Sorption of perfluorinated compounds from ... - Springer Link

23 downloads 294 Views 181KB Size Report
Jan 12, 2010 - Results For perfluorooctane sulfonate and perfluorooctanoic acid Freundlich sorption coefficients, log KiF, for powdered activated carbon (PAC) ...
J Soils Sediments (2010) 10:179–185 DOI 10.1007/s11368-009-0172-z

SEDIMENT MANAGEMENT IN NORWAY • RESEARCH ARTICLE

Sorption of perfluorinated compounds from contaminated water to activated carbon Mona C. Hansen & Marion H. Børresen & Martin Schlabach & Gerard Cornelissen

Received: 9 October 2009 / Accepted: 11 December 2009 / Published online: 12 January 2010 # Springer-Verlag 2010

Abstract Introduction Perfluorinated compounds (PFC) are toxic and bioaccumulative compounds that are ubiquitous in the environment. It is important to develop effective techniques to remove PFC from water. This study is the first to investigate sorption of PFC to activated carbon (AC) at environmentally relevant nanogram per liter concentrations. Methods Batch AC sorption isotherms were measured for water from a contaminated groundwater well, for three perfluorosulfonates and five perfluoroacetic acids. Results For perfluorooctane sulfonate and perfluorooctanoic acid Freundlich sorption coefficients, log KiF, for powdered activated carbon (PAC) were 4.0 and 3.8 (ng/g)(ng/L)–n, respectively, and for granular activated carbon (GAC) were 2.7 and 2.3 (ng/g)(ng/L)–n, respectively. Sorption was nonlinear, with Freundlich n coefficients generally around 0.5. The KiF on both GAC and PAC were PFC chain-length dependant, with increasing number of carbon yielding increasing KiF. This chain-length dependence appeared stronger for perfluorosulfonates than for perfluoroacetic acids. Tests with short (10 min) adsorption times still yielded substantial PFC removal (20–40% for GAC, 60–90% for PAC) and revealed that AC is probably suitable for PFC Responsible editor: Gijs D. Breedveld M. C. Hansen : M. H. Børresen : G. Cornelissen (*) Norwegian Geotechnical Institute (NGI), Ullevaal Stadion, P.O. Box 3930, 0806 Oslo, Norway e-mail: [email protected] G. Cornelissen Department of Applied Environmental Sciences (ITM), Stockholm University, 10691 Stockholm, Sweden M. Schlabach Norwegian Institute for Air Research, P.O. Box 100, 2027 Kjeller, Norway

removal in flow-through systems. A perfluorinated polymer, Teflon, was also tested as a PFC removal agent but proved not to be effective for PFC-contaminated water purification. Keywords Activated carbon . Contaminated water . Isotherms . Perfluorinated compounds

1 Introduction Perfluorinated compounds (PFC) are industrially fabricated compounds used in many industrial and commercial processes, such as flame retardants, paints, and textiles. They are extremely stable compounds and resistant to transformation in the environment. In recent years, PFC has been found to be increasing in concentration throughout the environment, including water bodies as well as biota. The ecological and biological effects are only starting to be understood, but there are indications that they are potentially toxic. Hazardous concentrations for 5% are set by the Norwegian government to 25,000 ng/L (SFT 2007). No observable effect concentration (NOEC) values for 48-h immobility for Daphnia magna and Daphnia pulicaria were, respectively, 0.8 and 13.6 mg/L (Boudreau et al. 2003). Studies show that they can be harmful to human health (Johansson et al. 2009; Kovirova and Svobodova 2008; Kyunghee et al. 2008). The NOEC values are higher than the concentrations found in nature all over, but dispersion of anthropogenic contaminants in nature are not fortunate. The long-term effects are not yet fully discovered, like changes in hormones and sex. The background level of perfluorooctane sulfonate (PFOS) in surface water in Norway is 0.1–0.7 ng/L, and the level in untreated leachate from waste disposal sites is 4–67 ng/L (SFT 2005). Surface water polluted by industry in Germany has been shown to contain PFOS concentrations up to 4,390 ng/L (Schaefer 2006), surface water from fire drill area in Sweden

180

contained between 200–2,200 ng/L PFOS (Swedish EPA 2004), and open Great Lake in the USA have been shown to contain PFOS concentrations between 21 and 70 ng/L (Boulanger et al. 2005). Levels of perfluorooctanoic acid (PFOA) are in the same order of magnitude. Measurements of other PFCs are scarcer. On the whole, it can be concluded that the environmentally relevant range is in the order of nanogram per liter. Activated carbon (AC) is a manufactured type of charcoal, which is a strongly sorbing and has a large surface area, up to 1,000 m2/g AC, and primarily been used as a water cleaning agent. However, recently its usefulness for remediation of soils and sediments has also been documented (Zimmerman et al. 2004; Brändli et al. 2008). AC has proven to possess a strong sorption affinity for hydrophobic organic compounds such as polycyclic aromatic hydrocarbons (PAH) (Cornelissen et al. 2006). It is an open question how well it sorbs PFCs. To our knowledge, there have been three recent studies that indicate that AC adsorption is a promising treatment technique for the removal of PFC from water (Martial 2006; OchoaHerrera and Sierra-Alvarez 2008; Yu et al. 2009). The study by Ochoa-Herrera and Sierra-Alvarez (2008) evaluated the sorption of three PFC compounds onto granular activated carbon (particle diameter 0.85–1.70 mm), zeolite, and sludge. Adsorption isotherms were determined with spiked water samples, in the concentration range 15–150 mg L–1, which is three to six orders of magnitude above environmentally relevant concentrations. Of the three sorbates, AC showed the highest affinity for perfluorooctane sulfonate (PFOS; Freundlich KF =36.7–60.9 (milligrams PFC per gram sorbent)(milligram PFC per liter)–n. The study by Yu et al. (2009) was carried out with spiked samples of PFOS and PFOA in the concentration range 20–250 mg L–1 with powder (powdered activated carbon (PAC)), granular activated carbon (GAC), and anion-exchange resin. The sorption isotherms from the study by Yu et al. (2009) showed that GAC had the lowest capacity among the three adsorbents for both PFOS and PFOA, while PAC possessed the highest sorption capacity for PFOS of 1.04 mmol g–1, and the anion-exchange resin had the highest sorption capacity for PFOA of 2.92 mmol g–1, both according to the Langmuir fitting (Yu et al. 2009). The overall aim in the following study was to evaluate the sorption of PFC to AC. Sorption isotherms were determined to test the potential of AC for use in flowthrough systems. An additional aim was to evaluate Teflon as a PFC sorbent. In the present study, we evaluated PAC and GAC with field-sampled, PFC-contaminated water, in contrast to earlier studies that has been carried out with laboratoryspiked PFC. Hence, the PFC concentrations were in the environmentally relevant nanogram-to-microgram per liter range. The currently used concentrations of maximally

J Soils Sediments (2010) 10:179–185

1,000 ng/L, although still low, are about the maximum observed in the environment. Further, we focused on the relationship between adsorption and the PFC of different chain lengths, by measuring three perfluorosulfonates and five perfluoroacetic acids. In order to use AC for the purification of PFCcontaminated water, its adsorption has to be rapid. By also measuring short AC–PFC contact times, we attempted to evaluate the practical feasibility of AC adsorption for remediation via flow-through systems. Apart from AC, also Teflon has been tested as a sorbent in this study. Teflon is a perfluorinated polymer which is structurally related to PFC, increasing the likelihood of favorable sorbent–sorbate Vanderwaals interactions and correspondingly strong absorption. This is the first study that attempts to deploy Teflon as a PFC sorbent.

2 Methods 2.1 Water sampling Natively PFC-contaminated water was sampled at 9.5-m depth from a well downstream a factory where water-resistant clothing is fabricated (geographical coordinates 32 N6589900 E596499). The well was 15 m deep, with a groundwater level at 8.7 m. Water was pumped from the well until stable pH and temperature were reached. Water of 25 L was sampled (pH 6.7; conductivity 33 mS/m; temperature 15.0°C), in high-density polyethylene (HDPE) 1-L flasks (VWR, Kalbakken, Norway), which was used for adsorption studies with activated carbon. Eight different PFCs were detected in the well water; perfluorobutane sulfonate (PFBS), perfluorohexanesulfonate (PFHxS), PFOS, perfluorohexanoic acid (PFHxA), perfluoroheptanoic acid (PFHpA), PFOA, perfluorononanoic acid (PFNA), and perfluorodecanoic acid (PFDcA). The concentrations of the different compounds lie in the nanogram per liter to microgram per liter area, see Table 1. 2.2 Activated carbon Two different types of activated carbon were obtained from Clairs, Son, Norway. PAC, Silcarbon TH90 extra, originated from anthracite coal, with a fineness of 80%1,000 mg/g, the methylene blue adsorption was >25 g/100 g, and the specific surface area was ca. 1,100 m2/g. The GAC was also coal-derived (Silcarbon), with a fineness of 1%> 400 μm, 60% between 180 and 400 μm, 20% between 120 and 180 μm, 15% between 75 and 120 μm, and 4%< 75 μm. Average particle size was 226 µm. The iodine number of GAC was 1,250 mg/g, and specific surface area was 1,200 m2/g.

J Soils Sediments (2010) 10:179–185

181

Table 1 Original perfluorinated compounds and dissolved organic carbon concentration of the tested contaminated well water, with standard deviations in triplicate measurements Compound PFBS PFHxS PFOS PFHxA PFHpA PFOA PFNA PFDcA DOC

Concentration (ng/L) 73±11 470±80 1,400±200 280±40 320±40 1,400±130 67±5 40±20 5.27±0.12

PFBS perfluorobutane sulfonate, PFHxS perfluorohexanesulfonate, PFOS perfluorooctane sulfonate, PFHxA perfluorohexanoic acid, PFHpA perfluoroheptanoic acid, PFOA perfluorooctanoic acid, PFNA perfluorononanoic acid, PFDcA perfluorodecanoic acid, DOC dissolved organic carbon

2.3 AC sorption studies The sorption isotherm for each sorbent was measured for the undiluted PFC-contaminated well water, and for four dilutions (2, 5, 10, and 20 times diluted with non-PFC containing water from a similar well with 3–5 mg/L dissolved organic carbon (DOC)). PAC (0.02 g) or 0.1 g GAC was added to 800 ml total sample volume. One sample per concentration was shaken horizontally in the above mentioned HDPE flasks at 100 rpm at 20°C for 4 weeks. No adsorption of AC to the HDPE flask walls was observed. Only minor changes in pH (0.1–0.3 log units) were observed after dilutions and adding AC to the samples. After equilibration, the AC was allowed to settle for 4 days, after which the aqueous phase was decanted and the perfluoroalkyl substances (PFAS) concentrations in the water were measured. DOC contents of the water were measured after sorption, and 96–98% of the DOC was sorbed to the AC during the experiments. For AC sorption to be a feasible PFC remediation method, adsorption kinetics needs to be reasonably fast. Therefore, undiluted water samples (triplicates) were also shaken for only 10 min with AC, after which they were analyzed as described below.

2.5 Quality control Deionized water and AC were analyzed for PFCs, and levels were below detection. Blank tests with PFC (PFCcontaining water without AC; triplicates) were performed to check for evaporation and adsorption to the walls of the HDPE flasks and bacterial degradation. These samples were shaken in exactly the same manner as in the AC sorption isotherm studies (20°C; 28 days) and resulted in recoveries of 90–110%, showing that neither degradation, evaporation, nor HDPE adsorption occurred. 2.6 Sample preparation and analytical methods The samples were sent to Norwegian Institute for Air Research (NILU) for PFC analysis of the water samples. For sample preparation, the water samples were filtered with 142-mm Isolute SPE Accessories filters in order to collect the particles before the samples were extracted with 20 mL methanol (for gas chromatography, Suprasolv Merck). Prior to extraction, 0.1 ng/µL internal standard 13 C-labeled PFC was added to all samples. The extracts were concentrated to exactly 500 µL and further cleaned using 225 mg EnviCarb (Oasis HLP Plus SPE Cartridges). After centrifugation (1,677,000×g), 0.1 ng/µL BTPA recovery standard was added. For analytical methods, the PFCs were analyzed by high performance liquid chromatography (HPLC) HP1100, coupled to a mass spectrometer (HPLC-Q-MS) in electrospray ionization with time-of-flight (TOF) analyzer (LCT, Micromass). Injection volumes of 50 µl were used, with a flow rate of 200 µL/min and a column temperature of 30°C. The following gradient consisting of aqueous 2 mM ammonium acetate and methanol (MeOH) was applied: 0 min, 50% MeOH; 5 min, 85% MeOH; 10 min, 85% MeOH; 11 min, 99% MeOH; 20 min, 99% MeOH; 21 min, 50% MeOH; 28 min, 50% MeOH. The method detection limits for all analytes were determined on the basis of blank extraction experiments. The detection limit for PFBS, PFHxS, PFHxA, and PFNA was 0.1 ng L–1, for PFOS 0.3 ng L–1, for PFDcA 3 ng L–1, for PFHpA 4 ng L–1, and for PFOA 5 ng L–1.

3 Results and discussion 2.4 Teflon sorption studies 3.1 AC sorption isotherms Teflon was obtained in ∼1-kg cylinder-shaped blocks from Astrup AS (Norway). Strands of 55-µm thickness were sliced with high precision razor blade. Teflon (10 g; in 0.055-mm strings) was added to 1 L PFC-contaminated water and shaken for 3 days at 20°C, after which the water was analyzed as described below.

Two different models have been used to fit the experimental data to determine the adsorption capacity (Schwarzenbach et al. 2003). The Freundlich model :

n Cis ¼ KIF  C iW

ð1Þ

182

J Soils Sediments (2010) 10:179–185

where KIF is the Freundlich constant or capacity factor (nanogram PFAS per gram AC)(nanogram PFAS per liter)–n. Cis and Ciw are, respectively, the PFAS concentration on the solid (AC) and PFAS water concentration. n=Freundlich exponent. This model assumes multiple sites of adsorption working parallel with different free energies (Schwarzenbach et al. 2003). Langmuir model

Cis ¼

Qmax  KiL  Ciw 1 þ KiL  Ciw

ð2Þ

where Qmax is the total number of surface sites per mass of the sorbent. KiL is the Langmuir constant. This model assumes adsorption in a monolayer and a constant sorbate affinity for all surface sites (Schwarzenbach et al. 2003). Study of the Langmuir model was deemed relevant since the model is linear at low concentrations, and the AC is a homogeneous material. The Langmuir and Freundlich adsorption isotherms, derived using experimental data for PFAS compounds, are presented in Table 2 for PAC and Table 3 for GAC. Several compounds were below detection in the PAC-amended systems, and these are omitted from the analyses. The experimental data fitted best to the Freundlich model, and the Freundlich adsorption isotherms obtained for PFOA and PFOS with PAC and GAC are presented in Fig. 1. PAC generally showed better adsorption than GAC, and the sulfonates showed stronger sorption than the carboxylic acids. The surface area of GAC and PAC are comparable but probably better accessible for PAC (smaller particles). Either the equilibration times could be insufficient for GAC, or it could be more prone to fouling through pore throat clogging by native DOC in the well water. This “attenuation” of AC sorption in natural systems has often been observed before, especially in sediment-water systems (Cornelissen et al. 2006; Werner et al. 2006). An additional

explanation is that the rigidity of the CF2 backbone may render its sorption to the inner pore surface area of GAC energetically unfavorable. 3.2 Relation KiF and chain length By studying PFC compounds with different chain lengths, it could be investigated whether the AC adsorption is affected by the number of CF2 units. Values of log KiF (in nanogram per liter units (nanogram per gram)(nanogram per liter)–n) are plotted against the chain length (number of carbon atoms) in Fig. 2 for PAC and Fig. 3 for GAC. For those PFC–AC combinations where, out of the dilution series of five samples, only two samples were obtained above the detection limit, the log KiF values were calculated from these two. For PFNA– PAC, there was only one measurement above the detection limit. The measured value was extrapolated to 1 ng/L using the average n value for PAC of 0.61. The results point to an increase in KiF value with increasing chain length for the PFC for both GAC and PAC. Figure 2 indicates a larger increase in KiF with chain length for PFS then for PFA; however, due to the large standard deviations in the slopes and the low number of data points, this difference was not significant (t test 95%). PFOA had significantly higher sorption (t test, 95%) to GAC than the other evaluated perfluorinated carboxylic acids (PFA). Similar results were reported by Ochoa-Herrera and Sierra-Alvarez (2008). We have no clear-cut explanation as to why the perfluorinated sulfonic acid (PFS) sorption to AC is more strongly chain-length dependent than the PFA sorption. Influence of micelle formation is probably unlikely since our concentrations were so low (in the nanogram per liter range), but some study show that even for low concentrations, this could happen (Cheng et al. 2009). Apparently, the hydrophobic

Table 2 Powdered activated carbon: Langmuir isotherm constants Qmax(nanogram per gram) and KiL(liter per nanogram) coefficients and Freundlich isotherm constants n and KiF (nanogram per gram)(nanogram per liter)−n coefficients, derived from experimental data Contaminant

Chain length

Measured values (this study) PFOS 8 PFHxS 6 PFOA 8 Literature values PFOSa 8 PFOAa 8

Sorbent

Langmuir isotherm

Freundlich isotherm

Qmax

KiL

r2

log KiF

n

r2

PAC PAC PAC

10,000±18,000 13,000±5,000 20,000±50,000

2±8 10±6 0.4±1.4

0.05 0.95 0.33

4.0±0.2 3.99±0.03 3.82±0.14

0.9±0.5 0.32±0.03 0.6±0.2

0.53 0.99 0.81

PAC PAC

4.53×108 2.67×108

0.024 0.026

0.84 0.91

6.60 5.79

0.18 0.28

0.87 0.97

PFHxS perfluorohexanesulfonate, PFOS perfluorooctane sulfonate, PFOA perfluorooctanoic acid a

Yu et al. (2009); log KiF values extrapolated over eight orders of magnitude

J Soils Sediments (2010) 10:179–185

183

Table 3 Granular activated carbon: Langmuir isotherm constants Qmax− (nanogram per gram) and KiL(liter per nanogram) coefficients and Freundlich isotherm constants n and KiF (nanogram per gram)(nanogram per liter)−n coefficients, derived from experimental data Contaminant

Chain length

Measured values (this study) PFOS 8 PFHxS 6 PFNA 9 PFOA 8 PFHpA 7 PFHxA 6 Literature values PFOAa 8 a PFOS 8 PFOAb PFOSb

8 8

Sorbent

Langmuir isotherm

Freundlich isotherm

Qmax

KiL

r2

log KiF

n

r2

GAC GAC GAC GAC GAC GAC

5,300±1,400 600±180 190±30 1,100±150 55.6±1.1 220±80

0.08±0.02 0.09±0.05 0.5±0.3 0.030±0.009 1.1±0.4 0.04±0.02

0.99 0.87 0.92 0.94 0.89 0.98

2.73±0.05 1.8±0.2 1.90±0.05 2.28±0.11 1.68±0.04 1.2±0.2

0.50±0.03 0.50±0.13 0.25±0.04 0.29±0.05 0.04±0.03 0.55±0.17

0.99 0.82 0.97 0.93 0.71 0.92

GAC GAC

1.55×108 1.613×108

0.0079 0.017

0.97 0.96

7.07 6.92

0.18 0.20

0.96 0.95

F400 F400

1.12×108 2.36×108

3.8×10−8 1.24×10−7

0.97 0.96

4.41 6.05

0.443 0.289

0.96 0.97

PFHxS perfluorohexanesulfonate, PFOS perfluorooctane sulfonate, PFHxA perfluorohexanoic acid, PFHpA perfluoroheptanoic acid, PFOA perfluorooctanoic acid, PFNA perfluorononanoic acid a

Yu et al. (2009); log KiF values extrapolated over eight orders of magnitude.

b

Ochoa-Herrera and Sierra-Alvarez (2008); log KiF values extrapolated over six orders of magnitude.

perfluorinated chain influences overall interaction with the AC more strongly in the presence of an SO3− group than in the presence of a CO2– one.

The correlations to the Langmuir model were slightly better than to Freundlich model for GAC adsorption. PAC adsorption fitted better to Freundlich model. However, we favor the Freundlich model over the Langmuir one since our concentration range is so far below the Langmuir Qmax, reducing the possibilities to derive adequate values for this parameter.

Earlier studies were in a much higher concentration range, which makes comparison difficult, as extrapolation of the Freundlich KiF values over many orders of magnitude is necessary for this. The high-concentration literature studies probably yield more accurate Langmuir Qmax values and lower Freundlich n values, since the AC was closer to saturation. Of the different types of GAC evaluated by Ochoa-Herrera and Sierra-Alvarez (2008), the Falcon F400 GAC (particle diameter 0.85–1.70 mm) was the most effective sorbent for PFOS. Yu et al. (2009) evaluated GAC (particle diameter 0.9– 1.0 mm) as well as PAC (10 µm), for PFOS as well as PFOA.

Fig. 1 Sorption isotherms of PFOA and PFOS onto powdered activated carbon and granular activated carbon

Fig. 2 The relation between log KiF and chain length of the perfluorinated sulfonates (filled and empty triangle) and perfluorinated acids (filled and empty square). Closed symbols for values based on full isotherms, open symbols for values derived with less than three observations, extrapolated to 1 ng/L. All for powdered activated carbon

3.3 Comparison with earlier studies

184

J Soils Sediments (2010) 10:179–185 Table 4 Kd (liter per gram) values for perfluorinated compounds with powder activated carbon (PAC) and granular activated carbon (GAC) at native concentration Compounds

Fig. 3 The relation between log KiF and chain length of the perfluorinated sulfonates (filled and empty triangle) and perfluorinated acids (filled and empty square). Closed symbols for values based on full isotherms, open symbols for values derived with less than three observations, extrapolated to 1 ng/L. All for granular activated carbon

Probably because of the different concentration ranges, literature values for the Langmuir parameters tend to be vastly different from ours, with higher capacities Qmax and lower affinities KiL. Comparison of the Freundlich parameters indicates that the literature values show higher K values and lower n values. This seemingly stronger sorption is probably caused by the extrapolation of K values from the milligram per liter to the nanogram per liter range using low n values; these extrapolations were generally as high as 4–6 log units, rendering the comparison of our current values to literature values difficult. Another reason for the discrepancy between our current values and literature values might be that our water contained DOC that may have clogged or competed for AC sorption sites, reducing sorption compared to the literature studies in pure water. 3.4 AC sorption at 10-min shaking The results from sorption studies after 10-min shaking are presented in Table 4 with Kd values for the PFC at native concentrations. The Kd values for PAC sorption are much higher than for GAC. The differences are larger than for the isotherm study, which employed a 4-week equilibration time. This indicates that the PFC sorption to GAC is not just lower than for PAC but also slower. An earlier study on sorption kinetics shows that it required a longer time to reach the sorption equilibrium for GAC (0.9–1.0 mm) than for PAC (