Environ. Sci. Technol. 2010, 44, 7914–7920
Treatment of Perfluorinated Chemicals by Electro-Microfiltration YU-TING TSAI, ANGELA YU-CHEN LIN,* YU-HSIANG WENG, AND KUNG-CHEH LI Graduate Institute of Environmental Engineering, National Taiwan University, 71, Chou-Shan Rd., Taipei 106, Taiwan (R.O.C.)
Received October 28, 2009. Revised manuscript received August 22, 2010. Accepted September 15, 2010.
Perfluorinated compounds (PFCs) are negatively charged and have low pKa values in water; therefore, a laboratoryscale electro-microfiltration (EMF) unit that applies a directcurrent electrical field across its membrane can greatly enhance their removal from aqueous systems. We examined the effects of an aqueous inorganic matrix (pH: 4, 7, or 10; ionic strength: 0.4-4.8 mM; ionic composition: Na2SO4, NaCl, NH4Cl or CaCl2) and an organic matrix such as dissolved organic matter (DOM) on the ability of EMF to remove perfluorooctanoic acid (PFOA) and perfluorooctane sulfonate (PFOS). Decreased removal of PFOX (X ) A or S) was observed when the proton concentration and the ionic strength increased. When the applied electrical field strength was less than the critical electrical field strength (Ecritical, HA), PFOX removal was lower in the presence of DOM. We hypothesize that these matrices affect PFOX rejection by altering membrane zeta potential during filtration in the presence of an electrical field. In addition, EMF was found to remove three other PFCs effectively (perfluorodecanoic acid, perfluorohexane sulfonate, and perfluorohexanoic acid), and was also able to remove 70% PFOX and 80% DOC from real industrial wastewaters.
1. Introduction Since the 1950s, perfluorinated compounds (PFCs) have been used extensively in industrial and commercial products such as repellents, surfactants, fire-fighting foams, and cosmetics for their distinctive surface activity and chemical and thermal stability. Nevertheless, PFCs are considered to be a new organic pollutant due to their persistence, toxicity, and bioaccumulation (1, 2). Perfluorooctanoic acid (PFOA) and perfluorooctane sulfonate (PFOS) are the PFCs most often detected in aqueous environments (3, 4). It is difficult to remove PFOX (X ) A or S) from water due to their low concentrations (ng/L-µg/L) in aqueous environments. Therefore, it is essential to understand the fate of these trace organic contaminants and to develop effective treatment methods. Membrane technology has recently received extensive attention because it provides an alternative method for meeting stricter water treatment regulations. Microfiltration (MF) with large pore sizes has been used to efficiently remove fine particulates and pathogens from water (5). Ceramic MF membranes in particular allow higher flux and better chemical compatibility compared to organic membranes and are therefore suitable for treating wastewaters from various waste streams. In theory, PFCs are not excluded by MF * Corresponding author phone: (886)-2-33664386; fax: (886)-233669828; e-mail:
[email protected]. 7914
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 44, NO. 20, 2010
because they are smaller than the average membrane pore size. However, PFCs are negatively charged in water, suggesting that the performance of MF process may be improved by applying an electrical force that attracts these charged compounds. This combination technology is called electromicrofiltration (EMF). EMF is a pressure-driven membrane process in which an additional electrical current is applied simultaneously during mechanical filtration. Since the applied electrical field gradient is parallel to the flow, charged substances would be retained by the electrical force, thus increasing the performance of the membrane. Electrophoresis and electroosmosis, two important electrokinetic phenomena, participate concurrently in EMF. Electrophoresis is the movement of a charged species under the influence of an electrical field, whereas electroosmosis uses counterions under the influence of an electrical field to draw a liquid through a membrane. In addition, electrochemical reactions may occur if target compounds are exposed to the electrodes (6). Many studies have reported the application of EMF to water and wastewater treatments on laboratory and pilot scales (6-9). Although electrofiltration requires more energy than traditional membrane filtration does, the incremental flux that results from the applied electrical field strength may compensate for the increased energy consumption when there is significant flux enhancement. Furthermore, rejection of the pollutant is increased after voltage is applied, indicating that additional costly treatments may not be necessary. Water inorganic matrices (pH, ionic strength and composition) and organic matrices such as dissolved organic matter (DOM) may influence filtration flux and solute rejection by altering the membrane zeta potential or the chemical speciation of solutes; this might affect many solutemembrane or solute-solute interactions (10-15). When an electrical force is applied to the filtration system to counter the hydrodynamic force, the membrane zeta potential may significantly affect solute rejection. To the best of our knowledge, no reports have discussed the efficacy of PFC removal by EMF. In this work, we have used a laboratoryscale EMF unit to study PFC removal from water under various conditions. The effect of membrane zeta potential on EMF performance was examined under the following conditions (1): in acidic, neutral or basic feeding solutions (pH 4, 7, or 10) (2), in electrolyte solutions of various ionic strengths, and (3) in the presence of DOM. Finally, a more realistic trial was conducted by applying EMF to real wastewater.
2. Materials and Methods 2.1. Filtration Module. We used a tubular ceramic MF membrane with nominal pore size 0.1 µm. Membrane characteristics are listed in Supporting Information (SI) Table S1; Figure 1 shows a schematic layout of the experimental setup. The main filtration module consisted of an external housing (a hollow plastic tube), an internal cathodic layer encircling the MF membrane and a concentric anodic rod. The internal diameter of the filter and the diameter of the concentric anodic rod were 6 and 1.3 mm, respectively. The anode was made of platinum, and the cathode was made of titanium. This tubular ceramic membrane was assembled in the external housing to form a cross-flow cell with a filtration surface area of 20.7 cm2. 2.2. Feed Solution. The PFCs used in this study are listed in SI Table S2. In addition, the pKa and effective diameters are also reported (16-22). We purchased PFHxA, PFOA, and PFDA from Sigma-Aldrich (St. Louis, MO), and PFHxS and 10.1021/es101964y
2010 American Chemical Society
Published on Web 09/27/2010
FIGURE 1. Schematic diagram of a laboratory scale electro-microfiltration (EMF) system. PFOS from Fluka (Buchs, Switzerland). All PFC standards were of purity >97%. The initial concentration of each PFC was 100 µg/L in all experiments. Humic acid (HA) was used to represent the DOM in synthetic water. The stock solution was prepared by mixing 1 g of powdered sodium salt of HA (Sigma-Aldrich, St. Louis, MO, USA) with 1 L of Milli-Q water. The solution was filtered through a 0.45 µm filter to remove particulates and stored at 4 °C until usage. The concentration of dissolved organic carbon (DOC) in the feed solution was approximately 5 mg/ L. For EMF experiments with varying pH or DOM, NaCl was used to adjust the conductivity to 100 µS/cm (0.8 mM). The pH was adjusted separately. 2.3. Filtration Experiment. Five liters of PFC-containing solution with or without HA were prepared for each experiment. A direct-current (DC) power supply (Chroma, model 6210K-600/1000 W) provided electrical voltage. A peristaltic pump was used to introduce the feed solution into the filtration cell, and pressure was maintained at 49 kPa with a pressure gauge. The cross-flow velocity was 0.18 m/s. Permeate was collected in a polypropylene container, measured with an electrical balance, and recycled back into the system to maintain a constant feed concentration. Samples were taken for determination of PFC and DOM concentrations during the course of filtration. The membrane was cleaned by sequential flushing with 0.1 N NaOH, 50% aqueous methanol and Milli-Q water after each experiment. After each 10 min washing cycle, the membrane was soaked in a 0.1 N NaOH solution for 12 h and in a 50% aqueous methanol solution for another 12 h. Afterward, the membrane was immersed in Milli-Q water to remove residual methanol and NaOH. Lastly, the clean water flux was measured after Milli-Q water was filtered through for 1 h. After the cleaning process, the clean water flux recovery was 98-100%.
The membrane zeta potential (ζ) at different solution conditions can be determined from the electroosmotic fluxversus-electrical current curve described by the HelmholtzSmoluchowski equation (23): Dζ Q )I ηκ
(1)
where Q is the electroosmotic flux (m3/s), I is the applied electrical current (A), D is the constant ) ε0εr (C/Vm), ε0 and εr are the vacuum permittivity and relative dielectric constant of the medium, and η and κ are the solution viscosity (Ns/ m2) and the conductivity of the electrolytes in the bulk solution (S/m). The experimental setup for measurement of the electroosmotic flux was the same as for the EMF experiments. The typical measurement time was 5 min. 2.4. LC-ESI-MS/MS and Other Chemical Analysis. The concentrations of PFOX and three other PFCs were quantified using liquid chromatography tandem mass spectrometry (LCMS/MS). Chromatographic separation was performed using an Agilent 1200 HPLC (Agilent, Palo Alto, CA) equipped with a ZORBAX Eclipse XDB-C18 column (150 × 4.6 mm, 5 µm). Mass spectrometric measurements were carried out on a Sciex API 4000 (Applied Biosystems, Foster City, CA) equipped with an electrospray ionization interface in negative mode. Data acquisition was performed in multiple reaction monitoring mode with a dwell time of 30 ms and unit mass resolution on both mass analyzers. The other LC and MS/ MS parameters used in this study have been described in detail elsewhere (4). Quantification was based on a 7-point calibration curve with a linear range from 1 to 125 µg/L. DOC was analyzed with an organic carbon analyzer (model IL-550 TOC-TN, Lachat Instruments, Germany). The pH and conductivity were measured by an EC-212 pH meter and EC220 conductivity meter, respectively. VOL. 44, NO. 20, 2010 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
7915
TABLE 1. Zeta Potential Measurements for the Microfiltration Membrane at Different pH Values and Electrolyte Solutions electro-osmotic flux method (ionic strength of 0.8 mM) pH zeta zeta zeta zeta a
FIGURE 2. Removal efficiencies of (a) PFOA (b) PFOS at various electrical field strengths at pH 4, 7, and 10.
3. Results and Discussion 3.1. Effect of Electrical Field Strength on PFOX Removals. The removal efficiencies and rejections (η) of PFOX and the other PFCs were determined by the following equation: η)
C0 - Cp × 100% C0
(2)
where C0 is the initial PFOX concentration in the bulk solution, and Cp is the PFOX concentration in the permeate stream. PFOX rejection remained consistent over time, indicating that these compounds were separated immediately from water when an electrical field was applied (SI Figure S1). Figure 2 depicts the removal efficiencies of PFOX as a function of solution pH at three different electrical field strengths (0, 29, and 58 V/cm). In the absence of an electrical field, PFOX removal efficiencies were low, indicating that the MF membrane alone was insufficient to remove these pollutants. The effective diameters of PFOA and PFOS are 0.91 and 0.99 nm, respectively (SI Table S2). Therefore an MF membrane with a nominal size of 100 nm should not be able to remove either PFOX. These results also suggest insignificant PFOX adsorption onto the MF membrane. However, the application of a DC electrical field through the membrane greatly enhanced PFOX removal. At pH 4, as electrical field strength increased to 29 and 58 V/cm, the average removal efficiencies of PFOA and PFOS increased to 42.4% (PFOA) and 45.6% (PFOS) at 29 V/cm and to 61.3% (PFOA) and 65.6% (PFOS) at 58 V/cm. In addition, PFOA and PFOS removals at solution pH 7 increased to 56.3% and 57.6% in a 29 V/cm electrical field, and further yet to 72.3% and 72.2% in a 58 V/cm 7916
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 44, NO. 20, 2010
potential, potential, potential, potential,
4
7
10
NaCl (mV) -12.2 -16.7 (-13.5) NH4Cl (mV) -17.2 CaCl2 (mV) -11.3 Na2SO 4 (mV) -19.1
a
-23.1
Zeta potential of membrane after HA adsorption.
electrical field. Moreover, PFOA and PFOS rejections increased to >84% at pH 10 in a 58 V/cm electrical field. These results clearly demonstrate that EMF increased PFOX removal efficiencies and are consistent with findings from earlier studies that used EMF to treat water containing arsenic and oxide-chemical mechanical polishing (oxide-CMP) wastewater (7, 24). Weng et al. (24) reported that rejection of arsenate (V) increased from 30% to >90% after applying an electrical field, whereas Yang et al. (7) found that removals from oxide-CMP wastewater were high for turbidity and total organic carbon. To determine the mechanism of PFOX removal by EMF, we monitored PFOX concentrations in the tank, reasoning that if electrochemical degradation were important for PFOX removal, their concentrations in the feed tank would decrease significantly. From mass balance analysis, we found that recoveries of PFOA and PFOS were 101% and 99.5%, respectively, indicating insignificant electrochemical loss in the EMF system. Electrochemical degradation of PFOX by EMF has been previously excluded as a major removal mechanism when conductivity is low (100 µS/cm), filtration cell retention time is short (few seconds), or inert material (Pt) is used for the anode electrode (6, 25). Other studies have pointed out that PFOX are resistant to decomposition by advanced oxidation processes such as ozone, ozone/UV, ozone/H2O2 and the Fenton reagent (26, 27). Consequently, physical separation due to electrophoretic attraction during EMF may be the dominant mechanism in PFOX removal. 3.2. Effect of pH and Ions on PFOX Removals and Their Removal Mechanism. The effects of different pH levels (4, 7, and 10) on the speciation of PFOX were expected to be of minimal significance, since these compounds have fairly low pKas and are always in their negative forms. However, PFOX removal efficiency was strongly and positively correlated with solution pH (Figure 2). Solution pH may affect solute speciation and membrane zeta potential; these in turn may influence the electrostatic repulsion between the membrane and solute. First, the membrane zeta potential (ζ) was determined by using the electroosmotic flux method described in Section 2.3. The titanium dioxide/zirconium dioxide (TiO2/ZrO2) incorporated into the membrane carries a negative zeta potential at solution pH 4, 7, and 10. Results from the measurements were appeared in SI Figure S2 and Table S3. The ζ values at different pH environments were -23.1 mV (pH 10), -16.7 mV (pH 7), and -12.2 mV (pH 4) (Table 1), indicating that our membrane’s zeta potentials were negative at all pH levels studied. In addition, the higher the pH value, the more negative the ζ observed. This finding is consistent with an earlier study (10) that also used a TiO2/ZrO2 tubular ceramic membrane. Figure 3 depicts the three forces proposed to affect PFOX filtration: the electrostatic repulsion force that originates from membrane-solute interaction, the electrophoretic force arising from the presence of an electrical field, and the hydrodynamic force stemming from advective transport of PFOX, which in turn derives from the pressure gradient and
FIGURE 3. Proposed mechanism for removal of PFOA and PFOS anions during (a) microfiltration (b) electro-microfiltration. advective flow. In the absence of applied electrical field, the hydrodynamic force dominated PFOX transport through the membrane, as evidenced by the small change in PFOX removal efficiency when the membrane zeta potential changed (Figure 2). The minimal removal observed at pH 10 (Figure 2) was attributed to the membrane zeta potential. When an electrical field was applied, the resultant electrophoretic attraction force and the electrostatic repulsion force from the membrane zeta potential offset the hydrodynamic force. When an electrical voltage was applied at the same magnitude, a significant difference in PFOX rejection at various pH values was observed. The increased removal efficiency of PFOX with increasing pH value was attributed to high membrane zeta potential when solution pH increased from 4 to 10. Because acids, bases, and salts are used extensively to fabricate electronics, wastewaters from electronic industry can contain substantial amounts of electrolytes. High-ionicstrength wastewater may decrease the performance of EMF because of the high electricity consumption required. To understand the effect of ionic strength on PFOX rejection, EMF experiments were conducted in various solution environments, and results were compared with experiments conducted in the absence of background electrolytes. Figure 4 shows the effect of ionic strength on PFOX removal by EMF in a 58 V/cm electrical field. The electrolytes used were
Na2SO4, NaCl, NH4Cl or CaCl2, and concentrations ranged from 0.4 to 4.8 mM. Decreases in PFOX removal were observed as ionic strength increased, suggesting decreased membrane zeta potential due to compression of the electric double layer, in turn reducing the electrostatic repulsion force between the membrane and PFOX. In addition, PFOX removals in the following electrolyte solutions decreased (given in descending order): Na2SO4 > NaCl ∼ NH4Cl > CaCl2. This is consistent with the membrane zeta potentials previously measured for each solution (Table 1). 3.3. Effect of DOM on PFOX Removals and Filtration Flux. DOM is ubiquitous in wastewater. To explore the effect of water organic matrix on filtration performance, HA was used as background DOM and spiked into a solution containing 100 µg/L PFOX. Because DOM is a fouling agent that can cause flux decline during filtration, it is essential to monitor the flux during the filtration process. Figure 5a shows the normalized flux as a function of filtration time in the presence of HA at pH 7. A blank test with Milli-Q water demonstrated constant flux throughout filtration. In the absence of HA, no flux decline was observed during the MF of PFOX, with results similar to those from the blank test (data not shown). However, the flux decreased to 73% of its initial value during the MF of PFOX in the presence of 5 mg/L HA. Although the size of HA is smaller than the membrane’s pore size, its aggregation makes it one of the VOL. 44, NO. 20, 2010 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
7917
FIGURE 4. The ionic strength effect of Na2SO4, NaCl, NH4Cl or CaCl2 (0.4-4.8 mM) on the (a) PFOA (b) PFOS removal by electro-microfiltration (E ) 58 V/cm; pH 7). most common fouling agents found during MF filtration (28); alternatively, a highly concentrated layer of HA may form on the membrane surface during filtration. We found that the flux increased substantially and proportionally with an applied electrical field and plateaued within 25 min (20, 37, and 47% flux increases were observed after 120 min at electrical fields of 29, 43.5, and 58 V/cm, respectively). The increase in flux after electrical field application was attributed to electrophoresis, which pulled HA away from the membrane surface, thus reducing filtration resistance. In addition, the electroosmosis flux induced simultaneously contributed to the increase in the filtration flux. The flux during EMF of PFOX in the absence of HA at 58 V/cm was higher than the Milli-Q water flux by 25%. In the presence of an electrical field, the incremental flux below the Milli-Q water flux was mainly attributed to electrophoresis, while the flux exceeding the Milli-Q water flux was mainly ascribed to electroosmosis. To quantify the effect of electrical field strength on flux, we calculated the critical field strength (Ecritical, V/cm), that is, the electrical field strength that counterbalances the convective migration of charged species toward a membrane (29). Ecritical )
J µp
(3)
In eq 3, J denotes the permeate flux (cm/s), and µp (cm2/ (Vs)) is the electrophoretic mobility of the charged species. 7918
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 44, NO. 20, 2010
FIGURE 5. (a) Normalized flux in the presence of HA at various electrical field strengths at pH 7 (PFOS displays identical behavior). (b) Removal efficiencies of PFOX and DOC in the presence or absence of HA at pH 7. From eq 3, the Ecritical, HA was 41 V/cm, assuming an electrophoretic mobility of Aldrich HA’s µp value of 3.3 × 10-4 cm2/ (Vs) (23) and an averaged filtration flux of 1.35 × 10-2 cm/s. When the electrical field strength was less than the critical value, HA tended to move toward the membrane surface, resulting in a flux lower than with Milli-Q water (J/J0 ) 90%). Applying an electrical field close to Ecritical,HA enabled retention of charged compounds on the concentrate side, yielding a flux equal to that of Milli-Q water. Previous investigators have also noted that electroosmosis should be considered when the applied field strength exceeds the critical value (8). Figure 5b shows the PFOX and DOC removal efficiencies in the presence or absence of HA at pH 7. Without an electrical field, PFOX rejection was low (
