Estimation of In Situ Sediment-to-Water Fluxes of Polycyclic Aromatic

Mar 24, 2010 - The Netherlands, IMARES, Haringkade 1, 1976 CP IJmuiden,. The Netherlands, Alterra, P.O. Box 47, 6700AA, Wageningen,. The Netherlands ...
0 downloads 0 Views 248KB Size
Download PDF
Environ. Sci. Technol. 2010, 44, 3014–3020

Estimation of In Situ Sediment-to-Water Fluxes of Polycyclic Aromatic Hydrocarbons, Polychlorobiphenyls and Polybrominated Diphenylethers A L B E R T A . K O E L M A N S , * ,†,‡ ANTON POOT,† HENDRIKA J. DE LANGE,§ I L O N A V E L Z E B O E R , ‡ J O O P H A R M S E N , †,§ A N D P A U L C . M . V A N N O O R T †,| Aquatic Ecology and Water Quality Management Group, Wageningen University, P.O. Box 47, 6700 AA, Wageningen, The Netherlands, IMARES, Haringkade 1, 1976 CP IJmuiden, The Netherlands, Alterra, P.O. Box 47, 6700AA, Wageningen, The Netherlands, and Deltares, PO Box 85467, 3508 AL Utrecht, The Netherlands

Received December 28, 2009. Revised manuscript received March 10, 2010. Accepted March 11, 2010.

Sediment--water fluxes of hydrophobic organic chemicals (HOC) may affect the quality of surface waters. Here, we present an approach to derive such fluxes from (a) in situ HOC concentration gradients measured with passive samplers and (b) mass transfer coefficients measured with a novel flux method using Empore disks. For eight undisturbed sediments, this method identified whether the sediment acted as a source or as a sink for HOCs. The analysis also identified which type of transport resistance governed sediment water exchange. For seven inland locations, exchange was limited by benthic boundary layer transport, showing no dependencies on sediment or chemical properties other than concentration. For one river mouth location, exchange was limited by slow in-bed intraparticle diffusion. A biphasic dual compartment radial diffusion model adequately described the data for this location. Fast desorption was interpreted as molecular diffusion retarded by microscale dual domain sorption to amorphous as well as black carbon (BC). Slow desorption was invariant with LogKow and consistent with intraorganic matter diffusion through BC particles. Finally, it is discussed how these findings can be translated into a general framework for flux based exposure assessment.

Introduction The importance of contaminant release from sediment beds for the quality of surface waters has been recognized decades ago (1, 2), but received less attention compared to release from suspended sediments. Contaminant release from sediment beds can be quantified as a flux φsed, that is, as contaminant mass released from the sediment water interface per unit area and per unit of time. Being able to measure * Corresponding author phone: +31 317 483201; e-mail: [email protected]. † Wageningen University. ‡ IMARES. § Alterra. | Deltares. 3014

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 44, NO. 8, 2010

fluxes under field conditions is relevant for several reasons. First, addressing fluxes from bed sediments may better emphasize the transport aspect of bioavailability, which to date is largely overlooked. For nonequilibrium situations, Valsaraj et al. (3) convincingly argued that water column concentrations leading to exposure to aquatic organisms are proportional to the flux from the sediment, not the concentration in the sediment. For instance, when the in-bed transport resistance is molecular diffusion, steady state hydropobic organic compound (HOC) pore-water concentrations can be up to 2 orders of magnitude lower than equilibrium partitioning theory (EPT) predicts (3). For equilibrium situations, similarly low pore-water concentrations have been related to the presence of condensed carbon phases in sediment, hereafter referred to as “black carbon” (BC) (4). Measurement of in situ fluxes might be useful to distinguish between these two mechanisms. Second, HOC release from sediment beds is a key element in transport models used for evaluation of water quality management scenarios (2, 5). Evaluation of emission reduction scenarios requires comparison of HOC release from sediments compared to other loads such as atmospheric deposition or upstream riverine loadings (2, 6). Third, flux measurements could contribute to site characterization in the framework of remedial actions for contaminated sediments (7), or management actions following from the EU Water Framework Directive (WFD). One proven threat to reaching the chemical and ecological WFD targets is exposure to sedimentbound contaminants (8). Consequently, there is an increasing need for routine quantification of in situ sediment-to-water fluxes and their possible implications for contaminant fate and ecological quality. The flux of a HOC from the sediment pore-water to the overlying water φsed (µg × m-2 × d-1) can be described as a product of the concentration gradient between pore- and overlying water, and an aqueous mass transfer coefficient (eq 1) (3, 5, 6, 9): φsed ) KL(Cp - Cow)

(1)

with KL (m × d-1) the sediment-water mass transfer coefficient, and Cp and Cow (µg × m-3) the concentrations in pore and overlying water. The reciprocal of KL can be interpreted as an overall transport resistance, which is composed of transport resistances relating to in-bed molecular diffusion, benthic boundary layer transport, or bioturbation and -diffusion (3, 6). Only few laboratory studies provide values for KL (5, 9, 10), and also, values for field conditions are rare (5). HOC concentration or fugacity gradients between pore- and overlying water have been measured only recently, using novel passive sampler methodologies (7). As far as we know, no earlier studies combined such passive sampler based gradients with field based values for KL in order to estimate in situ fluxes. The primary aim of this study was to experimentally determine in situ sediment-to-water fluxes; direction as well as magnitude. The method combined (a) in situ concentration gradients measured with passive equilibrium samplers, with (b) laboratory KL values from flux measurements at nearresuspension conditions using fresh, unaltered sediment cores and Empore disks as sink for HOCs in the overlying water. Overlying water HOC concentrations (Cow) were determined on field locations with soft silicone rubber (SR) passive samplers that allow relatively fast equilibration. Porewater native HOC concentrations (Cp) were determined by negligible depletion equilibrium partitioning to hard poly10.1021/es903938z

 2010 American Chemical Society

Published on Web 03/24/2010

oxymethylene (POM) samplers deployed in agitated sediment slurries in the laboratory, which provide a better resistance to BC nanoparticles (11). Chemicals studied included native polycyclic aromatic hydrocarbons (PAH), polychlorobiphenyls (PCB) and polybromodiphenylethers (PBDE). Sediments were chosen to cover different HOC concentrations in The Netherlands, in order to assess the effects of concentration on fluxes. A second aim was to discuss the role of BC in explaining observed pore-water concentrations and sediment-water exchange kinetics.

in this study, transport may be expected to be water side controlled and KL can be assumed constant in time (2, 6). Although bioturbation was not observed, this process may have contributed to HOC transport in the sediment, thereby reducing the bed side resistance (14). If an Empore disk is present in the overlying water, an additional flux φED (µg × m-2 × d-1) from the overlying water to the disk is defined by

Materials and Methods

where KLED (m × d-1) is the mass transfer coefficient for HOC transfer from the overlying water to the Empore disk, CED is the concentration in the Empore disk (µg/kg) en KdED is the HOC disk to water equilibrium partition coefficient (m3/kg). Assuming the stirred overlying water to be uniformly mixed, the effect of the serial fluxes on the HOC concentration change in the overlying water can be calculated by combining eqs 1 and 2 and correcting for sediment surface area (Ased; m2), Empore disk surface area (AED; m2), and overlying water volume (Vow; m3):

Chemicals. Details on chemicals, materials, and pretreatment of materials are provided as Supporting Information (SI). Locations and Sampling. Six sediments were taken from IJmuiden Harbor (IJH), one from Canal Gent-Terneuzen (CGT) and one from a flood plain lake near the Rhine (ADW), using an Ekman sediment grab (SI Figure S1). IJH samples numbered 1-6 followed a concentration gradient with IJH1 having highest concentrations and IJH6, which is closest to the sea, having lowest concentrations. Sediments for laboratory flux experiments and determination of pore-water concentrations were taken using Jenkins core samplers. Sediment Analyses. Total organic carbon (TOC) and BC were determined in triplicate using chemothermal oxidation at 375 °C (12). Amorphous organic carbon (AOC) was calculated as AOC ) TOC-BC. PCB and PAH were cold extracted with acetone/petroleumether. PCBs were analyzed using GC-MS. PAH were analyzed using HPLC with fluorescence detection. PBDE (CGT samples only) were Soxhlet extracted with hexane/acetone and analyzed using GC-MS. Details on these analyses are provided as SI. In Situ Concentration Gradient. PAH and PCB concentrations in sediment pore water were determined in triplicate using POM passive samplers, following previously published procedures (11). In situ aqueous PCB, PAH, and PBDE concentrations in the overlying water column at locations ADW, CGT, IJH1, and IJH6 were measured using polydimethylsiloxane (PDMS) SR passive samplers (13). Further details are provided as SI. Laboratory Flux Experiments. Immediately after sampling, unaltered sediment cores with overlying water were brought to the laboratory. Cylindrical cores measured 70 cm (h) by 6 cm (diameter) and contained 30 cm sediment and 30 cm of overlying water. The overlying water was gently homogenized by Empore disk-stirrers connected to one motor by a chain, such that all eight sediments had the same stirring speed. Empore disk-stirrers consisted of an Empore disk held by an open frame of stainless steel, positioned at a distance of 10 cm from the sediment surface. Stirring was adjusted to the highest possible rate before sediment resupension occurred, as tested on a separate set of cores. Empore disks were replaced by clean disks after 4, 24, 72, 168, 336, and 504 h, and were Soxhlet extracted and analyzed as described for POM. Calculation of lab-KL Values from Laboratory Flux Experiments. The flux of HOCs from the sediment porewater to the overlying water in the cores φsed is described as in eq 1. In a first stage of release from homogeneous sediment, the mass transfer coefficient KL is constant in time, as no or limited gradients exist in the sediment and transport is controlled by the water side resistance in the benthic boundary layer. In this stage, KL can be equated to Dw/δL where Dw is a molecular diffusion coefficient and δL the thickness of the boundary layer (3, 6). After a critical time, tcrit, this regime is taken over by another regime in which in-bed diffusion becomes rate-limiting and the apparent KL decreases with square root of time (see also eq 9) (2, 6). For HOCs with Koc> 104-105 L/kg, tcrit has been calculated to be >1 year (6), which implies that at shorter time scales such as

φED ) KLED(Cow - CED /KdED)

dCow /dt )

(2)

KLAsed KLEDAED (Cp - Cow) + (CED /KdED - Cow) Vow Vow (3)

In our experiments, Empore disks were replaced by clean disks thus keeping CED low. Accordingly, backward transport from the disk to the overlying water was limited, which reduces mass balance eq 3 to a first order two-compartment model dCow /dt ≈ ksed(Cp - Cow) - kEDCow ) ksedCp (ksed + kED)Cow

(4)

in which ksed (d-1) equates to KLAsed/Vow, and kED ) KLEDAED/ Vow (d-1) quantifies effective uptake by the disks. Some earlier studies strived at the condition kED > 10 ksed to ensure sediment-water transport being rate limiting, allowing the further assumption of Cow ≈ 0. However, values for KLED are highly dependent on experimental conditions such as turbulence, disk housing, disk pretreatment or temperature and therefore carry considerable uncertainty (15). We therefore did not use preset values for kED but fitted ksed as well as kED for all individual sediment-chemical combinations. To this end, eq 4 was complemented by mass balance equations for the sediment-pore-water and the Empore disk compartments. For the pore-water, a mass balance similar to eq 3 is defined: dCp /dt )

KLAsed Vow k (C - Cp) (Cow - Cp) ) Vp Vp sed ow

(5)

If in a first stage of release (before tcrit) no substantial in-bed gradients exists, the pore-water concentration Cp can be approximated as Cp ≈ Csed/Kd where Csed (µg/kg) is the HOC concentration in the sediment and Kd (L/kg) is the sediment water equilibrium distribution coefficient (6). This strongly reduces depletion of the pore water because desorbing HOCs contribute to the flux. This further implies that the apparent pore-water volume (Vp) stands for the volume of water required to contain the initial sediment-bound HOC mass at initial concentration Cp, which in turn is equal to measured total HOC mass in the system (QT) minus HOC mass in the overlying water, that is, Vp × Cp,t)0 ) QT - Vow × Cow,t)0. By assuming identical initial concentrations in pore and overlying water (Cp,t)0 ) Cow,t)0 ) C0), Vp in eq 5 can be calculated as Vp ) [QT - (Vow × C0)]/C0. Finally, the cumulative quantity measured in the Empore disks (QED) can be calculated from VOL. 44, NO. 8, 2010 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

3015

the measured total quantity of HOC in the system (QT), Vp, Vow and the coupled eqs 4 and 5: QED ) QT - VpCp - VowCow

(6)

Equations 4-6 were numerically solved in Microsoft Excel using an Euler integrator and variable time step. KL, KLED and C0 were optimized using the Excel Solver tool with scaling of parameters and a relative least-squares criterion. The procedure was validated with the Software package Scientist (Micromath Inc., Salt Lake City, Utah) for eight representative cases, which all showed excellent agreement (SI Table S1). Following parameter estimation, 90% confidence intervals (CI90) for KL were obtained by fitting this parameter to meet the criterion (16): SS90 ) SSmin[1 + p/(n - p) × F(p, n - p, 90%)]

(7)

with SS90 is the residual sum of squares at the 90% confidence contour, SSmin is the sum of squares at the best estimate of KL, n is the number of data points in the cumulative plot of desorbed test compound vs time, p is the number of parameters, and F(p,n - p,90%) is the F distribution.

Results and Discussion Equilibrium Phase Distribution of PCBs, PBDEs, and PAHs. Sediment-to-water distribution ratios (KD, L/kg) were calculated from total concentrations (SI Table S2) and averaged triplicate pore-water concentrations (SI Table S3) and showed a clear relationship with LogKow (SI Figure S2). These KD values relate to redistribution of a small fraction of native HOCs from the fast desorbing HOC pool to water and POM passive samplers, in agitated slurries for a month (11). This implies that experimental conditions approximated the conditions as they would be encountered at microscale equilibrium in the field. Accordingly, the KD values can be referred to as in situ equilibrium coefficients. Nevertheless, KD values for different sediments show an order of magnitude variation and are 1-2 orders of magnitude higher than EPT sorption would predict (SI Figure S2). The high magnitude and variation in KD values can be explained by sorptive phases such as BC (4, 11). The KD values can be modeled using a dual domain sorption model that combines sorption to AOC according to the traditional EPT model with nonlinear sorption to BC (4, 17):

KD )

Cs BC ) focKoc + fBCKD ) focKoc + fBCKF,BCCpnF,BC-1 Cp (8)

in which Cs (µg/kg) is the measured HOC concentration in sediment, foc, and fBC are the AOC and BC weight fractions, Koc (L/kg) and KF,BC ([µg/kg]/[µg/L]nF,BC) are sorption constants for these respective phases, which are calculated from compound-class specific regressions with LogKow, and nF,BC is the Freundlich exponent for nonlinear sorption to BC (set to 0.7 4, 18, 19). As in earlier studies (4, 17-19), eq 8 yields good fits to the data (SI Figures S3, S4), with KF,BC parameters close to previous calibrations. In situ distribution coefficients for BC, KDBC (L/kg; eq 8), were 1 order of magnitude higher than whole-sediment KTOC values, and 2-3.5 orders of magnitude higher than EPT-based KOC values (SI Figure S2). Although the data are consistent with previous equilibrium studies, for the assessment whether equilibrium also exists in the field situation, in-bed diffusive rate limitations should be ruled out (3), as will be discussed below. In Situ Concentration Gradients. In situ concentration gradients were calculated as Cp-Cow. Given the measured uncertainty in Cp of (8-14% and reported general uncertainty in Cow of (50% (13), we used a criterion of Cp > 2 × 3016

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 44, NO. 8, 2010

Cow as a limit of detection for a positive gradient (leading to a flux from the sediment pore-water to the overlying water), and vice versa for a gradient of opposite sign. Using this criterion, positive PCB gradients were detected only for PCB28 at CGT and ADW (Table S4). For PCBs 101, 118, 138, 153, and 180, several negative gradients were found, which in all cases followed the order ADW > CGT > IJH6 > IJH1. PAH gradients were larger and positive at IJH1. For 3 and 4 ring PAHs (PHE, ANT, FLU, PYR) gradients are highest, also at location CGT, whereas at ADW only PHE passed the detection criterion. PBDEs were only measured in samples from CGT, where one positive and one negative gradient were detected for PBDE153 and PBDE47, respectively. Laboratory KL Values. HOC masses detected in Empore disks were used to construct plots of cumulative HOC mass removed per sediment. It appeared that concentrations for all PCBs and PBDEs were below detection limits, except those for BDE209 in the first two extraction steps. For PAH, the total masses per chemical captured by Empore disks, were first compared to the masses present in the overlying and pore water, as calculated from water volume and POM-SPE based measurement of Cp (SI Table S3). It appears that in most cases, aqueous PAH could account for less than 10-20% of extracted PAH (SI Figure S5), implying that more than 80-90% of chemical mass originated from the sediment particles. Further, the quantity of extracted PAHs (except NAPH) was much less than 1% of total mass initially present in the sediment. This implies that only readily available sediment-bound PAHs were removed, because fast desorbing HOC percentages are much higher (6). Removal percentages for NAPH were a little higher with 2-8% removed for IJH samples, 12% for CGT and 18% for ADW. Cumulative PAH removal followed the order NAPH > ACE > FLU > FLE > PHEN > PYR > ANT > CHR > BAA > BAP > BBF > BKF > INP > BGP > DBA, that is, heavier and more hydrophobic PAHs were released by the sediment to a lesser extent (SI Figures S6-S14). The experimental data were fitted to the model as condensed in eq 4-6, and showed good agreement. Due to the limited depletion of the sediment, the model results appeared insensitive to QT. This implies that any uncertainty with respect to the availability of PAH for diffusive release according to eqs 4-6 did not affect the modeling results. Model fitting resulted in a rather narrow range of KLED values (0.18 m × d-1, inter quartile range, IQR, ) 0.10 to 0.24 m × d-1) (SI Table S5). Similar KLED values can be expected because aqueous diffusivities, turbulence and other experimental conditions were identical for all laboratory flux experiments. This also confirms that backward transport from the disks played no role, because this would have lowered effective mass transfer especially for the least hydrophobic PAH. The median ratio of fitted KLED/KL was about 10 and was between 1 and 5 only in 15% of all cases. This confirms that mass transfer from the water to the Empore disk was faster than transport across the sediment-water interface, thus forcing a continuous concentration gradient between pore and overlying water. For locations IJH1, IJH2, IJH3, IJH4, IJH5, ADW, and CGT, KL values typically range between 0.01 and 0.1 m × d-1 (average 0.024 m × d-1, IQR ) 0.015 to 0.026 m × d-1), and are independent of LogKow (Figure 1, SI Table S6). However, for IJH6, lower apparent KL values and a clear negative trend with LogKow are found (Figure 1). CI90s for fitted KL values typically are (10-20%. For location IJH1, optimization of KL was not possible for ACE, FLE, PHEN, ANT, FLU, and PYR. For these cases, the aforementioned percentage of mass originally present in the overlying water probably was too high (SI Figure S5). This implies that the release curves were insufficiently determined by the sediment-water exchange process to accurately optimize KL. For BDE209 only data for 4 and 24 h release time were available. Tuning of KL yielded

FIGURE 1. PAH sediment water mass transfer coefficients as a function of LogKow for eight sediments. an exact fit at 0.03 m × d-1. This value has little statistical rigor but still agrees well with the values for PAH, suggesting that BDE209 did not behave differently. From the KL values for locations IJH1-5, ADW and CGT, several inferences on the nature of the exchange process can be made. They do not apply to IJH6, which will be discussed separately. First, it appears that each release curve can be fitted using eqs 4-6 and one single KL value, that is, there is no indication of a shift of boundary layer resistance turning into diffusion limitations throughout the time span of this experiment. This is also consistent with the fact that KL values show no trend with hydrophobicity of the PAH, even though LogKow values span more than 4 orders of magnitude. This is consistent with aqueous boundary layer transport at these locations because molecular diffusivities have limited range (6). However, this is not consistent with intra- or interparticle diffusion because hydrophobicity should have a large impact on the effective diffusivity governing these in-bed processes. Third, there is no clear relationship between KL values and other characteristics of these seven sediments (Figure 1, SI Table S6). KL values are close, often with overlapping confidence intervals. However, SI Figure S2 shows that variation in equilibrium partitioning coefficients is more than 1 order of magnitude. Also particle sizes of for instance IJH1 differed considerable from that of the other sediments (SI Figure S15). If sediment side diffusion limitations would be rate limiting, these different adsorption characteristics and particle size distributions would have caused distinct differences in effective diffusivities, thus yielding different release profiles among sediments. We explain the small differences in KL among the various sediments from differences in shear stress at the sediment-water interface resulting in differences in benthic boundary layer thicknesses. Finally, our KL values of 0.024 (IQR ) 0.015 - 0.026 m × d-1) agree well to earlier reported values, especially for systems with limited bioturbation. Eek et al (20) reported on flux experiments from which KL values of 0.02-0.027 m × d-1 can be calculated for a series of PAH. For PCB release from Hudson River sediments, Connolly et al reported KL values of 0.03 m × d-1 under winter and 0.14 m/d under summer conditions (21), whereas Erickson et al reported a range of 0.03-0.19 m × d-1 (14) for the same river. Granberg et al (9) reported a range of 0.07-0.53 m × d-1 for DDE, DDD, and a series of PCBs calculated from laboratory experiments, and a range of 0.12-1.06 m × d-1 for the same chemicals in bioturbated systems. Valsaraj et al (3) estimated a KL value for TCDD of 0.1 m × d-1. These ranges agree well to our data,

but are too high to be attributed to in-bed diffusion limitations, which typically yield transfer rates a factor of 1000 lower (3). Consequently, we conclude that our shortterm flux experiments with sediment cores from locations IJH1-5, ADW, and CGT were rate limited by benthic boundary layer transport. For location IJH6, however, apparent KL values decreased with LogKow according to LogKL (m × d-1) ) -0.5LogKow -0.16 (r2 ) 0.74) (Figure 1). Because diffusivity differences in the benthic boundary layer cannot explain this strong dependency on PAH hydrophobicity and because of the low apparent KL values, we suggest release from IJH6 must have been sediment-side controlled by diffusive limitations in the sediment bed. Such limitations may originate from retarded diffusive transport in interstitial pore space or from diffusive transport limitations in individual sediment particles or aggregates, either caused by intraparticle retarded diffusion or intraorganic matter diffusion (6, 22, 23). If retarded diffusion in interstitial pores is rate limiting, the exchange can be modeled using the geometry of a semi-infinite slab, and the mass transfer coefficient in eqs 1, 3, and 5 can be equated to (2, 6): δL 1 ) + KL Dw



π×t Dwε4/3Rf

(9)

where Rf ) ε + σKD, ε is sediment porosity (v:v) and σ is sediment density (kg/L). Equation 9 expresses the resistance to mass transfer as the sum of diffusive resistances in the benthic boundary layer (δL/Dw) and the sediment bed. Note that the retardation factor Rf is dependent on aqueous concentration as well as sediment BC content through eq 8. We evaluated this dual resistance model using the data for IJH6, by setting the aqueous boundary layer resistance Dw/ δL to the average measured value of 1/KL ) (0.024)-1 d × m-1, ε, σ and Kd to the measured values (SI Figure S2, Table S2), and optimizing Dw. This resulted in statistically satisfying fits to the data, with similar or slightly worse sums of squares compared to the single resistance model. The factor decrease of overall KL values according to eq 9, ranged from a factor four (FLU and PYR) to a factor 325 (DBA) during the 500 h time span of the experiments. Resultant value for Dw was realistic for NAPH (1.6 × 10-5 m2 × d-1) (6) but was too low for all other PAH, extending to very unrealistic values of 10-10-10-11 m2 × d-1 for the most hydrophobic PAH (SI Table S7). We therefore conclude that the assumptions underlying this approach do not apply to IJH6 either. So, because benthic boundary transport as well as diffusion in interstitial pores failed to explain observed PAH release from IJH6 sediment, we hypothesize that bed release was limited by intraparticle diffusion. Several recent reports provide evidence for biphasic desorption from amorphous organic matter and condensed organic matter domains, which can be described by radial diffusion kinetics from two types of particles according to the following (22, 24): QED /QT ) 1-

6 π2

∑ n1 (F ∞

2

rapexp

n)1

rap ∧ Deff )

Dw

[

rap -n2π2Deff t 2 rrap

]

+ (1 - Frap)exp

[

slow -n2π2Deff t 2 rslow

])

1 + KDrsw (10) rap Deff

and where Frap is the rapidly desorbing PAH fraction, slow are effective diffusion coefficients (m2/d) for rapidly Deff and slowly diffusing fractions, rrap and rslow are the radii of the corresponding equivalent spheres, and rsw is the solid to solution ratio of the sediment, equal to rsw ) σ(1 - ε)/ε. rap 2 slow 2 /rrap and Deff /rslow with Equation 10 was solved for Frap, Deff Simplex global optimization according to Nelder and Mead, VOL. 44, NO. 8, 2010 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

3017

TABLE 1. In Situ Flux Estimates Based on in Situ Concentration Gradients and Mass Transfer Coefficients Derived from Laboratory Flux Experiments flux estimates (ng × m-2 × d-1) location

rap 2 FIGURE 2. PAH LogD eff /rrap as modeled with a biphasic radial rap 2 /rrapestimated diffusion model (eq 10 first line, O), and D eff independently from measured KD, rrap, rsw, and Dw (eq 10 last line, ×), as a function of LogKow.

followed by local search with the Solver tool in Excel, accounting for 2000 terms in the infinite series (24) (SI Table S7). This diffusion model fitted the curves better than the two compartment first order model as condensed in eqs 4-6 (compare SI Figures S11 and S12). Rapidly desorbing fractions were small and were approximately equal to Empore disk extracted quantities for 2, 3, and most 4 ringed PAH (SI Table S7). For CHRYS, BAA, 5 and 6 ring PAH, Frap amounted to rap 2 /rrap 5-30% of totally extracted quantities. Values for LogDeff decreased linearly with increasing LogKow, which is consistent with retarded molecular diffusion through intraparticle pores 2 (Figure 2). Theoretically, the values forDrap eff /rrap estimated from the flux experiments can be compared with values independently calculated from measured or estimated rrap, Dm, KD, and rsw (eq 10) (6, 22). Such comparisons may be troubled by the uncertainties in KD and because diffusive path length r is indeterminate (24). Nevertheless, by taking Dm from (6), rsw from measured values, our measured BC-inclusive KD values (SI Figure S2) and a realistic value for rrap (110 µm) rap 2 /rrap was obtained (Figure good agreement with modeledDeff 2). For comparison, with a traditional KOC value according to LogKoc ) 0.97 × LogKow - 0.12 (18) (neglecting sorption to BC), a similarly good agreement could be obtained only by assuming an unrealistically high effective particle diameter of 2200 µm. From these observations we conclude that fast desorption was governed by molecular diffusion in intraparticle pores, retarded by dual domain sorption to amorphous as well as condensed organic matter. slow 2 rap 2 Values for D eff /rslow were much lower than D eff /rrap and showed no trend with LogKow (SI Table S7). Consequently, diffusion from the slow domain can be parametrized with slow 2 one average value of LogD eff /rslow(d-1) ) -11.1 ( 0.27. This behavior points to diffusion in condensed organic matter such as BC, as argued in several earlier studies (4, 23, 25). Consequently, apparent diffusivities for this material can be estimated by assuming a plausible equivalent particle size. PAH in IJmuiden harbor most probably originate from unburned coal particles because coal is unshipped in the harbor where the sediment was sampled. A typical size range for coal particles is 10-20 µm (26), which yields diffusivities in the range of (1-4) × 10-21 m2 × d-1. This range is comparable to ranges reported earlier for diffusivities in condensed carbon for PAH (10-16-10-18 m2 × d-1 (23); 10-14 m2 × d-1 (25)), and PCB (10-19-10-20 m2 × d-1 (27)). The fact that the main kinetic fraction of PAH appears to come from this pool (very low Frap) agrees to our finding that BC dominated sorption at equilibrium conditions, as discussed above. In summary, microscale binding to BC seems to determine slow release kinetics by substantially attenuating the effective diffusion coefficient governing the rate of 3018

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 44, NO. 8, 2010

PCB PCB28 PCB52 PCB101 PCB118 PCB138 PCB153 PCB180 PAH PHE ANT FLU PYR CHR BAA BBF/BKF BAP INP BGP DBA PBDE PBDE47 BDE153 BDE183

IJH1

IJH6

CGT

ADW

a a a a a -0,30 -0,047

a a -0,64 -0,30 -0,23 -0,55 -0,074

1,4 a -1,0 a a -0,75 -0,21

1,3 a -2,3 -0,72 -1,3 -2,4 -0,73

8400 2700 18000 13000 180 370 93 28 1,4 3,1 0,22

a a a a -61 -36 a -7,1 -0,96 -1,4 -0,18

490 52 840 970 -88 a a a a a a

320 a a a a a a a -0,77 a -0,24

b b b

b b b

-0,044 0,16 a

b b b

a Not significant. For criteria, see text. measured.

b

Cp or Cow not

intraparticle pore diffusion, as well as intra organic matter diffusion in BC particles. The question remains why the core taken at IJH6 behaved so differently compared to the other cores. We see no explanation in sediment properties such as particle size or BC content (SI Table S2). A possible explanation is that IJH6 was located close to open sea, that is, it was subjected to tidal currents and clean seawater, which may have reduced the benthic boundary layer thickness and effectively removed fast desorbing fractions such that particle scale desorption became rate limiting. Bioturbator activity also may have been less at this location. In Situ Fluxes. Following eq 1, in situ fluxes were estimated by multiplying the concentration gradients with the laboratory based KL values (Table 1). The laboratory PAH and BDE209 KL values did not differ significantly among chemicals and locations IJH1-5, ADW, and CGT and also agreed to literature values for PCB and PAH. Accordingly, we used the average value of 0.024 m × d-1 as input for flux calculations for all HOCs studied. This value could be taken as a general value at near-resuspension conditions, a condition which is often encountered in shallow aquatic systems. For PAH at IJH6, very low KL values were found for transport over the positive concentration gradient forced in the laboratory. However, in the field, the flux was directed toward the sediment so that the value of 0.024 m × d-1 was also used for this location. Considering all detected fluxes (Table 1), especially the PAH gradients at location IJH1 are important, confirming this locations’ status of “hot spot” with respect to sediment quality. The total ΣPAH flux is 42 × 103 ng × m2 × d-1 ()16 mg × m-2 × yr-1), compared to 2.3 × 103 ng × m2 × d-1 at CGT. Fluxes for IJH6 and ADW are low, absent or negative, indicating that on the system level, these sediments do not act as a source of PAH. The same can be concluded for PCBs and PBDEs for all locations.

Implications for Risk Assessment. Valsaraj et al (3) provided a flux based concept that included bed side resistances in the assessment of benthic exposure concentrations or associated sediment quality criteria:

(

1

K*D ) KD 1 +

air

KL

∑ 1/K y

y

)

(11)

-1 where ∑air y 1/Ky (d × m ) is the sum of all transport resistances between water column and atmosphere. Assuming constant flux across the sediment-water and water-air interface, eq 11 calculates the apparent KD* experienced by benthic organisms as a function of Σ1/Ky, the equilibrium KD, and the mass transfer coefficient KL. Note that KD is the BCinclusive equilibrium value, as in eq 8, which leads to low exposure if binding to BC is substantial. As discussed before, KL may also be affected by BC through the effective diffusion coefficient (eq 10). Therefore, eq 11 unifies bioavailability limitations due to strong equilibrium binding to BC, as well as bioavailability limitations caused by BC related in-bed transport resistances in one framework. Using eq 11, we calculated exposure concentrations relative to EPT based concentrations (Cw*/CwEPT ) KD/KD*), as a function of KL for PHEN, BAA, and DBA. Calculations and results are detailed in SI Figure S16. For our current data, we conclude that no substantial deviations from EPT based exposure concentrations are expected, because (a) concentration gradients are absent or small (ADW and IJH6), or (b) KL values are too high to cause a substantial transport limitation (IJH1 and CGT). After all, at IJH1 and CGT, LogKL ranges from -0.5 to -2 (KL in m × d-1), yielding Cw*/Cw of 0.2-1.0 (SI Figure S16). However, at locations such as IJH6 where LogKL is between -3 and -4, transport limitations might occur as soon as water column concentrations decrease, for instance by seasonal or gradual changes in hydraulics or external loadings. Exposure concentrations would decrease up to a factor 10-100 below EPT predictions, especially for less hydrophobic PAH such as PHEN (SI Figure S16). In summary, this study is relevant for risk assessment for several reasons. First, it shows that by combining passive sampler data on in situ gradients with mass transfer coefficients, in situ fluxes can be estimated. Furthermore, this work introduced a rather simple methodology to measure mass transfer coefficients using Empore disks. Future development of this flux methodology should focus on better agreement between turbulence conditions in laboratory and field, for instance using laser Doppler anemometry. Finally, this work showed that this method efficiently identifies potential transport limitations as they occur in undisturbed natural sediments, which is of great interest in sediment site characterization and sediment quality assessment.

Acknowledgments This research was financially supported by the Ministry of Transport, Public Works and Water Management, The Netherlands (ATB 10098850). A.K. and A.P. acknowledge funding from European Union FP6 Integrated Project AquaTerra (Project no. GOCE 505428). We thank Hannie Maas, Ewoud Klopstra, John Hin, and Cor Schipper for initiating and managing the project. Foppe Smedes, Frits Gillissen, Hans Zweers, Laura Buijse-Bogdan, Maadjieda Tjon Atsoi, Christiaan Kwadijk, and Judith van Hesselingen are acknowledged for their contributions to the experimental work.

Supporting Information Available Underlying data, experimental details, sediment characteristics, detailed modeling results, and additional results. This material is available free of charge via the Internet at http:// pubs.acs.org.

Literature Cited (1) Larsson, P. Contaminated sediments of lakes and oceans act as sources of chlorinated hydrocarbons for release to water and atmosphere. Nature 1985, 317, 347–349. (2) Thoma, G. J.; Reible, D. D.; Valsaraj, K. T.; Thibodeaux, L. J. Efficiency of capping sediments in situ. 2. Mathematics of diffusion-adsorption in the capping layer. Environ. Sci. Technol. 1993, 27, 2412–2419. (3) Valsaraj, K. T.; Thibodeaux, L. J.; Reible, D. D. A quasi-steadystate pollutant flux methodology for determining sediment quality criteria. Environ. Toxicol. Chem. 1997, 16, 391–396. (4) Koelmans, A. A.; Jonker, M. T. O.; Cornelissen, G.; Bucheli, T. D.; ¨ . Gustafsson, O ¨ . Black carbon: The reverse Van Noort, P. C. M.; O of its dark side. Chemosphere. 2006, 63, 365–377. (5) Thibodeaux, L. J. Recent advances in our understanding of sediment-to-water contaminant fluxes: The soluble release fraction. Aquat. Ecosyst. Health Manage. 2005, 8, 1–9. (6) Schwarzenbach, R. P.; Gschwend, P. M.; Imboden, D. M. Environmental Organic Chemistry; John Wiley & Sons: New York, 2003. (7) Cornelissen, G.; Pettersen, A.; Broman, D.; Mayer, P.; Breedveld, G. Field testing of equilibrium passive samplers to determine freely dissolved native polycyclic aromatic hydrocarbon concentrations. Environ. Toxicol. Chem. 2008, 27, 499–508. (8) Knott, N. A.; Aulbury, J. P.; Brown, T. H.; Johnston, E. L. Contemporary ecological threats from historical pollution sources: impacts of largescaleresuspensionofcontaminatedsedimentsonsessileinvertebrate recruitment. J. Appl. Ecology 2009, 46, 770–781. (9) Granberg, M. E.; Gunnarsson, J. S.; Hedman, J. E.; Rosenberg, R.; Jonsson,P.Bioturbation-drivenreleaseoforganiccontaminantsfrom Baltic Sea sediments mediated by the invading polychaete Marenzelleria neglecta. Environ. Sci. Technol. 2008, 42, 1058–1065. (10) Ortiz, E.; Luthy, R. G.; Dzombak, D. A.; Smith, J. R. Release of polychlorinated biphenyls from river sediment to water under low-flow conditions: Laboratory assessment. J. Envir. Engrg. 2004, 130, 126–135. (11) Jonker, M. T. O.; Koelmans, A. A. Polyoxymethylene solid phase extraction as a partitioning method for hydrophobic organic chemicals in sediment and soot. Environ. Sci. Technol. 2001, 35, 3742–3748. (12) Gustafsson, O; Haghseta, F.; Chan, C.; MacFarlane, J; Gschwend, P. M. Quantification of the dilute sedimentary soot phase: Implications for PAH speciation and bioavailability. Environ. Sci. Technol. 1997, 31, 203–209. (13) Smedes, F. Monitoring of chlorinated biphenyls and polycyclic aromatic hydrocarbons by passive sampling in concert with deployed mussels. In Passive Sampling Techniques in Environmental Monitoring; Greenwood, R, Mills, G.; Vrana, B. , Eds., Elsevier: Amsterdam, 2007; Vol. 48, pp 407-448. (14) Erickson, M. J.; Turner, C. L.; Thibodeaux, L. J. Field observation and modeling of dissolved fraction sediment-water exchange coefficients for PCBs in the Hudson River. Environ. Sci. Technol. 2005, 39, 549–556. (15) Kingston, J. K.; Greenwood, R.; Mills, G. A.; Morrison, G. M.; Persson, L. B. Development of a novel passive sampling system for the timeaveraged measurement of a range of organic pollutants in aquatic environments. J. Environ. Monit. 2000, 2, 487–495. (16) Draper N. R. Smith. H. Applied Regression Analysis; John Wiley & Sons: New York, 1981. (17) Accardi-Dey, A.; Gschwendt, P. M. Assessing the combined roles of natural organic matter and black carbon as sorbents in sediments. Environ. Sci. Technol. 2002, 36, 21–29. (18) Moermond, C. T. A.; Zwolsman, J. J. G.; Koelmans, A. A. Black carbon and ecological factors affect in situ biota sediment accumulation factors for hydrophobic organic compounds in flood plain lakes. Environ. Sci. Technol. 2005, 39, 3101–3109. (19) Armitage, J. M.; Cousins, I. T.; Persson, N. J.; Gustafsson, O.; Cornelissen, G.; Saloranta, T.; Broman, D.; Naes, K. Black carboninclusive modeling approaches for estimating the aquatic fate of dibenzo-p-dioxins and dibenzofurans. Environ. Sci. Technol. 2008, 42, 3697–3703. (20) Eek, E.; Cornelissen, G.; Kibsgaard, A.; Breedveld, G. D. Diffusion of PAH and PCB from contaminated sediments with and without mineral capping; measurement and modelling. Chemosphere 2008, 71, 1629–1638. (21) Connolly, J. P.; Zahakos, H. A.; Benaman, J.; Ziegler, C. K.; Rhea, J. R.; Russel, K. A model of PCB fate in the upper Hudson river. Environ. Sci. Technol. 2000, 34, 4076–4087. (22) Wu, S. C.; Gschwend, P. M. Sorption kinetics of hydrophobic organic compounds to natural sediments and soils. Environ. Sci. Technol. 1986, 20, 717–725. VOL. 44, NO. 8, 2010 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

3019

(23) Ghosh, U.; Tally, J. W.; Luthy, R. G. Particle scale investigation of PAH desorption and thermodynamics from sediment. Environ. Sci. Technol. 2001, 35, 3468–3457. (24) Johnson, M. D.; Keinath, T. M.; Weber, W. J. A distributed reactivity model for sorption by soils and sediments. 14. Characterization and modelling of phenanthrene desorption rates. Environ. Sci. Technol. 2001, 35, 1688–1695. (25) Chai, Y.; Kochetkov, A.; Reible, D. D. Modeling biphasic sorption and desorption of hydrophobic organic contaminants in sediments. Environ. Toxicol. Chem. 2006, 25, 3133–3140.

3020

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 44, NO. 8, 2010

(26) Jonker, M. T. O.; Koelmans, A. A. Sorption of polycyclic aromatic hydrocarbons and polychlorinated biphenyls to soot and sootlike materials in the aqueous environment. Mechanistic considerations. Environ. Sci. Technol. 2002, 36, 3725–3734. (27) Carroll, K. M.; Harkness, M. R.; Bracco, A. A.; Balcarcel, R. R. Application of a permeant/polymer diffusional model to the desorption of polychlorinated biphenyls from Hudson River sediments. Environ. Sci. Technol. 1994, 28, 253–258.

ES903938Z