David a . Alvarez 2002

7/28/2019 David a . Alvarez 2002

1/7

Development of the Permeability/Performance Reference CompoundApproach for In Situ Calibration ofSemipermeable Membrane Devices

J A M E S N . H U C K I N S , * ,

J I M M I E D . P E T T Y , J O N A . L E B O ,

F E R N A N D A V . A L M E I D A , K E E S B O O I J ,

D A V I D A . A L V A R E Z ,

W A L T E R L . C R A N O R ,

R A N D A L C . C L A R K , A N DB E T T Y B . M O G E N S E N |

Columbia Environmental Research Center, U.S. GeologicalSurvey, 4200 New Haven Road, Columbia, Missouri, 65201,Instituto de Quimica, State University of Campinas,13081-970 Campinas, Sao Paulo, Brazil, Netherlands Institute

for Sea Research, 179 0 AB Den Burg, The Netherlands, andNational Environmental Research Institute,Frederiksborgvej 339, DK-4000 Roskilde, Denmark

Permeability/performance reference compounds (PRCs)are analytically noninterfering organic compoundswith moderate to high fugacity from semipermeablemembrane devices (SPMDs) that are added to the lipidprior to membrane enclosure. Assuming that isotropicexchangekinetics (IEK)applyandthatSPM D-waterpartitioncoefficients are known, measurement of PRC dissipationrate constants during SPMD field exposures and laboratorycalibration studies permits the calculation of an exposureadjustment factor (EAF). In theory, PRC-derived EAFratios reflect changes in SPMD sampling rates (relative tolaboratorydata)duetodifferences in exposuretemperature,

membrane biofouling, and flow velocity-turbulence at themembrane surface. Thus, the PRC approach shouldallow for more accurate estimates of target solute/vaporconcentrationsinanexposuremedium.Under someexposureconditions, the impact of environmental variables onSPMD sampling rates may approach an order of magnitude.

The results of this study suggest that most of the effectsof temperature, facial velocity-turbulence, and biofoulingon the uptake rates of analytes with a wide range ofhydrophobicities can be deduced from PRCs with a muchnarrower range of hydrophobicities. Finally, our findingsindicate that the use of PRCs permits prediction of in situSPMD sampling rates within 2-fold of directly measuredvalues.

In t roduct ionTraditionally, passive sampling devices (PSDs) have beenused to monitor vapors of volatile organic chemicals (VOCs)in occupational environments (1). Detection limits are

generally not an issue for these applications because healthconcerns of VOCs are generally triggered at milligrams percubic meter to grams per cubic meter levels (2). However,PSD detection limits are often a concern when samplingsemivolatile organic compounds (SVOCs) in environmentalmedia. This is because trace levels of SVOCs (, milligramsper cubicmeter) maybe bioconcentrated in organismtissues

to potentially toxic levels.It is instructive to note that the maximum PSD uptake

rates areachievedwhen therate-limitingbarrier to chemicalvapor or solute transport is the external boundary layer (i.e.,a thin air or water layer between the sampler exterior andthe bulk environmental medium). In other words, the rateof mass transfer or PSD uptake across a series of barriers orlayers can be no greater than the rate of supply. However,changes in the flow velocity-turbulence of the exposuremediumaffect theeffectivethicknessof theexternal boundarylayer of a PSD. Since mass-transfer resistance is directlyproportionalto boundary layerthickness,the sampling ratesof analytes will vary with the hydrodynamics/aerodynamicsof the deployment site. Under boundary layer control, PSDdesign features other than the external surface area for

chemicalexchange willhave little or no effect on linear uptakerates.

Unlike most PSDs (3), lipid-containing semipermeablemembrane devices (SPMDs) are designed for high samplingrates of hydrophobic SVOCs in environmental media (4-10). In light of the above sampling rate discussion, it is notsurprising that work by Booij et al. (11) and Huckins et al.(12) indicates that the SPMD uptake rates of SVOCs with logoctanol-water partitioncoefficients (Kows)g4.4are generallyunder aqueous boundary layer control (assuming watervelocities at the SPMD membrane surface of 4.4 include nearly all polychlo-rinated biphenyl (PCB) congeners and chlorinated dioxins

and furans and most organochlorine pesticides (OCs) andpolycyclic aromatic hydrocarbons (PAHs). SPMD samplingrates are also affected by temperature and biofouling (12,13). Because of the large range of potential environmentaleffects on the sampling rates of high Kow SVOCs (11-17), itappears impractical to conduct calibration studies for allexposure scenarios. Thus, SPMD environmental exposuresrequiring more quantitative measures or media concentra-tions would be limited to a narrow range of exposureconditions that can be directly related to calibration data.

Huckins et al. (4, 5) proposed the use of permeability(membrane control of uptake rates)/performance (aqueousboundarylayercontrolof uptake rates) reference compounds(PRCs) to address the issue of biofouling effects on SPMDsampling rates. PRCs are analytically noninterfering organic

compounds with moderate to high Kows that are added totheSPMD lipid prior to membrane enclosure.The measuredrateof PRCloss during environmental exposuresis comparedto similar valuesderived from laboratory calibration studies.This novel in situ calibration approach is based on theoryand experimental evidence that PRC dissipation rate con-stants (kePRC-fs) at sampling sites are related to the uptakerate constants (kua-fs) of target compounds.In other words,isotropic exchange kinetics (IEK) govern the accumulationof hydrophobic SVOCs by SPMDs (4, 11, 13). Moreover,research (11, 12) has indicated that PRCs can be used toassess the effectsof sitehydrodynamics andtemperature onSPMD sampling rates.

* Correspondingauthorphone: (573)876-1879; fax: (573)876-1896;e-mail: [email protected].

U.S. Geological Survey. State University of Campinas. Netherlands Institute for Sea Research.| National Environmental Research Institute.

Environ. Sci. Technol. 2002, 36, 85-91

10.1021/es010991w CCC: $22.00 2002 American Chemical Society VOL. 36, NO. 1, 2002 / ENVIRONMENTAL SCIENCE & TECHNOLOGY 9 85Published on Web 11/22/2001


2/7


3/7

from actual exposures, the values used for KSPMDs may notbe acceptable for the derivation of in situ uptake rates ofanalytes.

Methods for Estimation of kua-f. Two PRC-based ap-proaches can be used to estimate in situ SPMD samplingrates(i.e.,kua-fs) of target compounds.Both methods requireknowledge of analyte KSPMD, and for some chemicals, theKSPMD may be temperature dependent. See the subsequentdiscussion on the role of temperature in the magnitude ofKSPMD. Regardlessof themethod used,PRCsmust be chosenwhose loss rates are governed by the same rate-controlling

mechanisms as the target analytes.Method 1, or the EAF approach, is largely based on the

assumption that the effects of environmental variables (e.g.,facialvelocity-turbulence andbiofouling) on theuptakeratesof chemicals under aqueousboundarylayercontrol (generallythose with log Kows in the range of about 4.4-8.0) can becloselyapproximated by the effects on theloss rates of PRCs,under the same environmental conditions. For boundarylayer-controlled chemicals, the range of PRC log Kows istypically limited to 4.4-5.5, which is due to the opposingrequirements of boundary layer control versus measurableresidue (PRC) losses from SPMDs. Implicit in the PRCapproach for chemicals under boundary layer control is thatthe effects of environmental variables on analyte exchangerates remain relatively constant across a wide range of log

Kows.Method 2 is based on the use of PRCs that bracket therange of analyte Kows, as described by Booij et al. (11), andcomescloser to approximatingan isotopic dilution approachoften used for the analysis of environmental samples. Thismethod may require the use of PRCs with logKows rangingfrom about 3.0 to 8.0. The approach is based on directderivation of analyte in situ kea-f by regression analysis ofdata from multiple PRCs (i.e., PRC log Kow vs log kePRC-f).However, KSPMDs still must be determined in the laboratory.After obt ainingkea-f and KSPMD values, in situ kua-f values ofanalytes are readily derived by eq 3. The key assumptionunderlying this approach is that kePRC-fs are measurable forcompounds with log Kows greater than 5.5. In practice,detectinglosses of PRCs with logKows > 5.5 are oftendifficultunless exposures of extended duration are conducted inwarm, highly turbulent environments. For example, earlierwork (12) has shown that the ke for benzo[b]fluoranthene is0.002 (d-1) at 26 C and a flow rate of


4/7

flow-through system to the 16 native PP PAHs. The watertemperature was maintained at 18C,andSPMDs(n) 4 foreach sample period) were sampled through time (days 0, 4,7, 14,and 21). Water concentrations of PP PAHs rangedfrom17 to 57 ng/L. Both the dissipation of perdeuterated PAHand the uptake of native PP PAHs were measured in thesame devices. Day 21 data was selected for modeling thelevel of uptake impedance relative to log Kow, as shown inFigure 1. However, levels of biofouling impedance weredetermined for all exposure periods.

Flow Effects. Booij et al. (11) have described the experi-ments conducted to assess the effects of flow velocity-turbulenceon PRCuptake andclearancerates.Briefly,waterwas pumped from a large chamber through a smallerchamber in a closed loop. A propeller stirrer was used tomaintain relatively high turbulence/flow in the large cham-ber, while much lower flow/turbulence was maintained bythe pumping action in the smaller chamber. Test chemicalconcentrations in the water were maintained by placingsediments spiked withPCBs,PAHs,and chlorinatedbenzenesinto the larger chamber at the beginning of the test. Thetemperature was maintained at 9C. The estimatedvelocitiesof water at the membrane surfaces in the smaller and largerchambers were0.03 and30 cm/s, respectively.A mixtureof PRCs including perdeuterated PAH and native PCBs werespikedintoSPMDlipid. SPMDs were sampled atsevenpointsin time up to and including 56 d.

Temperature Effects. Experiments to measure the effectsof temperature on KSPMDs were conducted in sealed glassjars. All test containers or jars (n) 9/temperature) contained1.6 L of deep-well water with a total organic carbonconcentration(TOC) of 0.26mg/L (personalcommunication,John McCarthy, Oak Ridge National Laboratory, Oak Ridge,TN), a Teflon-coated stir bar, and one 3.8-cm SPMD. [14C]-Phenanthrene,[14C]-2,2,5,5-tetrachlorobiphenyl(TCB), and[14C]-p,p-DDE were spiked separately into jars (n ) 3 foreachtest chemical and sampletime). Concentrations rangedfrom 0.48 to 1.6 g/L, which represent


5/7

same compounds were as follows: acenaphthene, M) 280cm3/dg and P) 320 cm3/dg; phenanthrene, M) 240 cm3/dg and P) 270 cm3/dg; and pyrene, M) 380 cm3/dg andP ) 370 cm3/dg. Computed values of phenanthrene andpyrene kePRCs from biofouled SPMDs were based on twopoint derivations (eq 7) because measurable losses of thesePRCs were not detected untild 21. However, eq 7 representsthe most practical and commonly used approach for deriva-tion of SPMD ke-PRC values, as it requires fewer samples.

By dividingMvaluesinto Pvalues (P/M),the bias or errorencompassing the difference between predicted and mea-

sured values of kua can be estimated. Application of thisapproach to our kua data gave the following P/M ratios:acenaphthene, 2.3 for clean and 1.2 for biofouled SPMDs;phenanthrene, 1.8 for clean and 1.1 for biofouled SPMDs;andpyrene, 1.6for cleanSPMD and1.0 forbiofouled SPMDs.Using earlier kinetics data from clean SPMDs with 99%triolein (5), we derived a P/Mratio for phenanthrene of 1.4.With only one exception, the predicted values of kuas werehigher than measured values. This possible bias appearedless significant for biofouled SPMDs. Overall, the mean P/Mratio was 1.5, and the CV was (30%.

Because three separate measurements are required toderive P/Mratios, error propagation is unavoidable. The CVof replicate values of the parameters used in the P/Mderivations were generallye (20%, but for computation

purposes, we assume an error of(

20% for each variablemeasured. Therefore, the cumulative variance associated witherror propagation (CVt) canbe estimated using thefollowingrelationship:

where the subscript t is tot al. Using eq 12, the CVt for thesederivations is (35%, which may explain much of the errorassociated with P/M estimates.

Constancy of EAFs Relative to Analyte Hydrophobicity.Compounds used for PRCs are usually selected to have logKows g 4.0 but e 5.5 (method 1). If the EAF values remainrelatively constant for analytes with logKows ranging fromabout 4.0 to 8.0, then EAFs will be directly applicable for

most target compounds. However, kua-cal values must beavailable forat least oneset of exposure conditions (e.g.,refs11, 12, and 17).

Biofouling. Earlier,Huckinset al.(13,17) showed that thelevel of biofouling-induced reduction in the loss rate of thePRC acenaphthene-d10 from SPMDs was similar to the levelof thereduction in theuptakerate of thenativeacenaphthene,i.e., IEK applied. However, the suitability of acenaphthene-d10 orotherPRCswithlogKowse5.5for estimating biofoulingimpedance of the uptake of SVOCs with logKows > 5.5 wasnot adequately addressed. To further explore this issue, weexamined thebiofoulingimpedance of PPPAH uptakerelativeto compound log Kow for four exposure periods. Figure 1summarizes data from the 21-d exposure period, and theresults suggest that the level of biofouling impedance isrelatedto PPPAH Kow. Equation 12 representsthe bestlinearregression model fit of the overall relationship:

The correlation coefficient (R2) is 0.82 and p e 0.0001.Obviously, % Ib is highly correlated to logKow, and separatelinearmodels could be applied to each phase ofthis apparentbiphasic relationship (Figure 1). The biphasic nature of therelationship appears to relate to the level of biofoulingimpedance when chemicals are under membrane andboundarylayer control. If an R2 of 0 (i.e., no significantchangein % Ib with increasingKow) was observed for the model fitof data representing PP PAHs withKows g 4.4, this would be

a clear indication that the exchange rates of PRCs with logKows g 4.4 but e 5.5 are directly applicable to high Kow PPPAHs. On the basis of the statistics given above, this is notthe case for biofouling impedance. Nevertheless, the slopeof the regression line shown in Figure 1 is not steep, as Ibincreases only10% per log unit. Thus, nomore thana 20%difference in uptake impedance would be expected for atarget compound with a logKow of 7.0 and a PRC with a logKow of 5.0. The CV for SVOC analyses often exceed (20%.

Examination of experimental data for all exposure times(4, 7, 14, and 21 d) showed that the level of % Ib gradually

increased throughout the exposure study. For example, the% Ib at day 7 ranged from 26 to 55%, and the mean value (xj) 45%) was 13% less than the mean of the day 21 values(xj ) 58%).Overthe range oftest compoundKows, differencesbetween theday 21 and theday 7 values remained constant,and deviations could be explained by variance associatedwith analytical measurements. In other words, the slopes ofthe logKow- % Ib relationship at days 7 and 21 were similar.Unfortunately, biofouling is generally heterogeneously dis-persed on an SPMD membrane. Therefore,it is not practicalto directly measure biofilm thickness for the estimation of% Ib.

Mass-transfer theory suggests that the effect of soluteKowon the flux across a biofilm would be relatively small andthat the level of biofouling impedance is more related to the

thickness of the water portion of the biofilm matrix. Thebiofilmcan be regarded as a water-richlayerwith embeddedbiotic and abiotic organic carbon. The biofilm partitioncoefficient Kbwis expectedto be representative of a weightedmean of periphytic organism bioconcentration factors andsediment organic carbon (biofilm)-water partition coef-ficients or the Koc. In general, we assume that the organiccarbon from sediment particles represent a much smallermass fraction than the organic carbon found in periphyticorganisms and chemical release follows first-order kinetics.The diffusion coefficients of SVOCs in biotic and abioticorganic carbon and water are expected to be inverselyproportional to chemical Kow. Examination of the group Ib/DbKbw, which models resistance to mass transfer (Ib) acrossa biofilm, shows that the impact on Ib of falling diffusion

coefficientsof SVOCswith proportionallyincreasingKbwsorKocs should be small and that the product of DbKbw wouldbe relatively constant. Remember that SPMD kua values areinversely proportional to Io. The slope of the regression lineshown in Figure 1 appears to be a measure of the biofilmorganic carbon deviation from IEK.

Assuming that nonideal interactions (e.g., biphasic de-sorption from sediment particles) do not control interphaseexchange rates of SVOC residues in the biofilm, water-phaseresistance may still be the rate-limiting step for chemicalswith logKows g 4.4. In this case, the overall thickness of thewater layer (i.e., the biofilm plus the aqueous boundary layer)largely controls the magnitude of the resistance. Thus, itfollows that the slope of the regression line is not steep forcompounds with increasingKows once water-phase resis-tance becomes dominant. Also, it follows, that as the biofilm

thickens with exposure time, the slope of the portion of thecurve representing boundary layer controlled compounds(log Kows g 4.4) should remain about the same, but themagnitude of the y intercept will increase.

Facial Velocity-Turbulence. Theoryandthe workof Booijet al. (11, 14, 15) indicate that the effects of changes in facialvelocity-turbulence are independent of log Kow when ananalyte is under boundary layer control. To confirm thishypothesis, Booij et al. (11) used two flow regimes andsimultaneously measured the amount of test chemicals(largely PCB congeners and native and perdeuterated PPPAHs) concentratedin or lost from SPMDs through time. Asdescribed earlier (11, 14, 15), several models were used to

CVt ) ([CVke]2

+ [CVKSPMD]2

+ [CVku]2)1/2 (12)

% Ib ) 100%(1 - kuab/kua) ) 10 logKow+ 3.45 (13)

VOL. 36, NO. 1, 2002 / ENVIRONMENTAL SCIENCE & TECHNOLOGY 9 89


6/7

analyzethe dataand to derivekuas.In thiswork,we comparedthe ratios or EAFs derived from the two flow-related kuas.Table 1 shows the results of test chemicals with a fairly widerange of Kows. The mean EAF ratio is 3.2 (n ) 15) and theCV) 20.8% for analytes with logKows from 4.14 to 7.67, ora range of3.5 log units. For pyrene (logKow ) 5.3), whichis under aqueous boundary layer control, the EAF ratio is

3.6.Therefore,the use of PRCs, suchas perdeuterated pyrene,with logKows in the range of 4.5-5.5 should be applicableto the derivation of flow turbulence-mediated EAFs or thekua-f of analytes with logKows ranging from about 4.5 to 7.7.

Temperature andthe KSPMD. Clearly exposure temperatureaffects exchange kinetics (12, 17, 20), but temperature alsomayaffect thermodynamic propertiessuch as theKSPMD.Leoet al. (21) have shown that the effect of temperature on Kowis small for a number of compounds (i.e., 0.001 -0.01 logunits/C) and may be either positive or negative. Both LDPEKmws and triolein-water partition coefficients (KLws) havebeen correlated to Kows (17-19, 22). In this study, weexaminedthe potentialeffectsof an environmentally relevantrange of exposure temperatures on model compound KSPMD.Table 2 summarizes the data.

On the basis of our findings, the effects of temperatureonthe magnitudeofKSPMDs appearto be molecularstructurespecific. For example, a 2.6-fold difference was observedbetween phenanthrene KSPMDs measured at 8 and 30 C,whereas no statistically significant temperature-related dif-ferences were observed for KSPMD values of 2,2,5,5-TCB andp,p-DDE. The results for 2,2,5,5-TCB and p,p-DDE areconsistent with those ofLeo et al.(21), androom temperatureKSPMDs measured for these nonplanar chemicals appear tobe satisfactory for use in eqs 2, 4, and 8-10. Furthermore,the use ofKSPMDs, derived from regression models relatinglogKmwsandlogKLws (typically at room temperature) to logKows (17, 19, 22), generally appear to be valid for use in eqs8-10. However, the apparent sensitivity of PAH KSPMDs to

temperature (possibly other planar compounds as well)suggests that their KSPMDs may have to be determined for arange of exposure temperatures before use in eqs 8-10. Apotential explanation for the apparent sensitivity of phenan-threne KSPMD values to experimental temperaturemay relateto stronger adsorption-type interactionsof planar moleculeswith the LDPE as the temperature falls. LDPE is about 50%crystalline at room temperature (23). Lower temperaturespotentially increase polymeric crystal content and thestrength of their surficial interactions with solutes. Clearly,additional research is needed to confirm this hypothesis.

Method Overview. This work and other investigations(11-16) suggest that the impacts of a wide range environ-mental exposure conditions on SPMD sampling rates maybe as high as 10-fold for facial velocity-turbulence effects,4-fold for temperature effects, and 3-4-fold for biofoulingeffects. Fortunately, our findings indicate that the effects ofthese variables on the uptake rates of SVOCs with a widerange of Kows are largely reflected by changes in the lossrates ofPRC with a much narrowerrange ofKows. In addition,it appears that the KSPMDs of some SVOCs will change littlewith most exposure temperatures (planar aromatic com-pounds such as PAHs seem to be an exception). Determi-nation of the exact magnitude of the error associated withSPMD/PRC-derived estimates of the environmental con-centrations of dissolved SVOCs is problematic because of

our inability to perform reliable matrix spikes in the fieldand the lack of independent methods for the selective andaccurate sampling of dissolved SVOCs. However, the experi-ments summarized in thiswork suggest that theoverall errorin SPMD-derived water concentration estimates can bereduced to about 2-fold by using PRCs.

In view of these research results,it appears that thereareno technical barriers to theuse of PRC data to adjust SPMD/PSD kua-cals to reflect the effects of exposure conditions onin situ sampling rates (i.e., kua-fs). Thus, the application ofPRCsto estimate site-specific samplingrates of SVOCs shouldgenerally improve the accuracy of water concentrationestimatesand reduce the amount of calibration datarequiredfor the use of SPMDs and PSDs.

AcknowledgmentsWe thankthe American Petroleum Institute for partialsupportin this project. Also, we appreciate the valuable input andmanuscript reviewsof Robert Gale andKathyMcCarthy, andwe appreciate Pam Haverland and Mark Ellersieck for thestatistical analysis of data from the biofouling study.

Literature Cited

(1) Fowler, W. K. Am. Lab 1982, 14, 80-87.

(2) U.S. Department ofHealthand Human Services.NIOSH PocketGuide to Chemical Hazards; Publication 94-116; NIOSH Pub-lications: Cincinnati, OH, 1994.

(3) Underhill, D. W.; Fiegley, C. E. Anal. Chem. 1991, 63, 1011-1013.

(4) Huckins, J. N.; Meadows, J. C.; Manuweera, G. K.; Lebo, J. A.12thAnnualMeetingof theSocietyof EnvironmentalToxicologyand Chemistry; November 3, 1991; P315.

(5) Huckins, J. N.;Manuweera, G. K.;Petty, J. D.;Mackay, D.;Lebo,J. A. Environ. Sci. Technol. 1993, 27, 2489-2496.

(6) Petty, J. D.; Huckins, J. N.; Zajicek, J. L. Chemosphere1993, 27,1609-1624.

(7) Prest, H. F.; Huckins, J. N.; Petty, J. D.; Herve, S.; Paasivirta, J.;Heinonen, P. Mar. Pollut. Bull. 1995, 31, 306-312.

(8) Lebo, J. A.; Gale, R. W.; Petty, J. D.; Tillitt, D. E.; Huckins, J. N.;Meadows, J. C.; Orazio, C. E.; Echols, K. R.; Schroeder, D. F.;Inmon, L. E. Environ. Sci. Technol. 1995, 29, 2886-2892.

(9) Gale,R. W.;Huckins,J. N.;Petty,J. D.;Peterman,P. H.;Williams,L.L.; Morse,D.; Schwartz,T. R.;Tillitt, D.E. Environ.Sci. Technol.1997, 31, 178-187.

TABLE 1. Constancy of EAFs Relative to Flow -Turbulence andHydrophobicity

compoundlogKowa

EAF(kua-30cm/kua-0.03cm)b

1,2,3-trichlorobenzene 4.14 2.3hexachlorobenzene 5.73 4.12,4-dichlorobiphenyl 5.09 3.92,4,4-trichlorobiphenyl 5.60 3.92,2,5,5-tetrachlorobiphenyl 6.10 3.82,2,4,4,5-pentachlorobiphenyl 6.60 3.5

2,2,3,3,6,6-hexachlorobiphenyl 6.70 3.42,2,4,4,5,5-hexachlorobiphenyl 6.90 2.92,2,3,4,4,5,5-heptachlorobiphenyl 7.67 2.32,2,3,3,4,4,5,5-octachlorobiphenyl 7.67 2.9phenanthrene 4.46 3.4pyrene 5.30 3.6chrysene 5.61 3.5benzo[a]pyrene 6.35 2.4benzo[g,h,i]perylene 6.90 2.1

a Values largely from Mackay et al. (24, 25). b From eq 9, theEAF isactually (kua-30cm/KSPMD)/(kua-0.01 cm/KSPMD), but KSPMD can be i gnoredbecause it did notchange, i.e., temperaturewas thesamefor bothflowvelocity-turbulence studies.

TABLE 2. Effects of Temperature on Log10 KSPMD Values (xj (

CV (%)) of Selected Compoundsexposuretemp(C) phenanthrene 2,2,5,5-TCB p,p-DDE

8 4.70 ((0.6) 5.55 ((17.0) 5.92 ((4.2)18 4.53 ((1.7) 5.66 ((9.1) 6.04 ((18)30 4.29 ((4.5) 5.53 ((10.0) 5.93 ((2.4)

90 9 ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 36, NO. 1, 2002


7/7

Documents

David a . Alvarez 2002