PCB-containing paint used in the bioavailability evaluation
The chlorinated rubber marine base paint used in this bioavailability experiment, prepared as described in Uhler et al. (2021), was made up to a concentration of approximately 2% Aroclor 1254 (w/w, liquid), which is equivalent to 4% Aroclor 1254 on a dry paint basis. This 2% formulation was used in order to optimize the concentration of PCB in sediment with a maximum density (mass and number) of PC, thereby improving PC homogeneity among experimental sediment treatments. The formulated paint was applied to a steel panel, dynamically aged in seawater for one month following methods described in Kojima et al. (2016), air-dried, and removed using a razor blade (Uhler et al. 2021).
Preparation of paint chip size classes
Flakes of PCB-containing paint were ground using a mortar and pestle and dry sieved using stainless steel sieves to achieve 3 separate PC size classes (“coarse” = 2-5 mm; ASTM #10; “medium” = 0.250-0.300 mm; ASTM #50-60; “fine” = <0.045 mm; ASTM #325; Figure S1). Following grinding, the particles were stored in amber glass jars at 4 °C prior to addition to sediment for experimental purposes.
Due to the role of surface area in chemical reactivity (e.g., sorption and desorption), both specific surface area and surface-area-to-volume ratios were calculated for the three size classes of PC. For the “coarse” PC, the specific surface area (cm2/g) was estimated based on the geometry of a flat (plate-like) structure (Equation 1; Pennel 2016).
Where ρ is the density of PC (῀2.8 g/cm3 for dry 4% Aroclor 1254 PC), a is the length, b is the width, and V is the volume (i.e., length (a)*width (b)*height (c)) of the structure. Length (a) and width (b) estimates were based on sieve diameter sizes. The average thickness (c) of the dried PC was approximately 0.078 mm.
For the ground PC medium and fine size classes, the particle shape was assumed to approximate a sphere, where the specific surface area (cm2/g) was estimated using Equation 2 (Pennel 2016).
Where ρ is the density of PC (῀2.8 g/cm3), and r is the radius of the particle (based on sieve equivalent diameter sizes).
Estimates of surface-area-to-volume (S:V) ratios were 1,333, ῀220, and ῀250 for the fine, medium and coarse fractions, respectively (Table 1). It should be noted that these values are estimates as they do not account for surface texture, non-spherical particles, or finer particles that may be present (Pennel 2016).
Table 1. Specific surface area and S:V estimations for paint particles.
Size Class
|
Diameter (mm)
|
Estimated Specific Surface Areaa (cm2/g)
|
S:V
|
Coarse
|
2 - 5
|
92a
|
256c
|
Medium
|
0.25 - 0.3
|
71 - 86b
|
200 - 240d
|
Fine
|
0.045
|
476b
|
1,333d
|
abased on equation (1); bbased on equation (2); cS:V = 2(ab)/V; dS:V = [4πr2]/[(4/3)πr3] or 3/r
|
Sediments
Sediment with very low levels of PCBs was collected from a single location at Horseshoe Lake (HSL) an oxbow lake alongside the Mississippi River (Warren County, MS), with no known sources of PCBs. The HSL sediment was comprised of 34.2% solids, with a predominance of fine particles and OC content of 3.6%. The sediment was thoroughly homogenized with a propeller mixer (Lightnin® Vari-Mix portable mixer; Mixing Equipment) and stored at 4 °C in clean polyurethane buckets before use. The concentration of ∑PCBs (sum of 33 detected congeners) was 0.013 mg/kg dry wt.
Sediment contaminated with relatively high concentration of PCBs was collected from several locations in the Manistique Harbor (MH) Superfund site located in Manistique, Michigan on the southern shore of Michigan’s Upper Peninsula. The primary sources of contamination at this site include release of PCBs from former paper mill and lumber mill operations, discharge from area industrial facilities and nonpoint sources. Fish collected at the site have elevated levels of PCBs, indicating bioavailability and bioaccumulation of contaminants from the sediment (Gustavson 2014). The field-collected sediments were thoroughly homogenized with a propeller mixer and stored at 4 °C in clean polyurethane buckets before use. MH sediment was comprised of 35% solids, with a predominance of fine particles . The average concentration of ∑PCBs sum of 171 detected congeners) and OC content, based on sediments obtained from replicate jars after non-depletive ex situ passive sampling, were 5.84 ± 1.78 mg/kg and 6.7 ± 0.3%. The relatively low variability for PCB concentration and OC content across replicates indicate that sediment was adequately homogenized.
Ex situ passive sampling
Each PS consisted of a 2.5 cm X 6 cm (~25 mg) PE (17.2 μm thick, HDXTM brand) coupon. The PSs were cleaned by soaking at least three times in dichloromethane (DCM). Each soaking lasted 2 to 3 days. Clean DCM was used for each cleaning cycle. The PS were then rinsed with water multiple times, each for a period of days. Each time clean Milli-Q water was used. The PSs were stored in water (sealed in a jar) for approximately 8 weeks prior to use.
Sediment was portioned into approximately half-liter glass jars (16-oz) with Teflon lined caps. One PS was then added to each jar. PC were then added to create mixtures of sediment, PC and one PS. Each mixture was a fluid slurry and at least 25% of the volume of the jar was left as headspace. Hand shaking of the jars confirmed that the slurries would move inside the jars during mixing. The jars were rotated end-over-end at 30 revolutions per minute (rpm) most of the time. However, initially not all jars could fit on the end-over-end mixer, so some jars were placed on a slower mixer at 5 rpm (end-over-end). The jars were rotated across mixers, so that all jars spent at least two thirds of their mixing time at 30 rpm. Rotation was accomplished by removing jars from the mixers, hand shaking the jars, and placing them back on the mixers in different locations/configurations. The aim was to promote and maximize a well-mixed slurry environment in the jars for the longest time possible within the timeframe of the project. At the termination of the mixing period, PSs were retrieved from the jars using tweezers. The coupons were rinsed with deionized water and wiped with a lint-free paper tissue to remove sediment particles from the surface of the PS. The PSs were individually tightly wrapped in aluminum foil. Each wrapped PS was placed in a 40-mL amber glass vials with Teflon lids. A few drops of water were placed in the vials to maintain moisture (Apell and Gschwend 2016) to lessen loss of PCBs from the PS. The vials along with the sediment recovered from each jar were shipped on ice overnight to Alpha Analytical Laboratory (Mansfield, MA, USA) for PCB analysis.
Bioavailability of paint-associated PCBs in the presence of sediment:
Effect of paint chip particle size
To evaluate the effect of particle size on bioavailability, four ex situ passive sampling jars were set up for each PC size class. Each jar received 241.5 ± 0.2 g wet weight of the HSL sediment corresponding to 83 g of dry sediment, one PS and 10 ± 0.1 mg of PC targeting 4 mg/kg ∑PCBs as PC to create the class size treatments HSL+PCFINE , HSL+PCMEDIUM, and HSL+PCCOARSE. This concentration was selected to match the ∑PCBs concentration in the MH sediment according to historic data and was approximately 4 times higher than high-end concentrations reported for sites impacted by PCB-containing paint (Supplementary Information). For ∑PCBs in the bulk sediment, the concentration of PC PCBs exceeded the concentration of native PCBs the sediment by 308-fold. The contents of the jars (sediment, PC, and PS) were mixed for 60 d to allow for PCB redistribution between PC, sediment and PS. A time period of 60 d was previously reported as sufficient time for sediment-associated PCBs to attain thermodynamic equilibrium with 25-µm PS during ex situ passive sampling (Lohman et al. 2012; Apell and Gschwend 2016). To verify if equilibrium of PCBs was achieved between sediment and PC, and PS, additional jars were set up as described above but only for HSL+PCFINE targeting 4 mg/kg ∑PCBs as PC and mixed for 119 and 158 d, four jars per time point. Even though the concentrations of native PCBs in HSL sediment were low, they were above detection limit for 33 congeners. Therefore, HSL sediment without added PC were also evaluated for bioavailability using PS as described above.
Bioavailability evaluation using an historically contaminated sediment amended with paint chips
For this component of the investigation, two treatments for ex situ passive sampling with PE PS were set-up as described above: 1) MH sediment only, and 2) MH sediment and the fine PC fraction combined (MH+PCFINE). The primary goal was to conduct a bioavailability comparison for paint-associated PCBs and sediment-associated PCBs. Because some of the PCBs in MH were not present in PC, we also investigated whether PC would serve as a sorptive matrix for sediment-associated PCBs, reducing bioavailability of congeners not initially present in the PC.
For this experiment, each jar received 268 ± 0.2 g wet weight MH sediment corresponding to 92 g of dry sediment and one PS. The MH+PCFINE treatment was created by adding 10 ± 0.1 mg of the fine fraction of PC to MH sediment targeting 4 mg/kg ∑PCBs as PC (target total PCBs concentration = 9.8 mg/kg dry wt.). Four jars were set up for each treatment and the contents of the jars (sediment, PC, and PS) were mixed for 60 d. The average concentration of ∑PCBs (sum of 180 detected congeners) for the MH+PCFINE treatment, based on sediments obtained from replicate jars after ex situ passive sampling, was 8.25 ± 0.74 mg/kg.
Chemical analysis
Sediment extraction. Approximately 20 g of well homogenized sediment was weighed into a Teflon™ extraction jar, dried with sodium sulfate, fortified with surrogate compounds, and serially extracted three times with DCM using an end-over-end mixer. Extracts were filtered through glass wool containing sodium sulfate and concentrated on a hot water bath. Extracts were cleaned with activated copper to remove sulfur and processed through a silica gel column followed by high pressure liquid chromatography (HPLC) containing a size exclusion gel permeation column (GPC). Extracts were solvent exchanged into hexane and acid cleaned using sulfuric acid.
PS extraction. Each PS was placed into a 250 mL amber glass jar equipped with a Teflon™ liner, dried with sodium sulfate, fortified with surrogate compounds, and serially extracted three times with DCM using a shaker table. Extracts were filtered through glass wool containing sodium sulfate and concentrated on a hot water bath. Extracts were cleaned with activated copper to remove sulfur, solvent exchanged into hexane and acid cleaned using sulfuric acid.
Instrumental analysis. All extracts were fortified with an internal standard and analyzed using an Agilent HP6890 or equivalent equipped with a Restek RTX-PCB 60-m x 0.18 mm ID, 0.18 um film thickness, fused-silica capillary column and a mass spectrometer operated in the selected ion monitoring mode (SIM). The concentrations of the individual congeners were quantified versus internal (i.e., injection standards) standards, which were spiked into the sample extract prior to analysis. The target congener concentrations were quantified using average response factors generated from a minimum of a 6-point multi-level calibration curve. Sample extracts were analyzed for 209 PCB congeners using USEPA Method 680.
Total Organic Carbon (TOC). Approximately 10 mg of sample (pre-treated with 10% hydrochloric acid, dried, and homogenized) was weighed into a tin capsule, and analyzed using a CHNS/O Analyzer for TOC per USEPA 9060. All analysis was performed in duplicate and the average TOC value reported.
Quality Control
A series of quality control samples were included to monitor laboratory contamination, extraction efficiency, and reproducibility. This was accomplished through the use of procedural blanks, lab control samples/duplicates (LCS/LCSD), surrogates, lab duplicates, and National Institute of Standard Reference Material (NIST SRM).
No analyte was detected in the procedural blanks above the reporting limit. All laboratory spiked surrogates and LCS compounds met data quality objective (DQO) recoveries (50-125% and 40-140% respectively). The lab duplicates had a relative percent difference (RPD) <30% for over 90% of the analytes detected above the reporting limit and the NIST SRMs met the lab DQO recoveries (40-140%) for all certified analytes detected above the reporting limit (see supplemental information).Detailed information on quality control and quality assurance is provided in the Supplementary Information.
Data analysis
Statistical comparisons were performed using SigmaStat v3.5 software (SSPS, Chicago, IL, USA). Normality was confirmed by the Shapiro-Wilk test and equal variance was confirmed using the Brown-Forsythe test. One-way ANOVA were performed to determine statistically significant differences (α = 0.05) across three or more treatments. The Holm-Sidak method was employed for pairwise multiple comparisons to determine statistical significance between treatments. When assumptions of parametric ANOVA were not met, the data was log-transformed. The nonparametric Kruskal–Wallis one-way ANOVA on ranks was applied when assumptions of parametric ANOVA were not met for log-transformed data. The Dunn’s method was employed for pairwise multiple comparisons to determine statistical significance between treatments. The Student’s t test was used to determine whether statistically significant differences existed between treatment groups (α = 0.05). When assumptions of parametric t test were not met, the nonparametric Mann–Whitney Rank Sum test was applied.