. First, the permeate valve was opened. Next the membranes were flushed for 3 min with DI water at 0.34 MPa and 200 mL/min. After the DI water flush, the cleaning solution was recirculated at 0.34 MPa for 30 min. A slow ramp up of the flowrate was used to avoid clogging the system by the foulant removal and was 200 mL/min for 5 min, than 400 mL/min for 5 min, and finally 600 mL/min for the remaining 20 min. The coupons were then left in the test skid saturated in cleaning solution for a 30 min soak, followed by the recirculation of the cleaning solution for another 30 min at 0.34 MPa and the same ramp of flowrates as in the initial recirculation. The final step was a 3-min flush with DI water at 0.34 MPa and 200 mL/min.
Contact angle experiments
After Phase II experiments were completed, captive bubble contact angle measurements were collected using a standard goniometer (Model No. 250-U1, ramé-hart instrument co., Succasunna, NJ) with environmental fixture (Product # 100-14, ramé-hart instrument co., Succasunna, NJ). The instrument was calibrated using a combo calibration device (Product # 100-27-31-U, ramé-hart instrument co., Succasunna, NJ), and Teflon was used as a reference measurement prior to analysis of the membranes. Each membrane was cut to approximately 0.64 cm × 1.9 cm, adhered to the stage using double-sided tape, and submerged in ultrapure water. A 20 μL bubble was generated using an inverted stainless-steel needle (sd = 3.6 μL). Image analysis was performed using ImageJ software v. 1.53.
RESULTS AND DISCUSSION
Phase I accelerated fouling experiments
Flux and pressure data from accelerated fouling experiments for two (a) ESNA1-LF NF membranes and (b) ESPA2 RO membranes. To produce replicates for the treatability experiments, both cells were fouled using the same feed solution of 20 mg/L alginate and 20 mg/L total organic carbon natural organic matter.
Flux and pressure data from accelerated fouling experiments for two (a) ESNA1-LF NF membranes and (b) ESPA2 RO membranes. To produce replicates for the treatability experiments, both cells were fouled using the same feed solution of 20 mg/L alginate and 20 mg/L total organic carbon natural organic matter.
Phase II salt solution
Salt performance data (% rejection and water flux) for pristine and fouled ESPA2 RO and ESNA1-LF NF before and after cleaning with (a) 2% (w) of STPP, 0.8% (w) of Na-EDTA, and pH adjusted to 10 and (b) pH adjusted to 11.5 using NaOH. Note that salt solution flux was not measured for the ESNA1-LF pre-cleaning test with cleaning solution A.
Salt performance data (% rejection and water flux) for pristine and fouled ESPA2 RO and ESNA1-LF NF before and after cleaning with (a) 2% (w) of STPP, 0.8% (w) of Na-EDTA, and pH adjusted to 10 and (b) pH adjusted to 11.5 using NaOH. Note that salt solution flux was not measured for the ESNA1-LF pre-cleaning test with cleaning solution A.
Phase II malathion solution
Malathion performance data (% rejection and water flux) for pristine and fouled RO and NF before and after cleaning with (a) 2% (w) of STPP, 0.8% (w) of Na-EDTA, and pH adjusted to 10 and (b) pH adjusted to 11.5 using NaOH.
Malathion performance data (% rejection and water flux) for pristine and fouled RO and NF before and after cleaning with (a) 2% (w) of STPP, 0.8% (w) of Na-EDTA, and pH adjusted to 10 and (b) pH adjusted to 11.5 using NaOH.
Across all conditions, malathion rejection ranged from 84 to 95% for the ESNA1-LF NF membrane and 77 to 94% for the ESPA2 RO membrane. These results are comparable to malathion rejection using different types of water treatment membranes. For example, when malathion rejection was assessed using a NF ceramic membrane and feed concentrations of malathion of 5.1–17.1 mg/L, performance was 93.5–99.4% removal of malathion (Sorour & Shaalan 2013). Zhang and Pagilla studied malathion rejection of two NF polyamide thin-film composite membranes and one NF polypiperazine amide thin-film composite membrane and observed rejections ranging from approximately 55–99% depending on the transmembrane pressure and membrane pore size (Zhang & Pagilla 2010). They observed a size exclusion mechanism where the highest rejection was observed for the membrane with the lower pore size. This correlation was not observed in our study as others have reported similar pore sizes of 0.439 nm for the ESNA1-LF NF membrane (Tanne et al. 2019) and 0.289 nm for the ESPA2 RO membrane (Fujioka et al. 2013b). Kiso et al. studied four different membranes (3 NF and 1 RO); two were made from poly(vinyl alcohol)/polyamide and two were sulfonated polyethersulfone-based. Their malathion rejection ranged from 41 to 99.14%, with the RO membrane having the highest rejection. The sulfonated polyethersulfone NF membrane with the lowest NaCl rejection had the worst malathion rejection. Furthermore, they observed adsorption of pesticide on the membranes, with more absorption occurring on the sulfonated polyethersulfone than the poly(vinyl alcohol)polyamide (Kiso et al. 2000). In our work, salt rejection was high for both types of membranes, so no such correlation between desalting capabilities and malathion rejection was observed. Another study also found a considerable amount of pesticide adsorbed onto RO membranes, but the polymers of the membranes were cellulose acetate and a cross-lined polyethylenimine, which presently represent a smaller market share of actively used membranes (Chian et al. 1975). The rejection of malathion for both types of membranes in Chain et al. was greater than 99%. The use of foulants and clean-in-place procedures in our study may have mitigated some adsorption that provided additional treatment in previous studies (Chian et al. 1975; Kiso et al. 2000). In our study, contact angle measurements do not indicate a uniformity in surface properties that might occur if a large amount of chemical remained on the surface. However, the rebound of the salt solution flux after cleaning, but not the malathion solution flux indicates potential changes in the membranes' selective layer that had a different impact on solutions containing smaller ions than longer organic molecules. The impact of the naphtha contained in the Spectracide was not investigated in this study. Not observed in this study, others have found that emulsified oils foul RO membranes (Kasemset et al. 2013).
Contact angle
The ESNA1-LF NF membrane had a contact angle of 30° and the ESPA2 RO membrane of 35° prior to any experimentation (Table 1). Regardless of exposure to cleaning solution A or B, the pristine membranes' contact angles remained relatively unchanged (±2°). The fouled membranes exposed to cleaning solution A saw a modest increase in the contact angle (+5° for ESNA1-LF NF and +3° for ESPA2 RO). The fouled membranes exposed to cleaning solution B were relatively unchanged for ESNA1-LF NF but saw a decrease of 10° for ESPA2 RO. The contact angle measured for ESNA1-LF NF was very similar to contact angles measured using the captive bubble technique in the literature (Muthu et al. 2014), but captive bubble contact angle measurements have been previously reported as higher for ESPA2 RO at 61.3° (Varin et al. 2013). From the data provided by Varin et al. no clear indication arose as to the discrepancy, although there have been several iterations of ESPA2 products, so perhaps it was a different generation of the membrane. Contact angle measurements collected in this work do not appear to correlate to any changes in membrane performance (which stayed relatively constant in this study). They do suggest minor material changes when fouled and different changes due to the interactions of cleaning solutions but within the uncertainty of the measurement. If rejection does begin to suffer, contact angle measurement may be a useful qualitative tool for diagnosing if changes to the membrane surface are contributing to a decline in performance.
Contact angle measurements for a variety of membrane exposure conditions
Exposure conditions . | Contact angle (°) . | |
---|---|---|
ESNA1-LF NF . | ESPA2 RO . | |
Not used in experiments | 30 | 35 |
Pristine, cleaning solution A | 29 | 33 |
Fouled, cleaning solution A | 35 | 38 |
Pristine, cleaning solution B | 29 | 36 |
Fouled, cleaning solution B | 29 | 25 |
Exposure conditions . | Contact angle (°) . | |
---|---|---|
ESNA1-LF NF . | ESPA2 RO . | |
Not used in experiments | 30 | 35 |
Pristine, cleaning solution A | 29 | 33 |
Fouled, cleaning solution A | 35 | 38 |
Pristine, cleaning solution B | 29 | 36 |
Fouled, cleaning solution B | 29 | 25 |
CONCLUSIONS
This study adds data to the limited existing literature that indicates polyamide-based NF and RO membranes are able to remove high percentages of malathion from water. The study introduced foulant representative of secondary wastewater effluent and observed that malathion rejection improved. This foulant applies to potable reuse facilities that are being increasingly implemented in water-stressed regions. The membranes were also exposed to two different clean-in-place procedures that helped restore flux while maintaining malathion rejection. While this project was limited in scope/funding, future research should be conducted to obtain a comprehensive exploration of different operating parameters (e.g., pressure, pH, and temperature) and extended fouling experiments with different types of foulants (e.g., different types of scale and biological material). The data presented here suggest that polyamide-based NF and RO membranes are an effective tool for utilities and emergency responders to treat malathion-contaminated waters. After an accidental or intentional release of pesticides or chemical weapons, an understanding of emergency water treatment capabilities is necessary so that an effective remediation strategy can be launched as soon as possible.
ACKNOWLEDGEMENTS
The U.S. Environmental Protection Agency (EPA) through its Office of Research and Development directed the research described herein under Interagency Agreement DW-089-92528301-7 with Idaho National Laboratory. It has been reviewed by the Agency but does not necessarily reflect the Agency's views. No official endorsement should be inferred. EPA does not endorse the purchase or sale of any commercial products or services. The authors would like to thank Ramona Sherman of EPA for her quality review and Jim Goodrich and Tae Lee of EPA for their internal technical reviews.
DATA AVAILABILITY STATEMENT
All relevant data are included in the paper or its Supplementary Information.
CONFLICT OF INTEREST
The authors declare there is no conflict.
REFERENCES
Author notes
These authors contributed equally to this work.