Environ. Sci. Technol. 2006, 40, 6668-6673
Partitioning of CPs, PCDEs, and PCDD/Fs between Particulate and Experimentally Enhanced Dissolved Natural Organic Matter in a Contaminated Soil SOFIA FRANKKI,† YLVA PERSSON,‡ MATS TYSKLIND,‡ AND U L F S K Y L L B E R G * ,† Department of Forest Ecology, Swedish University of Agricultural Sciences, SE-901 83 Umeå, Sweden, Environmental Chemistry, Department of Chemistry, Umeå University, SE-901 87 Umeå, Sweden
We determined the distribution of hydrophobic organic contaminants (HOCs) to fractions of natural organic matter in a soil contaminated by chlorophenol wood preservatives more than 30 years ago. The concentration of dissolved organic matter (DOM) was enhanced in soil suspensions by raising pH to 6.8-9.1. After 48 h of desorption/equilibration, the DOM fraction was separated from the particulate organic matter (POM) of the soil by filtration. In the next step, DOM was flocculated by Al-nitrate, and free concentrations of HOCs were determined in the aqueous phase. The HOCs associated with DOM and POM were extracted with toluene. No significant differences in gross carbon chemistry were detected between DOM and POM, using X-ray photoelectron spectroscopy (XPS). Normalized to organic C, chlorophenols (CPs) showed a similar degree of partitioning between DOM and POM, whereas the partitioning of polychlorinated diphenyl ethers (PCDEs), polychlorinated dibenzo-p-dioxins, and furans (PCDD/Fs) was highly shifted toward POM. The partitioning to POM, relative to DOM, increased in the order PCDE < PCDF < PCDD, reflecting the hydrophobicity of the compounds.
Introduction Use of chlorophenol (CP) formulations for wood impregnation has led to soil contamination at industrial sites. Technical chlorophenol formulations contain a number of chlorinated compounds, formed as byproducts in the production, such as polychlorinated dibenzo-p-dioxins (PCDDs), polychlorinated dibenzofurans (PCDFs), polychlorinated phenoxy phenols (PCPPs), and polychlorinated diphenyl ethers (PCDEs) (1). There is great concern to which extent these contaminants are mobile or retained in the soil at industrial sites where chlorophenol formulations have been used. The PCDE and PCDD/Fs are often designated hydrophobic organic compounds (HOC), and together with the more water soluble CPs they cover a range of physicochemical properties such as water solubilities and n-octanol-water partitioning (KOW). The log KOW of the compounds shows the * Corresponding author phone: +46-(0)90-7868460; fax: +46(0)90-7868163; e-mail:
[email protected]. † Swedish University of Agricultural Sciences. ‡ Umeå University. 6668
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 40, NO. 21, 2006
following ranges: CPs 2.2-5.1 (2, 3), PCDEs 5.5-8.2 (4), PCDFs 6.1-8.0 (2), and PCDDs 6.6-8.2 (2). The HOCs are well-known to associate primarily with soil organic matter (SOM) via a hydrophobic partitioning interaction (5). Soil organic matter is present in the solid phase of the soil as particulate organic matter (POM) and in the aqueous phase as dissolved organic matter (DOM). The POM and DOM fractions are operationally defined and separated either by the pore size of a filter or a defined speed of centrifugation. Because of this, the DOM fraction in addition to truly dissolved macro- and biomolecules also contains nanoparticles and colloids. The physicochemical composition and concentration of POM and DOM are constantly changing, due to input of litter from plants, decomposition by microorganisms, sorption-desorption, and leaching processes. The POM fraction may be regarded as a potential pool for DOM (6). Experimentally, the solubility of SOM can be manipulated and the POM fraction may be mobilized as DOM by increasing pH and by removing flocculating di- and polyvalent cations (7). Reproducible experimental systems in which manipulated concentrations of POM and DOM, as well as compounds associated to these two fractions, reach an (pseudo-) equilibrium within days have successfully been used to determine the binding and partitioning of organic compounds to POM and DOM fractions (8, 9). Despite a consensus about the importance of SOM for mobilization and retention of HOCs in soil, experimentally determined KOC values are nonexisting for PCDE and limited for PCDD/Fs. This holds both for DOM (KDOC) and for POM (KPOC) and reported KOC values are almost exclusively estimated from relationships with KOW. In seawater from the Baltic Sea, KPOC values for PCDD/F associations to particulate matter were determined to be 2-3 orders of magnitude greater than KDOC (10). The importance of dissolved humic acids as a transport agent for HOCs has been demonstrated by enhanced leaching of PCDD/Fs from fly ash (11). In addition there are a few measured KOC values reported for PCDD/Fs associations to humic and fulvic acids (12-15), and to organic carbon in sea sediments (16). For CPs, more data are available, and in a recent study using an experimentally manipulated POM-DOM system, KPOC was shown to increase with an increasing degree of chlorination, whereas KDOC remained constant and smaller (9). There is a fundamental difference between adsorption experiments in which added organic compounds are allowed to equilibrate for hours to days, and investigations of contaminated soils in which organic compounds have been “aged” for years or decades. The aging process is known to cause a hysteresis between adsorption and desorption (17). Studies of interactions between HOCs and SOM in aged contaminated soils are very few. The objective of this study was to determine concentrations of CPs, PCDEs, PCDFs, and PCDDs associated to fractions of POM and DOM, and free concentrations of the compounds, in an experimental system using a soil contaminated by the compounds of interest 1964-1975. The pH was raised in order to mobilize and release DOM enough to determine HOCs associated to DOM. Thus, the DOM fraction represents a potentially mobile pool of SOM. The partitioning between POM and DOM will give insight into the likelihood for contaminants to become mobilized and/or for the soil to be remediated. The separation of free analyte in aqueous phase and analyte associated with DOM was obtained by flocculation of DOM by Al3+ ions, a method that has been successfully used to determine KOC for the binding of PAHs to dissolved humic acids (18). 10.1021/es0608850 CCC: $33.50
2006 American Chemical Society Published on Web 09/21/2006
Experimental Soil and Site. A soil contaminated by a technical chlorophenol formulation was sampled at Sikeå (59° 32′ N, 12° 37′ E) in north Sweden. At the site, chlorophenols were used to treat wood during the years 1964-1975. The mean annual air temperature was 2.6 °C and mean annual precipitation 560 mm during the period 1961-1990. At a place with previously detected chlorophenol contamination, a bulk sample was collected with a spade at 5-30 cm depth. The whole bulk sample was passed through a 2 mm sieve, homogenized and stored in a glass jar at -20 °C until further use. The soil profile at the site is disturbed by industrial activities and wood fiber is mixed into the soil. The soil can not, therefore, be classified according to common soil classification systems, but the undisturbed soil adjacent to the site is a Spodosol (19) developed in glacial till formed from gneissic bedrock. The soil pH (H2O) was 5.8. Extractions and Separation of DOM, POM, and Aqueous Phase Analytes. To enhance the signal of analytes associated with DOM, we used alkaline conditions to extract DOM. To evaluate possible effects of pH four different levels were used: pH 6.8, 7.1, 7.7, and 9.1. For pH 7.7, three replicates were used, resulting in a total of six soil suspensions. Thirty gram of moist soil (corresponding to 23 g dry mass) was mixed with 100 mL of 10 mM NaCl. The pH was adjusted by addition of 1 M NaOH and the suspensions were incubated for 48 h at room temperature (20 °C) on a reciprocal shaker. The pH was measured by a pH meter (AZ Instruments, Taiwan). A 0.7 µm filter (GF/F Whatman) combined with a 2.7 µm (GF/D Whatman) pre-filter was used to separate the solid and liquid phases. Glass fiber filters were chosen to minimize possible interactions with the compounds of interest. We used a pore size of 0.7 µm in order to avoid severe clogging of the filters by the highly concentrated DOM solutions (600-1500 mg C L-1. The soil suspensions were pored onto the filters in a Bu ¨ chner funnel and filtration was conducted under vacuum. The soil captured by the filter was rinsed with Milli-Q water until the filtrate was colorless (by visual inspection). The filtrate (including the leachates from rinsing steps) was weighed and a sub sample was taken for total organic carbon (TOC) analysis (TOC analyzer, Shimadzu 5000) to quantify released DOM. Thus, the DOM fraction was defined as SOM passing a 0.7 µm GF/F filter and POM was the fraction of SOM not passing the filter. The DOM in the filtrate was flocculated by an addition of Al(NO3)3 (corresponding to approximately 2 mM Al) and enough NaOH to keep pH at 8. At this pH, approximately 99% of soil extracted DOM is re-flocculated by Al (20). The flocculated DOM was thereafter separated from the free dissolved analytes by filtration through a 0.7 µm GF/F filter under vacuum. The remaining solution was analyzed for TOC and for free dissolved analytes. The filters with captured POM and DOM, respectively, were stored at -20 °C until further extraction. All glassware used was cleaned with toluene (glass distilled, Burdick and Jackson, Muskegon, MI). Soil Organic Matter Characterization. The loss on ignition (LOI) was determined to be 13 mass-% of dry soil after 6h at 550 °C. Total organic C was determined in POM and DOM samples (TOC analyzer, Shimadzu 5000, Japan), after treatment with 1.0 M HCl, before and after heating at 375 °C. The latter procedure removes non-pyrogenic OC and gives an estimate of black carbon (BC), following the procedure of Gustafsson et al. (21). In POM total OC was 5.7% and BC was 0.6% of total OC. X-ray photoelectron spectroscopy (XPS) was used to characterize the gross carbon chemistry of POM and DOM. Spectra were collected with an electron spectrometer (Kratos Axis Ultra) using a monochromated Al KR source operated at 180 W. To compensate for surface charging, a low-energy
electron gun was used. The binding energy (BE) scale was referenced to the C 1s line of aliphatic carbon, set at 285.0 eV. The spectra were processed using Kratos software. Four binding energies were quantified for carbon: 285 eV (representing C-C and C-H bonds), 286.5 eV (C-OH, C-N and C-O-C bonds), 288 eV (CdO and N-CdO bonds) and 289.3 eV (O-CdO bonds) (22). The precision for each energy region was ( 2 atomic %. Analytical Procedures. Extraction, clean up, and fractionation of CPs, PCDEs, and PCDD/Fs was performed according to Persson et al. (23). In short, a sub-sample of 1 g of POM, and the total mass of DOM, flocculated by Al, was extracted using Soxhlet-Dean-Stark equipment with toluene as solvent. Prior to the extraction the samples were acidified to increase the extraction efficiency of the phenolic compounds and to dissolve Al(OH)3 formed during flocculation of DOM. The water phase containing dissolved compounds (after Al flocculation) was acidified (pH < 3) and extracted by liquid-liquid extraction using dichloromethane. Internal standards (IS) were added after Soxhlet extraction of DOM and POM samples, but prior to extraction of the aqueous phase filtrate. In the CP analysis, 13C-labeled mono-, di-, tri-, tetra-, and pentachlorophenol were used (Cambridge Isotope Lab, Andover, MA). In the PCDE analyses, Ten 13C-labeled PCBs (Cambridge Isotope Lab, Andover, MA) were used as IS. In the PCDD/F analysis, 16 13C-labeled 2,3,7,8-substituted PCDD/Fs (Wellington Laboratories, Ontario, Canada) were used as IS. 13C labeled PCBs no. 97 and no. 188 (Cambride Isotope Lab, Andover, MA) were used as recovery standards (RS) in the analyses of CPs and PCDEs. 1,2,3,4-TCDF and 1,2,3,4,7,8,9-HpCDF were used as RS in the PCDD/F analyses (Wellington Laboratories, Ontario, Canada). Recovery standards were added after cleanup, prior to instrumental analysis to account for losses during cleanup. The CPs were separated from the toluene extracts using lithium hydroxide followed by acetylation using acetic anhydride. PCDEs and PCDD/Fs went through a cleanup using a multilayered silica column (treated with sulfuric acid and potassium hydroxide, respectively). The PCDEs were separated from PCDD/Fs on a carbon column. The CPs and PCDEs were analyzed with gas chromatography coupled to mass spectrometry (HRGC/LRMS) (Fisons GC 8000 coupled to a Fisons MD 800 mass selective detector) instrument equipped with a DB-5MS column (J&W Scientific, CA) 30 m × 0.25 mm and 0.25 µm film thickness. For determination of PCDD/F, a HRGC/HRMS instrument (HP 6890 coupled to a Waters Autospec) with a resolution of 10 000 was used. The GC was equipped with a 60 m × 0.32 mm ID (DB-5) column with 0.25 µm film thickness (J&W Scientific, CA). Regarding PCDD/Fs both 2,3,7,8-substituted PCDD/Fs and non 2,3,7,8-substituted PCDD/Fs were included in the calculations. Non 2,3,7,8-substituted PCDD/Fs were identified based on retention order determined on a DB-5 column and using a fly ash sample containing most PCDD/F congeners that were run in connection with the sample presented here. The quality of the chemical analyses was assured by monitoring of two ions in the single ion monitoring (SIM) in the mass spectral analyses; the ratio of the two ions was within 15% of the theoretical chlorine cluster. Signal-to-noise ratio was set to 3. The amounts of compounds associated to DOM and POM were normalized to organic carbon, while the free analytes in the water samples are reported as concentrations in the soil suspensions. KOC values were calculated for specific congeners determined both in the aqueous phase and in association with DOM (KDOC) and with POM (KPOC), respectively. Detection limits were for CPs 0.05µg/L (aqueous phase) and 0.5 g/g OC (DOM and POM), for PCDEs 0.05 ng/L (aqueous phase) and 30 ng/g OC (DOM and POM) and for PCDD/Fs 0.1ng/L (aqueous phase), 1.5 ng/g OC (DOM) and 50 ng/g OC (POM). VOL. 40, NO. 21, 2006 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
6669
Results and Discussion Effects of pH and Incomplete Phase Separation. The methodology used to enhance DOM concentrations and to separate POM and DOM associated compounds may have affected the partitioning of CPs and HOCs. We recognize three possible effects that need to be discussed: (1) pHeffects, (2) effects of incomplete phase separation, and (3) effects caused by the Al-flocculation step. The increase of pH enhanced the soluble SOM fraction (designated DOM) from 0.6 g DOC × L-1 at pH 6.8 to 1.5 g DOC × L-1 at pH 9.1, corresponding to 5-13% of the total OC content of the soil. Our intention was to compare the partitioning of HOCs to POM and DOM fractions as a function of pH and DOC concentrations. However, because no significant differences in KDOC or KPOC were observed in relation to pH or DOC concentrations, the POM-DOM suspensions obtained at pH 6.8, 7.1, 7.7, and 9.1 were treated as six replicates. It is well-known that DOM becomes less condensed with increased pH owing to a dissociation of mainly carboxyl and phenol functional groups. This could in turn affect the binding affinity of more polar organic compounds and decrease the partitioning of HOCs. The fact that we could see no apparent effects on KDOC and KPOC values for any of the studied compounds indicates that changes in DOM and POM conformations in the pH range had only small effects on the binding and partitioning. Even if the procedure used to release DOM partly may resemble the method recommended by IHSS for extraction of humic substances from soil (24), pH was kept lower in order to minimize irreversible effects on OM structure, and no means were undertaken to purify the DOM extract. Determinations revealed that approximately 3% of DOC remained in solution after the Al-flocculation step. Because of this incomplete phase separation, free concentrations of analytes could be biased if analytes in association with the remaining DOM were released in the liquid-liquid extraction step, as recognized by Schrap et al. (25) and by Lee et al. (26). A theoretic calculation, based on concentrations of analytes associated to the flocculated DOM and the concentration of remaining DOC, revealed that the free concentration of the CPs, in the worst case, only would have been increased by an insignificant amount (
