lable at ScienceDirect

Environmental Pollution 158 (2010) 142–147

Contents lists avai

Environmental Pollution

journal homepage: www.elsevier .com/locate/envpol

The potential effect of differential ambient and deployment chambertemperatures on PRC derived sampling rates with polyurethane foam(PUF) passive air samplers

Karen Kennedy a,*, Darryl W Hawker b, Michael E Bartkow a, Steve Carter c,Yukari Ishikawa a, Jochen F Mueller a

a The University of Queensland, EnTox (The National Research Centre for Environmental Toxicology), 39 Kessels Rd., Coopers Plains QLD 4108, Australiab Griffith University, School of Environment, Nathan QLD 4111, Australiac Queensland Health Forensic and Scientific Services, Coopers Plains QLD 4108, Australia

Internal deployment chamber temperatures rather than ambient tempe

ratures may be required to accurately estimate PRC-based sampling rates.

a r t i c l e i n f o

Article history:Received 19 May 2009Received in revised form21 July 2009Accepted 22 July 2009

Keywords:Passive air samplingSampling ratePerformance reference compoundWind speedTemperature

* Corresponding author. Tel.: þ61 7 32749009; fax:E-mail address: k.kennedy@uq.edu.au (K. Kennedy

0269-7491/$ – see front matter � 2009 Elsevier Ltd.doi:10.1016/j.envpol.2009.07.031

a b s t r a c t

Performance reference compound (PRC) derived sampling rates were determined for polyurethane foam(PUF) passive air samplers in both sub-tropical and temperate locations across Australia. These estimateswere on average a factor of 2.7 times higher in summer than winter. The known effects of wind speedand temperature on mass transfer coefficients could not account for this observation. Sampling rates areoften derived using ambient temperatures, not the actual temperatures within deployment chambers. Ifdeployment chamber temperatures are in fact higher than ambient temperatures, estimated sampler-airpartition coefficients would be greater than actual partition coefficients resulting in an overestimationof PRC derived sampling rates. Sampling rates determined under measured ambient temperaturesand estimated deployment chamber temperatures in summer ranged from 7.1 to 10 m3 day�1 and2.2–6.8 m3 day�1 respectively. These results suggest that potential differences between ambient anddeployment chamber temperatures should be considered when deriving PRC-based sampling rates.

� 2009 Elsevier Ltd. All rights reserved.

1. Introduction

Polyurethane foam (PUF) passive air samplers (Shoeib andHarner, 2002) are one of the most widely used passive air samplersfor monitoring semivolatile organic chemicals (SVOCs) in air andhave been used to cover broad spatial scales, including remoteareas and background sites (Harner et al., 2006a; Jaward et al.,2004a, 2005; Pozo et al., 2006). Apart from investigating large scaleambient distribution patterns, these samplers have been applied tothe assessment of urban–rural gradients (Harner et al., 2004,2006b), seasonal exposure studies (Gouin et al., 2005b; Klanovaet al., 2006; Motelay-Massei et al., 2005) and to study indoor andoutdoor exposure patterns for different classes of SVOCs (Wilfordet al., 2004). The low cost and ease of application of these samplersensures these types of sampling strategies can expeditiouslyprovide information about the levels of SVOCs in a range ofcontexts.

þ61 7 32749003.).

All rights reserved.

To improve the accuracy of ambient concentration estimatesderived from the amount accumulated by PUF samplers, it is impor-tant to be able to account for the influence of ambient environmentalparameters on sampling rates. Parameters which may influence thevolume of air sampled in-situ include ambient temperature and windspeed. In order to determine in-situ sampling rates, performancereference compounds (PRCs) (Bartkow et al., 2004; Booij et al., 1998;Huckins et al., 2002) also referred to as depuration compounds maybe loaded into PUF samplers prior to deployment (Gouin et al., 2005a,2008; Moeckel et al., 2009; Pozo et al., 2004, 2006; Santiago andCayetano, 2007). The elimination of PRCs from the PUF samplershould respond isotropically, with respect to target compoundaccumulation, to changes in environmental conditions experienced inthe field.

The total air volume sampled VA (m3) may be estimated from theloss of PRCs during the deployment period (Equations (1) and (2)).In these expressions, ke PRC is the first

VA ¼ RS t (1)

RS ¼ ke PRC KPUF-AIRVS (2)

K. Kennedy et al. / Environmental Pollution 158 (2010) 142–147 143

order elimination rate constant for PRC loss from the PUF samplers(day�1), RS is the sampling rate or volume of air sampled per day(m3 day�1), t is the deployment period (days), KPUF-AIR is thesampler-air partition coefficient (dimensionless) and VS is thevolume of the PUF sampler. Elimination rate constants may beestimated from the slope of lnðN0 PRC

Nt PRCÞ vs: t, where N0 PRC and Nt PRC

are the average time zero and post-deployment period amounts ofPRC in the PUF sample respectively.

If air side resistance to mass transfer is dominant for PUFsamplers, expressing ke PRC in terms of the air side mass transfercoefficient (kA (m s�1) Equation (3)) enables the identification ofthe potential influence of ambient environmental parameters onsampling rates (Bartkow et al., 2005, Shoeib and Harner, 2002).

RS ¼

kA

KPUF-AIR ðCHAMB TÞ

AS

VS

!KPUF-AIR ðAMB TÞVS

¼ kAAS ¼DA

dAAS (3)

Molecular diffusivity (DA m2 s�1) in the air side boundary layer isa weak function of temperature while wind speed will reduce thethickness of the air side boundary layer (dA), so increasing eithertemperature or wind speed can potentially increase sampling rates.In order for our PRC-based estimates of in-situ sampling rates (RS)to be accurate a ratio of unity should exist between the KPUF-AIR thatare derived using laboratory based calibration data (Shoeib andHarner, 2002) and ambient temperature estimates (KPUF-AIR (AMB T))and the actual KPUF-AIR under the influence of temperature pre-vailing in the deployment chamber (KPUF-AIR (CHAMB T)), which isincorporated into ke PRC.

Current theory indicates that a temperature difference ofapproximately 10 �C at ambient temperature ranges such as mightbe expected between summer and winter deployments in sub-tropical and temperate Australia for example, should result inminimal effects on sampling rates where mass transfer is viapassive diffusion. This is based on a semi-empirical relationship formolecular diffusivity as a weak function of ambient temperature (T)(Fuller et al., 1966). Since the relationship between temperatureand sampling rates using theoretical predictions and field basedstudies is not expected to be significant, the same sampling rate(3.5–4 m3 day�1) in different seasons is often assumed (Gouin et al.,2005b; Harner et al., 2006b; Klanova et al., 2006; Motelay-Masseiet al., 2005). Agreement between PUF derived ambient concen-tration based on these assumptions and simultaneous high volumemeasurements, is generally reasonable (Gouin et al., 2005b; Jawardet al., 2004b; Klanova et al., 2006).

The influence of meteorological parameters in the fieldincluding wind speed, has previously been assessed for a large dataset of 42 passive sampling events (28 days duration) and 168 highvolume samples (4 days or 14% of event duration) at one site(Klanova et al., 2008). This study found sampling rates (Equation(4)) of approximately 7 m3 day�1 (N 3038 Gumotex Breclav PUF) or5.2 m3 day�1 (corrected to planar surface area of Tisch Environ-mental PUF), for predominantly vapour phase compounds such asPCBs, where NS is the amount accumulated in PUF in linear phasesampling; and CA is the ambient concentration determined fromhigh volume sampling.

RS ¼NS

CAt(4)

Vapour phase sampling rates for the site in this study werea factor of two higher in winter and this was attributed to a factor oftwo increase in wind speeds in this season. A more recent study of

40 days duration at one site in Uppsala Sweden with a height basedwind speed gradient, validated PRC-based sampling rates at twoout of eight heights using co-deployed low volume active airsamplers (Moeckel et al., 2009). This study confirmed that PRC lossreflected the wind profile and therefore yielded a concentrationprofile consistent with the active samplers. Relationships betweenwind speed and sampling rates have been established for PUFdeployed in typical ‘‘flying saucer’’ chambers in a wind tunnel(Tuduri et al., 2006).

In this study we derive in-situ PRC-based sampling rates insummer and winter at eight sites across Australia. Differencesbetween sampling rate estimates in different seasons at the samesite are evaluated by quantifying the potential influence of windspeed and temperature between seasons based on establishedsemi-empirical relationships and calibration studies. The relation-ships for temperature and wind speed are based on the influence oftemperature ratios on diffusivity and laboratory calibrations of theinfluence of wind speed on sampling rates respectively. Thepotential of known wind and temperature effects to account forobserved differences in PRC-based sampling rates seasonally wasassessed. Confounding effects related to the application of ambienttemperature measurements to derive partition coefficients used toestimate the PRC-based volume of air sampled are examined.

2. Materials and methods

2.1. Passive air sampling seasonal deployments

PUF passive air samplers (Tisch Environmental TE-1014; AS ¼ 3.62 � 10�2 m2;VS ¼ 1.95 � 10�4 m3; rS ¼ 2.36 � 104 g m�3, where rS is sampler density) weredeployed at each site in summer (January–February 2007) and winter (July–August2007) for between 40 and 50 days. Samplers were deployed in a two-disc config-uration (Thompson et al., 2007) per sampling chamber (‘‘flying saucer’’ design) withtwo replicate chambers/samples (2-PUF sample�1) at each site.

Three replicate field blank samples were prepared and transported to deploy-ment sites in each of the three states of Australia (Western Australia (WA),Queensland (QLD) and South Australia (SA)) but were not exposed to ambient air atthe sites. The sampling sites used within this study fall into several climate classi-fication zones based on temperature, relative humidity and seasonal rainfall asdetailed in electronic supplementary material (ESM) Table S1 and Figure S1. Mostsites used are classified as temperate climates with two sites (Brisbane and Glad-stone) being sub-tropical. PUF pre-extraction/extraction and analysis are describedin detail in ESM.

2.2. PRC loading

PRCs (polychlorinated biphenyls (PCBs): 21, 30, 145, 204) were loaded in 20 mLof HPLC grade hexane (10 mL side�1) and dried (5 min side�1) under purifiednitrogen. Loading rates were higher (by a factor of 4) in summer for the relativelymore volatile PCBs to ensure analytically significant amounts remaining in thesamplers upon exposure. The N0 PRC and Nt PRC estimates were based on the averageof the replicate field blanks (n ¼ 3) and replicate exposed samples from each site(n ¼ 2) respectively. Recoveries of PCB 204 averaged 103% in exposed samples andwere used to correct for differences in loading rate. Reproducibility of PRC loadingwas good with average (and range) relative standard deviations of 2.7% (0.37–6.0%).

3. Results and discussion

3.1. Seasonality of PRC loss

PRC-based sampling rates at each site could only be determined inboth seasons from the elimination of PCB 30. This was primarily dueto PCB 30 being the only PRC with any significant loss from the PUFsampler in winter (an average percentage loss of 32%). In summer theaverage percentage loss for this PRC was 85%, while PCB 21 averaged44%. Due to the high summer loading rates however there was noconcern as to analytical uncertainty associated with high percentageloss of PCB 30 (average 32 ng recovered), which is potentiallya problem when elimination is high (>80%) (Pozo et al., 2004). Referto ESM for KPUF-AIR (AMB T) estimates for PCB 30 in summer and winter

K. Kennedy et al. / Environmental Pollution 158 (2010) 142–147144

and PCB 21 in summer (Table S2) and for the average elimination rateconstants, half-lives and times to equilibrium (teq 99%) for the loss ofthese PRCs from PUF during these deployments (Table S3). The teq 99%

(days) estimated averaged 104, 517 and 334 days for PCB 30 insummer and winter and PCB 21 in summer respectively.

3.2. PRC derived sampling rate estimates in summer and winter

Deployment periods and seasonally averaged temperatures areprovided in Table 1. These deployment periods are less than thetimes to effective equilibrium indicated by the elimination of thisPRC. PCB 30-based, average sampling rate estimates (Equation (3))for each site and season are shown in Table 1. This treatmentassumes the value of KPUF-AIR based on laboratory calibrations andambient temperatures (KPUF-AIR (AMB T)) (refer ESM Table S2) is thesame as the value related to the elimination rate constant (ke)which is a function of the temperature in the deployment chamberthrough its relationship to KPUF-AIR (CHAMB T).

Derived in this manner, the sampling rate averages � standarddeviations (and ranges) in summer and winter were 8.2 � 1.3(7.1–10) and 3.2 � 0.96 (2.3–5.1) m3 day�1 PUF�1 respectively. Thehalf-lives for the elimination of PCB 30 in summer (average 16 days)would indicate that elimination rate constants have been estimatedfrom the curvilinear region of the elimination profile. It is unlikelythat a potential approach to equilibrium for this PRC is accountingfor elevated PRC derived sampling rate estimates in summer as theslopes used to derive the elimination rate constants would bereduced in this case and hence if anything sampling rates would beunderestimated. The sampling rate estimates based on theelimination of the higher log KPUF-AIR (AMB T) PRC PCB 21 which isunlikely to be approaching equilibrium (refer ESM Table S4)average � standard deviations (and ranges) were 10 � 2.0(8.1–13) m3 day�1 PUF�1 provide further confirmation of elevatedsampling rate estimates for these sites in summer.

The PCB 30-based sampling rate estimates were significantlydifferent for each site in different seasons (paired t-test; P< 0.0001)and were higher for the summer deployments at each location.There was no statistically significant relationship between PCB30-based RS EST (m3 day�1) and ambient temperature within eachseason. However, the linear regression for all measurementsincluding data from both seasons was significant (R2 ¼ 0.89;P < 0.0001) (refer ESM Fig. S2). On a seasonal basis temperaturewould appear to be influencing the PCB 30-based sampling rate

Table 1PCB 30-based sampling rate estimates determined using average ambient temper-ature for summer and winter.

Sampling sites Summer Winter

Ta [�C] tb [days] RS ESTc

[m3day�1]T [�C] t [days] RS EST

[m3day�1]

Perth 24 40 7.3 14 44 2.3Bunbury 22 40 8.1 14 44 3.4Gladstone 26 40 10 17 41 5.1Brisbane 25 41 10 16 45 3.6Mutdapilly 25 41 7.8 14 45 3.7Gawler 24 42 7.5 11 50 2.5Adelaide 24 42 7.7 12 50 2.3Mount Gambier 19 41 7.1 9 46 2.9

a Average ambient temperature from 10 to 30 min observations.b Deployment period.c PRC-based sampling rate estimates (Equation (3) based on surface area of

a single PUF disc; KOA from relationship with relative retention time (RRT) PCB30 ¼ 0.3598 (Harju et al., 1998) for multi-ortho PCBs (log KOA ¼ a þ b (RRT) wherea ¼ �2.0296 þ 1310.7/T and b ¼ �5.5305 þ 5879.8/T where T is temperature inKelvin (Harner and Bidleman, 1996)); KPUF-AIR¼ rS� 10^ ((0.64� log KOA)� 3.1774)(Shoeib and Harner, 2002) provided in ESM Table S2.

estimates prompting a re-examination of the role of temperature ininfluencing these estimated values.

3.3. The role of wind speed and temperature in the observeddifferences in PRC-based seasonal sampling rate estimates

PCB 30-based sampling rates derived using ambient tempera-tures were on average a factor of 2.7 times higher in summer thanin winter. By using a single PRC in each season the observeddifferences in sampling rate should reflect the influence of in-situconditions and not compound specific differences. From Equation(3) (Rs ¼ kA AS) the parameter which may be influenced by ambienttemperature and wind speed is the air side mass transfer coefficientkA. The ratio of air side mass transfer coefficients between summerand winter is equivalent to the seasonal sampling rate ratio. Byexpressing kA in each season in terms of diffusion coefficients (DA)and air side boundary layer thickness ðdAÞ we can define ratioswhich account for the influence of temperature and wind speedeffects respectively, between seasons. The product of these ratios isa seasonal wind and temperature factor (WTF) (Equation (5)).

RS SUMMER

RS WINTER¼ kA SUMMER

kA WINTER

¼�

DA SUMMER

DA WINTER

�TEMP

�dA WINTER

dA SUMMER

�WIND

¼ WTF (5)

The potential influence of temperature on diffusion coefficientsmay be quantified using the ratio of ambient temperatures (K) ineach season to the power of 1.75 (Fuller et al., 1966). Similarly, usingcalibration data relating wind speed to sampling rates measured ina laboratory wind tunnel (Tuduri et al., 2006) the wind factorbetween seasons may be quantified from the ratio of thesesampling rate estimates. Using average ambient temperatures(Table 1) and average ambient wind speed measurements for eachsite (Table 2) the influence of each of these factors has beenquantified separately and then combined together and expressed asa seasonal WTF for each site (Table 2). The proportions of theobserved seasonal factor difference which can be accounted for bythe WTF are also provided in Table 2.

The predicted effects of temperature on diffusion coefficientsonly account for an average factor increase of 1.1 seasonally. TheWTFs quantified predict that sampling rates should increaseslightly between winter and summer in most cases but not to theextent observed. In certain cases (Bunbury and Gawler) a reductionin wind speed should keep sampling rates relatively consistent oreven decrease sampling rates between winter and summer. Theconsistent and unexpectedly large increase over and above thatpredicted by established relationships with temperature and windspeed suggests that the basis on which sampling rates werecalculated may be incorrect.

3.4. The influence of temperature on sampling rates

An important factor to be considered within the context ofambient passive sampling in outdoor environments is the influenceof actual field temperatures experienced by the PUF samplersdeployed in metal sampling chambers. It is possible that the inte-rior chamber temperature is influenced by conduction of heat fromthe metal chamber surface or from radiation reflected as heat bythe chamber, or other surfaces within the vicinity of the chamber tothe surrounding air. While the chamber effectively attenuates theinfluence of wind speed (Bartkow et al., 2004; Tuduri et al., 2006)and potentially photo-transformation (Bartkow et al., 2006), thesesame design features may be contributing to increased internaltemperatures within the chamber particularly when wind speeds

Table 2Seasonal sampling rate ratios, individual temperature and wind speed factors and the combined estimated seasonal wind and temperature factor for each site.

Sampling SitesRS EST SUMMER

RS EST WINTER

a

DA SUMMER

DA WINTER

!TEMP

b

Wind SpeedSummer [m s�1]c

Wind SpeedWinter [m s�1]

dA WINTER

dA SUMMER

!WIND

d

WTF e % f

Perth 3.2 1.06 5.3 4.8 1.26 1.34 42Bunbury 2.4 1.05 3.5 3.9 0.94 0.99 –Gladstone 2.0 1.05 2.9 1.7 1.26 1.33 66Brisbane 2.8 1.05 2.3 1.5 1.18 1.25 45Mutdapilly 2.1 1.07 2.9 2.3 1.12 1.19 58Gawler 3.0 1.08 4.3 4.6 0.83 0.90 –Adelaide 3.4 1.07 3.3 2.8 1.09 1.17 35Mt. Gambier 2.4 1.06 5.1 4.8 1.16 1.23 50

a Seasonal ratio of PRC-based sampling rate estimates (Table 1).b Temperature (K) ratio of average ambient temperature at each site (from Table 1) to the power of 1.75 (Fuller et al., 1966).c Average ambient wind speeds from 10 to 30 min observations.d The relative influence of wind speed on boundary layer thickness in different seasons estimated from the ratio of sampling rates estimated using wind tunnel calibration

relationships (Tuduri et al., 2006). In detail: vair int¼ vair ext/4; for vair ext > 4 m s�1 RS¼ 38.81 vair int� 27.92 (estimated based on values (2 points) provided in this study) while forvair ext < 4 m s�1 RS ¼ 6.35 vair int þ 4.54 where vair ext is the external air velocity and vair int is the internal chamber air velocity.

e Seasonal wind and temperature factor (Equation (5)).f Proportion of observed seasonal ratio that WTF may account for when an increase in sampling rates in summer is predicted (WTF > 1).

K. Kennedy et al. / Environmental Pollution 158 (2010) 142–147 145

are relatively low. Discrepancies between ambient temperaturemeasurements and actual air temperatures within and around thedeployment chamber are likely, since ambient measurements aretaken so they are not influenced by direct sunlight and heat.

It is reasonable to assume that the influence of differentialsampling chamber and ambient temperatures on actual diffusioncoefficients would be negligible compared to the effects on PUF-AIRpartitioning. Temperature differences of 20 �C vs. 10 �C for examplewill still only influence molecular diffusivity by a similar factorwhile the influence on partition coefficients and hence ke would bemuch higher.

Estimated partition coefficients (KPUF-AIR) are typically deter-mined using available ambient temperature data, while partitioncoefficients reflecting actual in-situ temperature within thedeployment chamber are incorporated into the elimination rateconstant for PRC loss (Equation (3). If sampling chamber temper-atures were greater than ambient temperatures then KPUF-AIR (AMB T)

would be greater than KPUF-AIR (CHAMB T), resulting in a ratio for theseparameters >1. Therefore, PRC-based sampling rates estimatedusing Equation (3) would be higher by this factor difference. Somerecent calibration data suggests that field based measurements ofKPUF-AIR may in fact be lower than expected which supports thisidea (Chaemfa et al., 2008). It is unlikely that the use of the 2-discconfiguration is contributing to this temperature effect as whenvalidated previously (Thompson et al., 2007) elimination rateconstants for PCB PRCs were equivalent for both the single disc and

Table 3Derivation of KPUF-AIR under the influence of in-situ deployment chamber temperatures tsampling rates at these sites in summer.

Site Sampling ratesa

RS SUMMER (m3 day�1)Elimination rate constantsb

ke SUMMER (days�1)

Perth 3.0 0.041Bunbury 3.4 0.041Gladstone 6.8 0.067Brisbane 4.5 0.061Mutdapilly 4.5 0.048Gawler 2.2 0.042Adelaide 2.7 0.044Mt. Gambier 3.5 0.030

a Equation (6).b Elimination rate constants from the slope of ln(N0 PCB 30/Nt PCB 30) vs. time.c Summer partition coefficient under the influence of deployment chamber temperatu

Table S2.d Deployment chamber temperatures estimated using log KOA values based on regr

footnotes to Table 1.

2-disc configuration suggesting partition coefficients and hencetemperatures within the deployment chambers are the same.

3.5. Quantifying potential errors in PRC-based sampling rates andestimating sampler chamber temperature

As a simplification it may be assumed that any temperatureinfluence on PRC-based sampling rates is minimised in winterwhen average daily solar exposure is lower. In this case the parti-tion coefficients derived from ambient temperature measurementsin winter provide a reasonable approximation to those influencedby deployment chamber temperature. Under such conditions,sampling rates derived from Equation (3) should be accurate andreflect the true sampling rate. If it can be further assumed that theratio of mass transfer coefficients between summer and winter isreasonably described by the WTF defined earlier and no otherprocesses affect mass transfer, then the true or corrected summersampling rate should be the product of the winter sampling rateand the WTF factor as shown in Equation (6), where ke SUMMER

(day�1) is the PRC elimination rate

RS SUMMER ¼kA SUMMER

kA WINTERRS WINTER ¼ RS WINTERWTF

¼ ke SUMMER KPUF-AIR ðCHAMB TÞVS (6)

constant in summer. Corrected summer sampling rates derived

ogether with these estimated deployment chamber temperatures and the predicted

Partition coefficientsc logKPUF-AIR (CHAMB T)

In-situ deploymentchamber temperatured T (�C)

(dimensionless) (m3 g�1)

5.6 1.2 405.6 1.3 385.7 1.3 345.6 1.2 405.7 1.3 365.4 1.1 475.5 1.1 445.8 1.4 31

re (Equation (7)), for comparison log KPUF-AIR (AMB T) for PCB 30 is provided in ESM

ession with log KPUF-AIR (m3 g�1) and relationship with temperature described in

Fig.1. Internal chamber temperatures and ambient temperatures during a one week deployment at the Brisbane site in summer.

K. Kennedy et al. / Environmental Pollution 158 (2010) 142–147146

using Equation (6) i.e. winter sampling rates together with the WTFfactor, averaged 3.8 � 1.4 m3 day�1 and ranged from 2.2 to6.8 m3 day�1 (Table 3).

Rearranging Equation (6), the actual partition coefficient valueunder the influence of in-situ deployment chamber temperaturesin summer KPUF-AIR (CHAMB T) can be derived (Equation (7)):

KPUF-AIRðCHAMB TÞ ¼RS SUMMER

ke SUMMERVS¼ RS WINTERWTF

ke SUMMERVS(7)

From the partition coefficient thus derived, the prevailingtemperature in the sampler chamber can be determined using therelationships described previously (refer Table 1 footnotes). Datainvolved in the partition coefficient calculation and predictedtemperatures in the sampler chamber are found in Table 3. Theseestimated deployment chamber temperatures in summer rangedfrom 8 to 23 �C higher than the average ambient temperature foreach site (Table 1) based on ambient 10–30 min data.

3.6. Field validation of sample chamber temperature estimates

In order to assess how realistic the estimated sample chambertemperatures might be, a short term field study was conducted atthe Brisbane site for a period of 7 days in the 2008/9 summer(10–18th December). Two replicate chambers were deployed atthis site, containing PUF samplers in the same configuration asthe earlier deployment. Temperature loggers were deployed tomeasure temperature in each chamber and ambient temperaturedata for this site was also obtained. Temperature profiles over thetime of deployment are shown in Fig. 1.

Since this deployment was retrospective and field based, it wasnot possible to replicate the exact environmental exposure condi-tions of the previous summer’s sampling period. There was frequentstorm activity during this latter period and a multi-storey buildinghad been erected on the western side of the site since the previoussummer meaning afternoon shading patterns were altered. Thesefactors notwithstanding, the data in Fig. 1 does show differencesbetween internal deployment chamber temperatures and ambienttemperatures, with chamber temperatures higher during the dayand often approaching the 40 �C estimated above for this site.

4. Conclusion

The development and application of the use of PRCs for in-situexposure assessment is desirable for accurate exposure assessments.In this study, sampling rates derived using ambient temperatures to

estimate KPUF-AIR were not adequately accounted for by knowninfluences of wind speed and temperature on mass transfer coeffi-cients. The influence of internal deployment chamber temperatureonpartition coefficients was investigated as an explanation for this. Aretrospective field study comparing internal deployment chamberwith ambient temperature measurements has shown that theseinternal temperatures may be higher and consistent with predictedtemperatures. These temperature increases may result in a consid-erable reduction in actual sampling rates. Using ambient tempera-tures, sampling rates at the temperate and sub-tropical sites in thiswork ranged from 7.1 to 10 m3 day�1. However, using an estimateof internal deployment chamber temperatures, the range was2.2–6.8 m3 day�1. Ambient air concentrations can therefore beunderestimated if ambient temperature is less than that of theinternal deployment chamber used to house the PUF passive airsamplers. It is therefore likely that more accurate PRC-based esti-mates of sampling rates would require that KPUF-AIR be estimatedusing internal chamber temperatures obtained from temperatureloggers deployed in-situ. Further seasonal work at a range of lati-tudes which is more consistent with recent guidelines for theapplication of PRCs (Moeckel et al., 2009) is needed to morecomprehensively assess the implications of these differentialtemperature profiles on PRC-based sampling rates. It is also likelythat this work would benefit from a validation of current relation-ships used to estimate KPUF-AIR (AMB T) for a range of temperatures.What is evident in this study is that our capacity to satisfy either theinitial 20–80% or the more recent >40% (Moeckel et al., 2009)recommendation for PRC elimination may be difficult for the samePRCs seasonally. An alternative approach which uses ‘‘curve fitting ofthe retained fraction of all PRCs as a function of PRC properties’’ (Booijand Smedes 2009) is currently being evaluated and may providealternative methods for the consistent application of PRCs in thefuture.

Acknowledgements

EnTox is a partnership between Queensland Health and theUniversity of Queensland. This work is funded by an ARC LinkageGrant (LP0560619). The authors acknowledge David Love(Queensland Environmental Protection Agency), Rob Mitchell,Chris Powell and Konrad Banasiak (South Australia EnvironmentProtection Authority), Jim Fitzgerald (South Australia Departmentof Health), Tina Runnion (Western Australia Department of Envi-ronment) for project facilitation and technical support and DavidGooding (AUSGIS) for mapping assistance.

K. Kennedy et al. / Environmental Pollution 158 (2010) 142–147 147

Appendix. Supplementary material

Supplementary data associated with this article can be found, inthe online version, at doi:10.1016/j.envpol.2009.07.031.

References

Bartkow, M.,Booij, K., Kennedy, K.E., Muller, J., Hawker, D., 2005. Passive sampling theoryfor atmospheric semivolatile organic compounds. Chemosphere 60, 170–176.

Bartkow, M., Kennedy, K.E., Huckins, J.N., Holling, N., Komarova, T., Muller, J.F., 2006.Photodegradation of polyaromatic hydrocarbons in passive air samplers: fieldtesting different deployment chambers. Environmental Pollution 144, 371–376.

Bartkow, M.E., Jones, K.C., Kennedy, K.E., Holling, N., Hawker, D., Mueller, J., 2004.Evaluation of performance reference compounds in PUF passive air samplers atdifferent wind speeds. Organohalogen Compounds 66, 139–144.

Booij, K., Sleiderink, H.M., Smedes, F., 1998. Calibrating the uptake kinetics ofsemipermeable membrane devices using exposure standards. EnvironmentalToxicology and Chemistry 17, 1236–1245.

Booij, K., Smedes, F., 2009. Uncertainties in PRC-based sampling rate estimation forhydrophobic samplers. In: Ocelka, T., Vrana, B., Booij, K., Koci, V., Strakos, K.(Eds.), 3rd International Passive Sampling Workshop and Symposium, May 27–May 30, 2009. Prague, The Czech Republic, pp. 50.

Chaemfa, C., Barber, J.L., Gocht, T., Harner, T., Holoubek, I., Klanova, J., Jones, K.C.,2008. Field calibration of polyurethane foam (PUF) disk passive air samplers forPCBs and OC pesticides. Environmental Pollution 156, 1290–1297.

Fuller, E.N., Schettler, P.D., Giddings, J.C., 1966. New method for prediction of binarygas-phase diffusion coefficients. Industrial and Engineering Chemistry 58, 18–27.

Gouin, T., Harner, T., Blanchard, P., Mackay, D., 2005a. Passive and active air samplersas complementary methods for investigating persistent organic pollutants in theGreat Lakes Basin. Environmental Science and Technology 39, 9115–9122.

Gouin, T., Harner, T., Daly, G.L., Wania, F., Mackay, D., Jones, K.C., 2005b. Variability ofconcentrations of polybrominated diphenyl ethers and polychlorinated biphe-nyls in air: implications for monitoring, modelling and control. AtmosphericEnvironment 39, 151–166.

Gouin, T., Wania, F., Ruepert, C., Castillo, L.E., 2008. Field testing passive air samplersfor current use pesticides in a tropical environment. Environmental Science andTechnology 42, 6625–6630.

Harju, M.T., Haglund, P., Naikwadi, K.P., 1998. Gas chromatographic properties of the209 PCB congeners on non-polar, chiral, and liquid-crystal columns. Organo-halogen Compounds 35, 111–113.

Harner, T., Bidleman, T.F., 1996. Measurements of octanol-air partition coefficients forpolychlorinated biphenyls. Journal of Chemical and Engineering Data 41, 895–899.

Harner, T., Shoeib, M., Diamond, M.L., Stern, G.A., Rosenberg, B., 2004. Using passiveair samplers to assess urban–rural trends for persistent organic pollutants. 1.Polychlorinated biphenyls and organochlorine pesticides. EnvironmentalScience and Technology 38, 4474–4483.

Harner, T., Pozo, K., Gouin, T., Macdonald, A.-M., Hung, H., Cainey, J., Peters, A.,2006a. Global pilot study for persistent organic pollutants (POPs) using PUFdisk passive air samplers. Environmental Pollution 144, 445–452.

Harner, T., Shoeib, M., Diamond, M.L., Ikonomou, M.G., Stern, G., 2006b. Passivesampler derived air concentrations of PBDEs along an urban–rural transect:spatial and temporal trends. Chemosphere 64, 262–267.

Huckins, J.N., Petty, J.D., Lebo, J.A., Almeida, F.V., Booij, K., Alvarez, D.A., Cranor, W.L.,Clarck, R.C., Mogensen, B.B., 2002. Development of the permeability/perfor-mance reference compound approach for in situ calibration of semipermeablemembrane devices. Environmental Science and Technology 36, 85–91.

Jaward, F.M., Farrar, N.J., Harner, T., Sweetman, A.J., Jones, K.C., 2004a. Passive airsampling of PCBs, PBDEs, and organochlorine pesticides across Europe. Envi-ronmental Science and Technology 38, 34–41.

Jaward, F.M., Farrar, N.J., Harner, T., Sweetman, A.J., Jones, K.C., 2004b. Passiveair sampling of polycyclic aromatic hydrocarbons and polychlorinatednapthalenes across Europe. Environmental Toxicology and Chemistry 23,1355–1364.

Jaward, F.M., Zhang, G., Nam, J.J., Sweetman, A.J., Obbard, J.P., Kobara, Y., Jones, K.C.,2005. Passive air sampling of polychlorinated biphenyls, organochlorinecompounds and polybrominated diphenyl ethers across Asia. EnvironmentalScience and Technology 39, 8638–8645.

Klanova, J., Kohoutek, J., Hamplova, L., Urbanova, P., Holoubek, I., 2006. Passive airsamplers as a tool for long-term air pollution monitoring: part 1. Performanceassessment for seasonal and spatial variations. Environmental Pollution 144,393–405.

Klanova, J., Eupr, P., Kohoutek, J., Harner, T., 2008. Assessing the influence ofmeteorological parameters on the performance of polyurethane foam-basedpassive air samplers. Environmental Science and Technology 42, 550–555.

Moeckel, C., Harner, T., Nizzetto, L., Strandberg, B., Lindroth, A., Jones, K.C., 2009. Useof depuration compounds in passive air samplers: results from active airsampling-supported field deployment, potential uses and recommendations.Environmental Science and Technology 43, 3227–3232.

Motelay-Massei, A., Harner, T., Shoeib, M., Diamond, M., Stern, G., Rosenberg, B.,2005. Using passive air samplers to assess urban–rural trends for persistentorganic pollutants and polycyclic aromatic hydrocarbons. 2. Seasonal trends forPAHs, PCBs and organochlorine pesticides. Environmental Science and Tech-nology 39, 5763–5773.

Pozo, K., Harner, T., Shoieb, M., Urrutia, R., Barra, R., Parra, O., Focardi, S., 2004.Passive-sampler derived air concentrations of persistent organic pollutants ona North–South transect in Chile. Environmental Science and Technology 38,6529–6537.

Pozo, K., Harner, T., Wania, F., Muir, D.C.G., Jones, K.C., Barrier, L.A., 2006. Towarda global network for persistent organic pollutants in air: results from the GAPsstudy. Environmental Science and Technology 40, 4867–4873.

Santiago, E.C., Cayetano, M.G., 2007. Polycyclic aromatic hydrocarbons in ambientair in the Philippines derived from passive sampler with polyurethane foamdisk. Atmospheric Environment 41, 4138–4147.

Shoeib, M., Harner, T., 2002. Characterisation and comparison of three passive airsamplers for persistent organic pollutants. Environmental Science andTechnology 36, 4142–4151.

Thompson, J., Kennedy, K., Hawker, D.W., Mueller, J.F., Bartkow, M.E., 2007.Comparison of polyurethane foam-based passive air samplers: single vs doubledisk. Organohalogen Compounds 69, 1034–1037.

Tuduri, L., Harner, T., Hung, H., 2006. Polyurethane foam (PUF) disks passive airsamplers: wind effect on sampling rates. Environmental Pollution 144,377–383.

Wilford, B.H., Harner, T., Zhu, J., Shoeib, M., Jones, K.C., 2004. Passive samplingsurvey of polybrominated diphenyl ether flame retardants in indoor andoutdoor air in Ottawa, Canada: implications for sources and exposure. Envi-ronmental Science and Technology 38, 5312–5318.