Environ. Sci. Technol. 1989,23, 1302-1306
NOTES Effect of Suspended Sedlment Concentration on the Sediment to Water Partition Coefficient for 1,3,6,8-Tetrachtorodibenzo-p-dioxin Mark I?.Servos" and Derek C. G. Muir
Freshwater Institute, Canada Department of Fisheries and Oceans, 50 1 University Crescent, Winnipeg, Manitoba R3T 2N6 Canada The sediment to water partition coefficient (K,) for 1,3,6,8-tetrachlorodibenzo-p-dioxin was determined for suspended sediment concentrations ranging over 4 orders of magnitude. The truly dissolved (free) concentrations were estimated by four different methods: centrifugation at 6000g for 15 min, centrifugation at 20000g for 30 min, reverse-phase C18 cartridges, and dynamic headspace analysis. Centrifugation at either 6000g or 2oooOg resulted in a negative relationship between log Kp and log (suspended sediment concentration) (SSC) (i.e., slopes of -1.46 and -1.54), but both the C18 cartridge and dynamic headspace methods resulted in significantly less negative slopes (-0.46, -0.65). Dissolved organic carbon inflated the apparent free water concentations, leading to an underestimation of Kp' However, measuring the truly dissolved concentrations did not completely eliminate the negative correlation between log K p and log (SSC). Another mechanism such as particle interaction or methodological bias may explain the deviation from zero slope predicted by a linear two-phase sorption equilibrium.
Introduction The extent to which hydrophobic organic pollutants sorb to aquatic sediments will dramatically influence their environmental fate. Accurate determination of the sediment to water partition coefficient (K,) is critical for modeling the fate and potential hazard of chemicals in aquatic ecosystems. Sorption of pollutants may affect their availability for biological uptake, volatilization, sedimentation, or reaction (1-3). Based on the assumption of a linear two-phase sorption equilibrium, Kp should be constant with increasing sediment concentrations. However, numerous authors have shown that Kp usually declines when sediment concentrations increase (4-8). This inconsistency led DiToro (4) and Mackay and Power (9) to hypothesize a particle interaction induced desorption of chemicals from suspended sediments, which is intensified as the suspended sediment concentration increases. An alternative hypothesis put forward by Gschwend and Wu (10) and Voice and Weber (11) explains the decrease in Kp as an artifact of the methodology used. In particular they suggested that sorption of the chemical to a third phase (colloids) results in overestimation of the concentration in the aqueous solution. These authors showed that as the sediment concentrations increase there is a corresponding increase in the concentration of dissolved organic carbon. It has been shown that dissolved organic carbon can sorb hydrophobic *Present address: Lakes Research Branch, National Water Research Institute, Environment Canada, 867 Lakeshore Road, Burlington, Ontario, Canada L7R 4A6. 1302
Environ. Sci. Technol., Vol. 23, No. 10, 1989
chemicals and inflate their apparent aqueous solubilities (12-14). It would appear that a third phase would explain at least part of the Kp versus suspended sediment effect. To date, measurement of third-phase and particle interaction effects have been conducted by centrifugation, and there has been no independent assay of the truly "free" aqueous concentration of hydrophobic chemicals in these sediment-water systems. Landrum et al. (15) described a reverse-phase cartridge method to directly measure truly dissolved (free) concentrations. Yin and Hassett (16) also recently described a dynamic headspace method to measure truly dissolved water concentrations. Both of these methods have been employed in this study to determine the Kp of 1,3,6,8tetrachlorodibenzo-p-dioxinin a sediment-water system and compared directly to determinations of Kp using conventional centrifugation techniques.
Experimental Section 14C-labeled1,3,6,8-tetrachlorodibenzo-p-dioxin (TCDD) with a specific activity of 24.16 mCi/mM was supplied by Pathfinder Laboratories (St. Louis, MO) and purified before use by thin-layer chromatography. The radiopurity of TCDD was determined to be >99.9% by reverse-phase HPLC (85% methanol/water, 1.5 mL/min). TCDD is a very hydrophobic compound with a reported log Kowof 7.13 (17). All water used was distilled deionized water passed through a Milli-Q and 0.2-pm filter system (Waters Scientific). Milli-Q water contained C0.24 mg/L dissolved organic carbon (DOC) measured by a high-temperature acid persulfate digestion, followed by infrared detection of COz on a 01 Corp. Model 700 carbon analyzer. Sediments were collected from the sublittoral zone (2 m) of Lake 304, Experimental Lakes Area (Canada Department of Fisheries and Oceans), in northwestern Ontario. Sediments were freeze-dried, sieved through a 100-mesh screen and stored at 4 "C. Sediments were 25.2% total organic carbon and 2.4% total nitrogen and had a clayey texture with 12% sand, 37% silt, and 51% clay. A preweighed sediment sample was added to 1 L of Milli-Q water in a tall-form gas-washing bottle. TCDD in methanol (10 pL) was spiked directly into the water column and allowed to equilibrate for 24 h with constant mixing by a Teflon stir bar. Ten microliters of methanol is not expected to have an effect on the partitioning behavior of TCDD (18). A t the start of each experiment four 20-mL water samples were collected from a depth of 5 cm below the water surface and placed in 25-mL Corex tubes. A 4-mL water sample was taken directly from each tube immediately after being mixed to prevent settling of suspended sediments. Samples were diluted with scintillation fluor (Atomlight, New England Nuclear) and assayed by liquid
0013-936X/89/0923-1302$01.50/0
0 1989 American Chemical Society
scintillation counting (LSC). Sorption of TCDD to the glass walls of the tubes in the presence of suspended sediments was minimal (Le., the concentrations of total TCDD measured in the tubes were not different from that in the experimental apparatus). The vertical distribution of radioactivity and suspended sediments was checked by sampling at four depths, and both varied by less than 10% in all cases. Two samples (i.e., tubes) were centrifuged at 6000g for 15 min and two samples centrifuged at 20000g for 30 min to remove the particulate fraction. Aliquots of the supernatant water (4 mL) were assayed directly by LSC to determine the proportion of radioactivity “in solution“. The proportion of radioactivity associated with the dissolved organic matter (DOM) was determined according to the method described by Landrum et al. (15). A 4-mL sample of the supernatant (20000g) was passed through a reverse-phase cartridge (cl8 SepPak, Waters Scientific) and the eluant assayed by LSC. The TCDD associated with the DOM will pass through the column while the truly dissolved TCDD will partition to the c18 and remain on the column (15). Greater than 95% of the DOC from Lake 304 epilimnetic water was determined to pass through the cartridge (14),similar to that reported by Landrum et al. (15). The precision of measuring the concentration of [14C]-TCDDin the various phases was approximately 20, 10, and 4% for SepPak extractable, centrifuged, and total, respectively. The free water concentration was also determined by the dynamic headspace method described by Yin and Hassett (16). The column was sparged with N2 (zero grade presaturated with HzO)with a coarse glass frit 25 cm below the water surface. Flow rates were approximately 300 mL/min and were measured with a bubble flow meter. Dunnivant et al. (19)reported that a column height of only 8 cm, with a flow of 1 L/min was adequate to achieve equilibrium of dichlorobiphenyls between the gas and liquid phases. Yin and Hassett (16) found that a 14-cm height was adequate for mirex to reach equilibrium at a flow rate of 560 mL/min. A 3 X 0.6 cm column of 60180 mesh Tenax-GC was placed into the outflow port of the headspace. After 3 h, the Tenax was removed, placed in a glass test tube, and later combusted on Packard 306 sample oxidizer. Recovery efficiency of TCDD from Tenax was determined to be 88.1 f 2.0%. Ninety-nine percent of the 14Ctrapped was on the first centimeter of the Tenax column. The Henry’s law constant (HLC) was determined in Milli-Q water by determining the removal rate of TCDD by sparging and using the model described by Mackay et al. (20)where HLC = -D$TV/G (1) and D, is the depletion rate constant, R is the ideal gas law constant, T = 283 K, V = 1 L, and G is the flow rate of the gas. The depletion rate constant was estimated by nonlinear least-squares regression (SAS-NLIN) from plots of cumulative nanograms of TCDD trapped on Tenax versus time (t). The HLC for TCDD was calculated to be (5.9 f 1.8) X lo4 atm m3/mol (N = 9) at 10 “C. Webster et al. (21)reported the HLC for TCDD to be 6.8 X atm m3/mol a t 23 “C, and Podoll et al. (22)estimated a value of 1.6 X loa atm m3/mol for 2,3,7,8-TCDD at 25 “C. The concentration of TCDD free in solution was then calculated as free = [trapped]RT/Gt(HLC) (2) The sediment-water partition coefficient ng/L bound/kg/L suspended sediments Kp = (3) ng/L “in solution”
Table I. Comparison of log K , versus log (Suspended Sediment Concentration) Regressions’
6om? 20000g CIS cartridge sparging
R2
zero intercept
slope
0.96 0.96 0.74 0.84
6.15* 6.21* 5.98* 6.23*
-1.46* -1.54* -0.46** -0).65**
Values followed by similar symbols are not significantly different (p < 0.01).
was determined four times for each experiment by use of different values for the concentration “in solution”: after centrifugation at 6000g or 20000g, using reverse-phase (218 cartridge, and gas,sparging. The log Kp versus log (SSC) regressions were compared by analysis of covariance (SAS-GLM). Kp estimated from DiToro’s equation (8) was calculated from Kp = fK&/[1+ (2E)mfKdI (4) where f is the organic carbon mass fraction, Kd is the organic carbon partition coefficient estimated from KO, by using the relationship reported by DiToro (4),2E = 0.7 where E is the collision efficiency term, and m is the concentration of suspended sediment. The collision efficiency term (E)was estimated for the Kpcalculated in this study by using the free water concentrations determined by fitting eq (4) by use of SAS-NLIN. Kp was also estimated from the equation described by Gschwend and Wu (10) KpObsd= K,t’ue/(l + K,joctNeDOC) (5) assuming that KO, = K, = KdW. Solubility enhancement by DOC was examined independently with a technique similar to that described by Chiou et al. (12). TCDD (26 ng) was spiked onto the walls of 25-mL Corex glass tubes. This amount is equivalent to 1300 ng/L or 4.1 times the reported water solubility (23). The solvent was evaporated and 20 mL of a Milli-Q-Aldrich humic acid solution added. Aldrich humic acid was precipitated twice with HC1, dialyzed in distilled water, and filtered through a 0.45-pm Millipore filter. The solution was shaken for 24 h and centrifuged at 20000g for 30 min; a 4-mL sample was taken and assayed by LSC. A 4-mL aliquot was passed through a cl8 cartridge as described previously to determine the concentration in true solution. Results and Discussion The Kp for TCDD declined as the suspended sediment concentration increased when conventional centrifugation techniques (Figure 1)were used. O’Connor and Connolly (1) and others (4-8) had previously observed this effect for other pollutants. Centrifugation at 2oooOg for 30 min resulted in the same relationship of log Kpversus log (SSC) as did centrifugation a t 6000g for 15 min (Table I). Removing the supernatant and centrifuging a second time also did not change the apparent aqueous concentration and therefore Kp’ Gschwend and Wu (10)also found that centrifugation at higher speeds and longer times (760g for 20 min vs 1700g for 60 min) did not affect Kp at the low suspended sediment concentrations used in this study (
