Download - Novel System for Controlled Investigation of Environmental Partitioning of Hydrophobic Compounds in Water

Published: August 08, 2011

r 2011 American Chemical Society 7834 dx.doi.org/10.1021/es201791x | Environ. Sci. Technol. 2011, 45, 7834–7840

ARTICLE

pubs.acs.org/est

Novel System for Controlled Investigation of EnvironmentalPartitioning of Hydrophobic Compounds in WaterLuca Nizzetto,*,† Rosalinda Gioia,‡ Claire L. Galea,‡ Jordi Dachs,§ and Kevin C. Jones‡

†Norwegian Institute for Water Research, Gaustadall�een 21, NO-0349, Oslo, Norway‡Centre of Chemicals Management, Lancaster Environment Centre, Lancaster University, Lancaster, LA1 4YQ , U.K.§Department of Environmental Chemistry, IDAEA-CSIC, Jordi Girona 18-24 barcelona 08034, Catalunya, Spain

bS Supporting Information

’ INTRODUCTION

A wide range of chemicals classified as priority environmentalpollutants are hydrophobic and semivolatile organic compounds.Examples are many of the persistent organic pollutants (POPs)and substances such as polycyclic aromatic hydrocarbons(PAHs), with demonstrated potential for environmental multi-phase partitioning, including bioconcentration.

The processes controlling the fate of these chemicals inaquatic ecosystems are fundamental for determining their overallenvironmental distribution and exposure of biota. Hydrophobi-city promotes partitioning with particulate1,2 and dissolved

organic matter3 present in the water column. Partitioning,directly or indirectly, has implications for key fate processes,such as the degradation rate,4 the rate of sedimentation,5,6

leaching and runoff,7 and bioavailability in the water columnand sediments.8,9 Partition coefficients, defined as the ratiobetween the concentration of the chemical in suspended

Received: May 26, 2011Accepted: August 8, 2011Revised: August 3, 2011

ABSTRACT: Partitioning behavior of hydrophobic and semivolatile chemicals(such as many POPs and PAHs) in water is key in controlling their environ-mental distribution and fate. A new equilibriummethod is presented here whichallows determination of the equilibrium partition coefficient of hexachloroben-zene with suspended particle (KSPM≈ 337 L gOC�1) in a complex bulk watersample by correcting for a number of sampling artifacts and for the presence ofdissolved matter. The method provides simultaneous experimental determina-tion of the fraction of chemical truly dissolved in water (representing in this caseabout 54% of the bulk water concentration) and that associated to DM (21%).The Henry’s law constant was also experimentally determined during the KSPM

measurements, providing information on the occurrence of partitioning equi-librium in the system for each single observation. Results showed that the highlevel of quality control and accuracy provided confidence intervals for the KSPM

estimates within 1 order of magnitude.

7835 dx.doi.org/10.1021/es201791x |Environ. Sci. Technol. 2011, 45, 7834–7840

Environmental Science & Technology ARTICLE

particulate matter (SPM) or dissolved matter (DM) and that inthe truly dissolved (TD) phase, at equilibrium, are fundamentalmetrics for the assessment and prediction of exposure. Addition-ally, partitioning into living biomass under steady state condi-tions can be described by the bioconcentration factor (BCF).This is a useful metric for determining concentrations in organ-isms at the base of the aquatic food chain, often used in regulatorybioaccumulation assessments.10

Obtaining reliable measurements of these coefficients istechnically challenging for hydrophobic chemicals. In the labora-tory there are difficulties associated with partitioning of the testchemical to the test system materials. For example, partitioningof the chemical can occur between the TD phase and the filter orfilter holder surfaces,11 or flask surfaces, potentially generatingsignificant artifacts in the measurements. The presence ofdissolved organic matter (naturally occurring in field samplesor derived from cell exudates or culture media), can also signi-ficantly affect the measurements.12 These effects cannot be easilyquantified without adequate quality control measures. Arnot andGobas13 have reviewed a large number of reports on BCFassessment and found that about 45% of them did not fulfill atleast one of the quality control criteria.

Here we present and test a novel benchtop experimentalsystem (later referred to as a Multi Media Test Chamber,MMTC), in which simultaneous determination of chemical con-centrations in TD, DM, SPM phases and the head space gasphase (GP) can be performed in a closed and controlled system.The MMTC is fully portable and can be used in the field. Theapproach, based on “single equilibrium”multimedia partitioningwas designed to allow determination of equilibrium conditions,correction for major sampling artifacts, and mass balance basedintrinsic control on the quality of data. This study is presented intwo parts: experiment 1, conducted to demonstrate the suit-ability of the method by determining the dimensionless air�water equilibrium partition coefficient (KAW) of HCB for Milli-Q,synthetic freshwater and seawater, and natural seawater inlaboratory and field conditions; and experiment 2, conductedto assess the suitability of the test system for determining SPM/water partition coefficients in a complex bulk water sample.

’EXPERIMENTAL SECTION

MMTC Description and Principles. The microcosms used inthis study consist of 2-L glass bottles with valved polytetrafluor-oethylene (PTFE) screw caps (Figure 1). The caps have a smallmetal clamp holding a 25-mL glass vial positioned in the headspace. This vial, hereafter called the “probe”, is filled with a phasewhich is equivalent to the inorganic truly dissolved fraction of thewater sample placed in the bottom of the bottle (in this study,pure seawater, fresh water, or Milli-Q water, depending on thetest to be performed). Test chemicals are added to the MMTCbefore the sample is introduced, using a carrier solvent (e.g.,toluene) with a vapor pressure several orders of magnitude higherthan the chemical. After the solvent is evaporated, the sample(e.g., bulk water including suspended particles and dissolvedorganic matter) is introduced in the system. During the incuba-tion, the chemical undergoes partitioning between the differentmedia (e.g., glassware�water�dissolved and suspended matter�gas phase); for example it may degas through the water/headspaceinterface, and from the gas phase into the probe medium. Thiscontinues until a steady state is reached. Assuming no degradationor other losses of the chemical, this state corresponds to the

multimedia single equilibrium partitioning. At the end of theincubation time, samples of the different media in the microcosmare collected and the concentrations in the probe (CP, μg L�1),gas phase (CG, μg L

�1), and bulk phase (CBulk, μg L�1) are

determined.At this stage, assuming that the chemical activity in the probe

and the truly dissolved phase of the water sample are equivalent,it is:

CP¼CTD ð1Þwhere CTD (μg L�1) is the concentration in the truly dissolvedphase of the bulk sample. This relationship stands if (i) thethermodynamic activities of the chemical in the freely dissolvedsolute fraction, the reversibly bound solute of suspended ordissolved fractions and the fraction in the TD phase are identicalat a given temperature;4 and (ii) the volatile exchangeablefraction of the natural dissolved organic matter in the bulk phasehas a negligible capacity for adsorbing the test chemicals. Basedon these assumptions the following coefficients can be deter-mined:

CG

CTD¼ KAW ð2Þ

CBulk � CTD

CTD¼ KBW ð3Þ

where KBW is the dimensionless equilibrium partition coefficientbetween the bulk phase dissolved or suspended in the watersample (e.g., suspended particulate + dissolved organic matter)and the TD phase in water.The experimental value of KAW (obtained for each individual

test) can be compared with separate measurements performedusing the pure water as the bulk phase and the probe, to ensureequilibrium partitioning is achieved between the probe and theGP and to validate eq 1. The occurrence of equilibrium betweenthe probe and gas phase provides information on the state ofequilibrium partitioning for the whole system. This assumes thatthe equilibration time between the probe and the GP is longerthan the glassware�surface water and the water�air exchange, assupported by previous results.12,14

Additionally, assuming that the concentration of the chemicalassociated with the SPM (CSPM) can be reliably measuredthrough filtration or other techniques, it is possible to solve the

Figure 1. MMTC design and concept.



mass balance of the chemical in the bulk water phase anddetermine the concentration associated only with the DM asfollows:

CDM ¼ CBulk � CSPM � CTD ð4Þ

MMTC Preparation. All glassware employed in this study wasthoroughly solvent cleaned and rinsed with Milli-Q water beforeuse. HCB was chosen as the test chemical for this study given itshydrophobic, semivolatile, and persistent character.15 To reduceanalytical effort and improve precision, radioisotope techniqueswere employed. 14C12�HCB (Institute of Isotopes Co., Ltd.,Budapest, Hungary) was used as the only source of HCB. Thechemical and radiochemical purity were >95%. The specificradioactivity was 2.58 KBq/μg.For the air/water partitioning test (experiment 1) about

38 000 Bq of 14C12�HCB (corresponding to 13.7 μg) wasdissolved in 2 mL of toluene and introduced into a 1-L Duranglass bottle. The bottle was kept open and gently swirled to letthe solvent evaporate while distributing the chemical uniformlyon the inner glass wall. After the solvent completely evaporated, 1L of Milli-Q water was added. The bottle was sealed with a capcarrying a PTFE insert and the solution was stirred through theexperiments using a glass-coated stirring bar. This processgenerates a saturated solution of the 14C12�HCB,18 hereaftercalled the stock solution. Aliquots of the stock solution were usedto spike the water sample in the MMTC.The spiking method described here provided levels of

14C12�HCB in the test chamber considerably below saturationand prevented the presence of significant residues of the organicsolvent in the microcosm during measurements. This allowed thefulfilment of fundamental quality criteria for BCF determination.13

Experiment 1: Air�Water Partitioning Test. This test wasperformed under a range of temperatures and salinities, both inlaboratory (Experiment 1a) and in controlled field conditions(Experiment 1b). The rationale for this was (i) to demonstratethe reliability of the method and robustness of assumptionsunder a simplified scenario, and (ii) to provide information onKAW under the conditions set for the later partitioning experi-ments on a complex water matrix. In the laboratory experiment(Experiment 1a), NaCl was added to Milli-Q water to obtainsalinities similar to those of fresh water (0.5 g/L) and seawater(35 g/L). The microcosms were spiked with 125 mL of the stocksolution and diluted with Milli-Q water, pseudo-fresh-water orpseudo-seawater, to reach the final volume of 1 L. Five mL of therespective water phases water were used as probe. Observationswere performed in triplicate. The microcosms were kept in athermostatic chamber on an orbital shaker at 100 rpm for 72 hbefore sampling. Observations were performed at a range oftemperatures between 30 and 0.5 �C.For the controlled field test (Experiment 1b), the same set up

as for the experiment 1a was used. Samples of surface seawaterwere collected on board the RV Hesperides in January�March2009 from 10 stations in proximity to the Antarctic Peninsula,using Niskin bottles. Water samples were prefiltered through a0.5-μm screen and successively processed through tangentialultrafiltration using two Vivaflow 200 (Sartorius Stedim Biotech)units assembled in parallel and equipped with membranes with acut off of 5000 Da, to eliminate most of the colloids, viruses, andsmall bacteria. Microcosms were prepared as described for thelaboratory experiments. Incubation was carried out in a pool

located on the ship main deck constantly flushed with seawater.Therefore, equilibration occurred at the same temperaturesmeasured in surface water. A net was used to cover the pool toreproduce radiation conditions similar to those for 15 m depth inthe ocean. A total of 10 observations were carried out.Experiment 2: Particle�Water Partitioning Test. The applic-

ability of the method to measure partitioning of 14C12�HCB incomplex water samples, e.g., in presence of suspended anddissolved organic matter, was tested. For this test, each micro-cosm bottle was spiked directly with 10 000 Bq of 14C12�HCB(corresponding to 3.88 μg), dissolved in 1 mL of dichloro-methane. After complete evaporation of the solvent, 700 mL ofseawater previously filtered through a 0.45-μm polycarbonatemembrane was added to each microcosm.Samples of SPM were collected and concentrated in the

middle part of Køngsfjord, Svalbard, Arctic (78.93 N, 11.93 E)using a phytoplankton net with a 25-μm nylon mesh. Sampleswere immediately transferred to the Kings Bay AS marinelaboratory in Ny-�alysund and gently screened through a 95-μmnylon mesh, to isolate the 25�95 μm fraction of the total SPM.An aliquot of this fraction was transferred into the triplicatemicrocosms previously spiked (as described above). The sampleswere then diluted to a final volume of 1 L using the filteredseawater. The final particle concentration in the microcosms wasquantified based on the particulate organic carbon (POC)concentration, which was 1.08 ( 0.16 mg OC/L (error repre-sents 2 SD). The probe solution consisted of 5 mL of the samewater used as sample but previously processed through tangentialfiltration as described earlier. Incubation was carried out over 96h at 0 �C, before the first sampling occurred.Sampling from the Microcosm. The gas phase in the micro-

cosm head space was sampled using a Sep-Pak Light C-18cartridge (Waters, Milford, MA, USA), connected through aPTFE luer lock adaptor to one of the valves of the screw cap. Gasphase was sampled (after venting the second valve of the cap)using a small pump running at a calibrated flow rate of 60( 2mLmin�1 at atmospheric pressure. Total sampled volume was 600( 20 mL. Immediately after sampling, the cartridges were eluted3 times with 2 mL of dichloromethane directly into 25-mLscintillation vials andmixed with scintillation liquid (SL) (UltimaGold XR). The last eluted fraction was measured separately toensure complete recovery of the chemical from the sorbent.Scintillation counting was performed over 10 min, using aCanberra Packard Tri Carb 2300 TR, UK, and standard calibra-tion and quench correction techniques. After sampling of thehead space was carried out, the microcosms were opened and theprobe was recovered. SLwas added directly into the probe vial forcounting. Collection of the sample from the microcosm wascarried out using a class A graduated glass pipet (previouslyrinsed with acetone and Milli-Q water). Five mL of the samplewas collected, transferred to a scintillation vial, andmixedwith SLfor counting.In experiment 2, 20-mL samples were collected through a glass

syringe and filtered through a 13-mm diameter glass fiber filter(GFF) (nominal pore size was 0.6 μm). Syringe filtration wasperformed gently using a stainless steel filter holder assembledwith two PTFE O-rings. The last 5 mL of the filtrated liquid wastransferred to a scintillation vial to count the radioactivity in thetotal dissolved fraction to obtain the value of the filtrateconcentration (CF, ng mL�1). After filtration, GFFs were care-fully removed from the filter holder and transferred to ascintillation vial for the measurements of the particle bound



fraction. Sampling of SPM was repeated 3 times at daily intervalsfrom each microcosm.Characterization and Correction of Filtration Artifacts. Ad-

sorption on GFFs and filter holder wet surfaces can represent aconsiderable source of artifacts, especially because PTFEO-ringshave a non-negligible sorptive capacity for hydrophobic chemi-cals in the truly dissolved phase.16 This was controlled byperforming an adsorption test on a set of 6 replicates under thesame temperature conditions used for experiment 2. The test wasperformed by sampling from a positive control (e.g., a testperformed with no addition of SPM) using the same procedureand materials as for the real samples. Filter adsorption wasdefined in terms of the ratio (KGFF-W, L) between the mass ofthe chemical adsorbed on the filter membrane (after filtering thevolume (VS = 20 mL) of the water sample) and the TD phaseconcentration, calculated as follows:

KGFF�W ¼ AGFFC

CPð5Þ

where AGFFC is the mass μg) of the chemical measured on thefilter surfaces.The concentration of 14C12�HCB associated with suspended

particles (CSPM, μg L�1) in experiment 2 was therefore calcu-lated by correcting for filter adsorption effects as follows:

CSPM ¼ ðAGFFS � ðCP 3KGFF�WÞÞVS

ð6Þ

Quality Assurance/Quality Control. All media and reagentswere measured for background β decay emissions. These valueswere used for blank correction.The effects of sorption of HCB on glassware surfaces on water

concentrations was monitored and the results are reported in textSI1 in the Supporting Information. Equilibration time betweenthe glass surfaces and the water phase was <10 h (see Figure SI5)In experiment 2 triplicate “positive” controls were added, in

which ultrafiltered seawater was used both as the sample and theprobe. This was done to ensure the occurrence of partitioningequilibrium conditions between air and gas phase. Only datafrom tests in which CBulk was not significantly different from CP

in the positive controls were taken into consideration. In addi-tion, the occurrence of partitioning equilibrium between SPMand the truly dissolved phase was checked by repeating twice themeasurement from each MMTC after 24 and 48 h from the firstsampling.Adsorption on the probe vial glass surfaces can cause artifacts

in the measurements or result in an apparent imbalance betweenCTD and CP, invalidating the results. This was tested on a subsetof observations (N = 10) by measuring the amount of radio-activity associated with the probe glass vial only, and the probecontent, separately. The radioactivity of the chemical adsorbedon the probe vial glass was below detection limits at 30 �C, butrepresented 15( 2% of the total radioactivity of the chemical inthe probe at 0 �C. Results for the 0 �C temperature werecorrected for this effect.The occurrence of artifacts associated with the use of the glass

pipet during sampling of the bulk water phase in the microcosmswere checked by rinsing the pipet (after sampling) with acetonedirectly into a scintillation vial in order to extract the chemicaladsorbed on the inner glass surface. No measurable levels ofradioactivity were sorbed on the pipet.

Themeasure of the gas phase concentration of 14C12�HCB inthe head space of the microcosm was corrected by the dilutioneffect produced by venting the chamber as described in the SI(text SI2). The possibility of breakthrough during gas phasesampling through the C18 cartridges was excluded based on testsperformed by deploying two cartridges in series. Elution effi-ciency was also tested by backup extraction tests.In experiment 2, the quality of each individual measurement

requiring filtration was checked by a mass balance approachwhere the mass of the chemical directly measured in an equiva-lent volume of the bulk test phase ABulk (μg) was compared withthe different fractions obtained from the filtration process andthe fraction adsorbed on the filter holder and sampling media asfollows:

ABulk ¼ Af ilt þ AGFF þ Asurf ð7Þ

where Afilt, AGFF, and Asurf (μg, in 20 mL of processed sample)are the masses (μg) of the chemical in the filtered water, GFF,syringe and filter holder surfaces, respectively.Asurf was measuredby rinsing the filter holders and syringe with methanol (afterfiltering and removing the GFFs) and collecting the solvent into ascintillation vial. The identity expressed by eq 7 was verified in allcases at the level of 98 ( 4% (error representing 95% con-fidence), confirming high levels of control on chemical losses dueto sample handling.

’RESULTS AND DISCUSSION

Data Quality. Detectability, Consistency, and Solubility. Theradioactivity of 14C12�HCB was measured in all the phases atlevels significantly above the background. Replicate measure-ments differed by a maximum of 16% and 24% from the meanvalue for CP and CG, respectively. For CBulk higher consistencywas achieved, with the maximum divergence from the meanbeing ∼7% across all the different measurements/experiments.In experiment 1 the concentration measured in the bulk phase

averaged 0.07 μg L�1 (at 0 �C) and 0.11 μg L�1 (at 25 �C). Inexperiment 2, before the addition of SPM, the concentration inthe bulk dissolved phase (CS) was 1.12 ( 0.18 μg L�1 (errorrepresents 2 SD). After the addition of the SPM the measuredCTD was 0.9 ( 0.15 μg L�1. Therefore all the tests wereconducted with 14C12�HCB concentrations between a factorof 10 and 5 below the reported solubility of 5 μg L�1.15

Occurrence of Equilibrium Partitioning. All samplings wereperformed at least 72 h after the beginning of the equilibrationtime. No statistically significant difference was observed betweenthe CP and CBulk in Experiment 1a and 1b and the positivecontrols in the Experiment 2 (e.g., test performed using Milli-Qor pure seawater both as probe and sample) (independently fromtemperature), suggesting the system had reached equilibriumpartitioning. The absolute value of the difference between CBulk

and CP (expressed in percentage) was always <17%. Thisvariability could be fully explained by the random error onreplicate measurements, and therefore was assumed as the limitfor ensuring equilibrium partitioning had been reached andvalidating the test.For the tests with the complex water sample (Experiment 2)

the occurrence of equilibrium was confirmed by the consistencyof the values of CP and CG from the 3 daily successive samplingsfrom each microcosm. This corroborates the validity of theassumptions made on the dynamics of the air water and particle



water exchange, suggesting that, at least for HCB, the method isable to determine that equilibrium has been achieved.Adsorption on Filters.The value ofKGFF-W averaged 1.16( 0.20.

Given the high consistency among replicate measurements, thesevalues were used in eq 6 to correct the results of CSPM.Experiment 1a: Air�Water Equilibrium Partitioning

(Laboratory Test). The value of KAW for HCB was determinedfor the different temperatures as the ratio between CG and CP orCTD. These were used to calculate the value of Henry’s lawconstant H (Pa m3 mol�1) as follows:

H ¼ KAW 3R 3T ð8Þwhere R is the ideal gas constant (m3 Pa K�1 mol�1) andT (K) isabsolute temperature.The values of H(25�) for the

14C12�HCB in Milli-Q water,fresh water, and seawater are reported in Table 1 along withliterature data. Reported errors represent estimates of the 95%confidence intervals calculated as 2 SD, assuming normality.Results obtained here are generally in good agreement withprevious reports.17�20 In particular our results are highly con-sistent withmost recent assessments based on data reanalysis andintrinsic thermodynamic consistency.21

Temperature dependence of H was assessed by plotting Ln Hvs 1/T for the water samples. Figure SI3 represents such analysisand plots the regression curves. Coefficients for the regressionfunction Ln H = A/T + B and quality of the regressions arereported in Table SI3. The product between the slope A and thegas constant R is the enthalpy of phase change (ΔWAH, K 3 J 3mol�1). The ΔWAH value measured here was consistent be-tween samples at different salinities and ranged between 53 and55 K 3 J 3mol

�1 (see Table SI3). These values are also consistentwith previous reports from Jantunen and Bidleman 17 and tenHulscher et al.18

The significant difference observed between the values H forthe Milli-Q (or freshwater) and seawater are due to the so-called“salting out” effect,22 which can be described by the followingrelationship:

Hsw ¼ HMilliQ 3 10Ks 3Csalt ð9Þ

whereKS (Lmol�1) is the Setchenow’s coefficient andCsalt is the

concentration of NaCl (g L�1) in the water. The value of KS,calculated using eq 9 and the experimental data, was 0.29 (0.10 L mol�1 (2 SD), which is in agreement with previousreports.22

The uncertainty on the value of H for many semivolatilecompounds (including HCB and many other POPs) currentlyrepresents a major limitation to adequately estimate and predictthe direction and magnitude of air�surface exchanges in theenvironment.23�26 This hampers assessment of the global fate ofthese compounds, particularly considering the potential funda-mental role of ocean surface water in influencing the fate ofsemivolatile pollutants.27,28 The present method provided resultsfor HCBwith a high level of control (i.e., verifying the occurrenceof partitioning equilibrium, consistency among subsequent mea-surements from the same microcosms, coupled gas phase andwater phase concentration determination) and fulfilling qualityassurance measures on the conditions.These results support the validity of the assumptions under-

pinning the method presented here and show that it can bereliably employed for studies on the partitioning processes ofhydrophobic chemicals in aqueous media. Next it is appropriateto consider more complex matrices in the system.Experiment 1b: Air�Water Equilibrium Partitioning

(Controlled Field Test). The average temperature at whichAntarctic water samples were incubated varied between �0.3and 3 �C, depending on the water masses encountered in thedifferent stations. LnH measured in natural seawater rangedbetween 1.65 and 2.57, and values were correlated with the

Table 1. Experimental Values of H Derived Here and Comparison with Literature Data

method H (Pa 3 m33 mol-1) at 25 �C

concentration in the water

during the test (μg L-1)

concentration in air

(μg L-1) at 25 �C reference

this study: Milli Q 58 ( 2 (2 SD) 0.10�0.11 0.0025�0.0028

seawater 91 ( 20 0.10�0.11 0.0035�0.0039

freshwater 62 ( 6 0.11�0.12 0.0028�0.0030

gas stripping 35 a 20�40 17

71 a NA 20

133 (23 �C) a NA 29

49 (20 �C) a 15 19

41 ( 4 (20 �C) 1 18

thermodynamic method 24 ( 8 NA 29

thermodynamic method/data compilation 65 a NA 21a Errors not reported. Errors represent 2SD.

Figure 2. Experimental values and estimates of the uncertainty dis-tribution for the HCB concentration in the different fractions of acomplex water sample. Curves represent the profile of the histograms ofthe estimates. Uncertainty was traced throughout eqs 2�6 using MonteCarlo analysis (N = 10 000) assuming normality.



inverse absolute temperature at the 90% confidence levels.Despite the very narrow range of temperature considered here,the temperature dependence and values were consistent withlaboratory data and data from literature (Table 1).These results support the viability of the MMTC method as a

useful tool to conduct experiments in controlled field conditions,starting from natural water media.Experiment 2: Partitioning Measurements in a Complex

Water Sample. The concentrations measured in the bulk watersample were significantly higher (p < 0.05) than CP. This con-dition was necessary in order to allow application of eqs 3 and 4.The ratio between CG and CP provided values of KAW consistentwith those measured in experiment 1.Reproducibility of measurements, expressed as 2 times the

relative standard deviation of triplicate observations, was asfollows: ( 8%, 32%, 40%, 53%, and 84% of the mean value ofCBulk (3.10 μg L

�1), CP (1.7 μg L�1), CG (0.004 μg L�1), CSPM

(0.65 μg L�1), and CDM (0.62 μg L�1), respectively. Theseresults are shown in Figure 2.CDM was assessed indirectly, through eq 4. Notably, assump-

tion of normality and propagation of uncertanties producednegative estimates of CDM with a frequency <5%. These mustbe regarded as nonsense values indicating the need of a higherlevel of resolution. However, it was possible to assess the qualityof the calculated central tendency of CDM through eq 4 bycomparing it with the mean value of CDM data derived from themeasurement of concentration in the eluted phase (Cfilt) duringthe filtration process.This was measured by correcting the value of the total HCB

concentration in the last 5 mL of the eluate for the effect ofadsorption on the filter membrane and filter holder and syringesurfaces, as follows:

CDMðexpÞ ¼ Cf ilt þ ðCP 3KGFF�W 3 0:05Þþ ðCP 3Ksurf�W 3 0:05Þ � CP ð10Þ

where Ksurf-W (L), similarly to KGFF-W, is the ratio between theamount of chemical adsorbed on the syringe and filter holdersurfaces and that in the TD phase, calculated as follows:

Ksurf�W ¼ Asurf C

CPð11Þ

using experimental data from the controls. Given that 20 mL ofthe samples was processed through filtration, the coefficient 0.05was introduced to scale the contribution of adsorption effects tothe volume of 1 mL (to be consistent with the concentrationunits), assuming linear release of the chemical from the TD phaseto the filter and sampling material surfaces. Notably, this experi-mental approach presented a wider uncertainty compared to theindirect approach of eq 4, however, there was no statisticallysignificant difference between CDM and CDM(exp), evidencingthe robustness of the method presented here.Particle Water Equilibrium Partition Coefficient (KSPM)

and Potential Sources of Error. Adsorption on GFF accountedfor about 45 ( 15% of the total radioactivity measured on thefilter, leading to a potential overestimation of CSPM (if correctionthrough eq 6 was not performed).The 33 ( 3% of the chemicals in filtered water was adsorbed

on sampling material surfaces (e.g., syringe and filter holder),potentially producing a significant underestimation of the

dissolved concentrations if Cfilt data were used as value for theconcentration in water.Finally the presence of dissolved organic matter had a strong

influence on the distribution of chemical in water, in fact CDM

accounted for about 21% of the total chemical in the bulk water.The particle water equilibrium partition coefficient (KSPM)

(L gOC�1) was calculated as follows:

KSPM ¼ CSPM

CTD 3 ½OC�ð12Þ

where [OC] (g L�1) is the concentration of organic carbonassociated to suspended particles. Itsmean valuewas 337 L gOC�1.The uncertainty of the assessment was traced through thecalculations foreseen by eq 6 and 12, using Monte Carlo analysis.Results are shown in Figure 3.In the method presented here, the variance associated to the

measurements of CTD represented the major contributor to thetotal uncertainty on the estimates of KSPM. It must be remarkedalso that the correction for artifacts on themeasurements ofCSPM

(such as sorption on filters and sampling materials) may con-siderably increase the accuracy of the method, however account-ing for the error associated to these artifacts in the uncertaintyanalysis produces broader boundaries of confidence for the finalKSPM value.In rare cases (P < 0.006) negative estimates of CSPM were

obtained. These obviously resulted in negative estimates ofKSPM.Negative values must be regarded as nonsense and were likelyderived from a poor resolution of the sampling in defining thedistribution of CPOC. A higher number of replicate measure-ments would likely reduce the frequency of negative estimates. Intheir extensive review on BCF, Arnot and Gobas13 reported thatfor hydrophobic compounds such as phenanthrene, p,p'-DDT, orHCB the uncertainty in the value of BCF can be about 2.5 ordersof magnitude, even when basic quality criteria were applied to thetest method. The present analysis showed the good methodperformance in terms of quality control, resulting in 95%confidence intervals for KSPM ranging within 1 order of magnitude.This variance appears to be built into these types of complexmeasurements having implications for bioaccumulation hazardassessment and regulations given that the approach commonlyused is often pass/fail, based on “bright line” criteria.13

This study describes a novel approach to the assessment of thepartitioning of hydrophobic chemicals in complex water samples.TheMMTC allows monitoring of truly dissolved water concentra-tions and coupling between water and air compartments. Thesepossibilities may be exploited in future for the implementation of

Figure 3. Frequency distribution of the equilibrium particle�waterpartition coefficient (KSPM) estimates. Green lines represent 5th and95th percentiles, red lines represent lower and upper quartiles, the blackline represents the average.



more detailed assessments of chemical fate and transfer insimplified plankton communities (for examples assessing parti-tioning and transfer of chemicals between a simple producer�primary consumer system or assessing the partitioning propertieson natural or engineered dissolved organic matter, etc.).

’ASSOCIATED CONTENT

bS Supporting Information. Further details on methodolo-gy and method performance. This information is available free ofcharge via the Internet at http://pubs.acs.org.

’AUTHOR INFORMATION

Corresponding Author*E-mail: [email protected]; tel.: +47 98215393; fax: +4722185200.

’ACKNOWLEDGMENT

We thank Dr. Katrine Borg�a from the Norwegian Institute forWater Research for the support during the tests in Ny-�aysund,Prof. Kirk Semple for the scientific advise and training and thecrew and colleagues on board the RV Hesperides. The ATOS-2cruise on board RV Hesperides was funded by the SpanishMinistry of Science and Innovation.

’REFERENCES

(1) Axelman, J.; Broman, D.; Naf, C. Field measurements of PCBpartitioning between water and planktonic organisms: Influence ofgrowth, particle size, and solute-solvent interactions. Environ. Sci.Technol. 1997, 31, 665–669.(2) Sobek, A.; Gustafsson, O.; Hajdu, S.; Larsson, U. Particle-water

partitioning of PCBs in the photic zone: a 25-month study in the openBaltic Sea. Environ. Sci. Technol. 2004, 38, 1375–1382.(3) Poerschmann, J.; Kopinke, F. D. Sorption of very hydrophobic

organic compounds (VHOCs) on dissolved humic organic matter(DOM). 2.Measurement of sorption and application of a Flory-Hugginsconcept to interpret the data. Environ. Sci. Technol. 2001, 35, 1142–1148.(4) Zwiernik, M. J.; Quensen, J. F.; Boyd, S. A. Residual petroleum in

sediments reduces the bioavailability and rate of reductive dechlorina-tion of aroclor 1242. Environ. Sci. Technol. 1999, 33, 3574–3578.(5) Larsson, P.; Andersson, A.; Broman, D.; Nordback, J.; Lundberg,

E. Persistent organic pollutants (POPs) in pelagic systems. Ambio 2000,29, 202–209.(6) Axelman, J.; Broman, D.; Naf, C. Vertical flux and particulate/

water dynamics of polychlorinated biphenyls (PCBs) in the open BalticSea. Ambio 2000, 29, 210–216.(7) Sakai, S.; Urano, S.; Takatsuki, H. Leaching behavior of PCBs

and PCDDs/DFs from some waste materials. Waste Manage. 2000,20, 241–247.(8) Fava, F.; Piccolo, A. Effects of humic substances on the bioavail-

ability and aerobic biodegradation of polychlorinated biphenyls in amodel soil. Biotechnol. Bioeng. 2002, 77, 204–211.(9) Ghosh, U.; Zimmerman, J. R.; Luthy, R. G. PCB and PAH

speciation among particle types in contaminated harbor sediments andeffects on PAH bioavailability. Environ. Sci. Technol. 2003, 37, 2209–2217.(10) OECD. Test No. 305: Bioconcentration: Flow-through Fish Test;

OECD Publishing.(11) Magnusson, K.; Magnusson, M.; Ostberg, P.; Granberg, M.;

Tiselius, P. Bioaccumulation of C-14-PCB 101 andC-14-PBDE 99 in themarine planktonic copepod Calanus finmarchicus under different foodregimes. Marine Environ. Res. 2007, 63, 67–81.

(12) Gerofke, A.; Komp, P.; McLachlan, M. S. Bioconcentration ofpersistent organic pollutants in four species of marine phytoplankton.Environ. Toxicol. Chem. 2005, 24, 2908–2917.

(13) Arnot, J. A.; Gobas, F.A.P.C A review of bioconcentration factor(BCF) and bioaccumulation factor (BAF) assessments for organicchemicals in aquatic organisms. Environ. Rev. 2006, 14, 257–297.

(14) Dachs, J.; Eisenreich, S. J.; Baker, J. E.; Ko, F. C.; Jeremiason,J. D. Coupling of phytoplankton uptake and air-water exchange ofpersistent organic pollutants. Environ. Sci. Technol. 1999, 33, 3653–3660.

(15) Mackay, D.; Shiu, W. Y.; Ma, K. C. Illustrated Handbook ofPhysical-Chemical Properties and Environmental Fate for Organic Chemi-cals; Lewis Publishers: Boca Raton, FL, 1991.

(16) Gerofke, A.; Komp, P.; McLachlan, M. S. Stir bar contamina-tion: A method to establish and maintain constant water concentrationsof poorly water-soluble chemicals in bioconcentration experiments.Water Res. 2004, 38, 3411–3419.

(17) Jantunen, L. M.; Bidleman, T. F. Henry’s law constants forhexachlorobenzene, p,p'-DDE and components of technical chlordaneand estimates of gas exchange for Lake Ontario. Chemosphere 2006, 62,1689–1696.

(18) Ten Hulscher, T. E. M. Temperature dependence of Henry’slaw constants for selected chlorobenzenes, polychlorinated biphenylsand polycyclic aromatic hydrocarbons. Environ. Toxicol. Chem. 1992,11, 1595–1603.

(19) Oliver, B. G. Desorption of chlorinated hydrocarbons fromspiked and anthropogenically contaminated sediments. Chemosphere1985, 14, 1087–1106.

(20) Atlas, E.; Velasco, A.; Sullivan, K.; Giam, C. S. A radiotracerstudy of air-water exchange of synthetic organic compounds. Chemo-sphere 1983, 12, 1251–1258.

(21) Shen, L.; Wania, F. Compilation, evaluation, and selection ofphysical-chemical property data for organochlorine pesticides. J. Chem.Eng. Data 2005, 50, 742–768.

(22) Schwarzenbach, R. P.; Gschwend, P. M.; Imboden, D. M.Environmental Organic Chemistry, 2nd ed.; JohnWiley & Sons: Hoboken,NJ, 2003.

(23) Nizzetto, L.; Lohmann, R.; Gioia, R.; Jahnke, A.; Temme, C.;Dachs, J.; Herckes, P.; Di Guardo, A.; Jones, K. C. PAHs in air andseawater along a North-South Atlantic transect: Trends, processes andpossible sources. Environ. Sci. Technol. 2008, 42, 1580–1585.

(24) Gioia, R.; Nizzetto, L.; Lohmann, R.; Dachs, J.; Temme, C.;Jones, K. C. Polychlorinated Biphenyls (PCBs) in air and seawater of theAtlantic Ocean: Sources, trends and processes. Environ. Sci. Technol.2008, 42, 1416–1422.

(25) Iwata, H.; Tanabe, S.; Sakal, N.; Tatsukawa, R. Distribution ofpersistent organochlorines in the air and surface seawater and the role ofocean on their global transport and fate. Environ. Sci. Technol. 1993,27, 1080–1098.

(26) MacLeod, M.; Fraser, A. J.; Mackay, D. Evaluating and expres-sing the propagation of uncertainty in chemical fate and bioaccumula-tion models. Environ. Toxicol. Chem. 2002, 21, 700–709.

(27) Dachs, J.; Lohmann, R.; Ockenden, W. A.; Mejanelle, L.;Eisenreich, S. J.; Jones, K. C. Oceanic biogeochemical controls on globaldynamics of persistent organic pollutants. Environ. Sci. Technol. 2002,36, 4229–4237.

(28) Nizzetto, L.; Lohmann, R.; Gioia, R.; Dachs, J.; Jones, K. C.Atlantic Ocean Surface Waters Buffer Declining Atmospheric Concen-trations of Persistent Organic Pollutants. Environ. Sci. Technol. 2010, 44,6978–6984.

(29) Atlas, E.; Foster, R.; Glam, C. S. Air sea exchange of highmolecular-weight organic pollutants - laboratory studies. Environ. Sci.Technol. 1982, 16, 283–286.