- Email: [email protected]
Effect of sampler material on the uptake of PAHs into passive sampling devices
Chemosphere 79 (2010) 470–475
Contents lists available at ScienceDirect
Chemosphere journal homepage: www.elsevier.com/locate/chemosphere
Effect of sampler material on the uptake of PAHs into passive sampling devices Ian J. Allan *, Christopher Harman, Alfhild Kringstad, Erling Bratsberg Norwegian Institute for Water Research (NIVA), Gaustadalléen 21, NO-0349 Oslo, Norway
a r t i c l e
i n f o
Article history: Received 3 October 2009 Received in revised form 13 January 2010 Accepted 14 January 2010
Keywords: Passive sampling Performance reference compounds Semipermeable membrane devices Low density polyethylene membrane Silicone strip
a b s t r a c t Increasing demand for simple and reliable passive samplers for monitoring hydrophobic organic contaminants in water has led to increased frequency of use of single-phase polymeric sampling devices. In this study, we evaluate the effect of sampler material on the passive sampling of polycyclic aromatic hydrocarbons (PAHs) in two Norwegian rivers. Low density polyethylene membranes (LDPE), silicone strips and semipermeable membrane devices (SPMDs) with the exact same surface area and conformation were exposed in the Drammen River for overlapping exposures of 24 and 51 d, under identical hydrodynamic conditions. Dissipation rates of performance reference compounds (PRCs) spiked in all samplers were consistent and demonstrated no significant differences in sampler–water analyte exchange kinetics between the two exposures. The transition to fully boundary layer-controlled uptake shown by PRC dissipation rates was confirmed by investigating PAH masses absorbed by the samplers. Masses of analytes with log Kow > 4.5 absorbed into the samplers were similar and independent of the sampler material used, generally indicating for these compounds that the boundary layer dominated the resistance to mass transfer. The very low variability in analyte masses absorbed across sampler types observed here indicates that much of the overall variability in dissolved contaminant concentrations seen in passive sampler intercomparison studies is likely the result of the uncertainty associated with sampler–water partition coefficients and PRC dissipation rates. PRC dissipation rates and ratios of masses absorbed over 51 and 24 d for these compounds demonstrated integrative sampling over 51 d and no major effects of biofouling on sampling. The equivalence of data obtained using silicone strips and SPMDs supports the use of single-phase polymeric passive sampling devices. Ó 2010 Elsevier Ltd. All rights reserved.
1. Introduction The selection of sampler design and choice of material are important factors that determine the performance of passive sampling devices for the measurement of hydrophobic organic contaminants dissolved in water (Rusina et al., 2007; Allan et al., 2009; Rusina, 2009). Samplers may be single-phase devices with combined membrane and receiving phases or more complex designs such as biphasic semipermeable membrane devices (SPMDs) comprising triolein lipids inside low density polyethylene (LDPE) tubing. Contaminant accumulation into passive samplers is a diffusive process driven by a gradient in chemical activity of the analyte of interest in water and the sampler initially free of analyte (Vrana et al., 2005). Therefore analyte diffusion coefficients in polymeric membranes and the sampler–water partition coefficient, Ksw affect sampler behaviour and analyte sampling rates (Rusina et al., 2007; Rusina, 2009; Smedes et al., 2009). For example, the diffusion coefficients of polycyclic aromatic hydrocarbons (PAH) have been found to be significantly lower in LDPE than in silicone material
* Corresponding author. Tel.: +47 22 18 51 00; fax: +47 22 18 52 00. E-mail address: [email protected] (I.J. Allan). 0045-6535/$ - see front matter Ó 2010 Elsevier Ltd. All rights reserved. doi:10.1016/j.chemosphere.2010.01.021
(Rusina, 2009). Such a difference is likely to result in a divergence between observed thresholds for the transition between membrane-influenced and boundary layer-controlled analyte uptake for samplers made of different materials (Booij et al., 1998, 2003; Allan et al., 2009; Rusina, 2009). Boundary layer-controlled uptake into silicone is likely to extend to compounds with lower octanol– water partitioning coefficients (Kow) than for LDPE (Rusina et al., 2007). The determination in situ of sampling rates for analytes of interest can be undertaken through the evaluation of dissipation kinetics of performance reference compounds (PRCs) spiked into the sampler prior to exposure (Booij et al., 1998; Huckins et al., 2002). When PRCs with an appropriate range of physicochemical properties are used, it is possible to establish which phases influence and control analyte exchange kinetics between samplers and water (Allan et al., 2009). The physical conformation of the samplers and deployment procedures (Booij et al., 2006) can have drastic effects on contaminant sampling rates since these can induce large differences in water turbulences and velocity at the surface of the samplers (Vrana and Schuurmann, 2002; Booij et al., 2006; Huckins et al., 2006). For example, samplers deployed outside exposure cages have shown increased sampling rates compared to those exposed within cages (Booij et al., 2006; Allan et al., 2009).
I.J. Allan et al. / Chemosphere 79 (2010) 470–475
Semipermeable membrane devices (SPMDs) have been successfully applied to the measurement of hydrophobic contaminant concentrations in water for almost two decades (Huckins et al., 1993, 2006). However, production of these samplers is a complex process and the presence of triolein and other impurities in SPMD extracts means a more comprehensive clean-up procedure is required. Single-phase polymeric devices made of LDPE or silicone may be used as a simpler alternative to SPMDs (Booij et al., 2002, 2003; Rusina et al., 2007; Anderson et al., 2008; Rusina, 2009) since more accurate values for parameters such as Ksw are available in the literature. Studies aiming to assess or demonstrate the equivalence of data obtained with different devices are scarce (Allan et al., 2009) and it is often difficult to control the many parameters involved during such comparative studies particularly when these are based on field exposure of the samplers. The present study aimed to evaluate the performance of LDPE membranes, silicone strips and SPMDs applied to the monitoring of PAHs in two Norwegian rivers. Effects of sampler material on the passive sampling of PAHs were evaluated by minimising the impact of sampler configuration. Possible impacts of sampler material used were gauged through the comparison of PRC dissipation rates and PAH masses absorbed by the different devices during two overlapping deployments in the Drammen River. In an attempt to control some of the many parameters that can influence measurements undertaken with different sampling materials, samplers used were prepared with exactly the same dimensions and exposed under identical hydrodynamic conditions. A 28 d exposure in another river, the Alna, was also undertaken, but with the use of two different types of deployment cages. 2. Methods 2.1. Reagents and glassware Polycyclic aromatic hydrocarbons and their deuterated analogues were obtained from Chiron (Trondheim, Norway) and were of analytical standard. Purities were >99% for PAHs and >99.5% for deuterated PAHs. Solvents were from Rathburn (Walkerburn, Scotland) except for cyclohexane (J.T. Baker, Deventer, Holland) and were of HPLC grade or better. Glassware was solvent rinsed and baked in a muffle furnace at 540 °C before use. Ultra pure water (Option 3, Elga™) was used for diluting solutions and rinsing equipment and for spiking PRC into silicone strips and LDPE membranes. 2.2. Passive sampler preparation LDPE membranes and silicone strips were prepared following standard SPMD dimensions and were spiked with a series of PRCs. SPMDs (nominal length of 91.4 cm between the inner LDPE welds and width of 2.5 cm) were purchased from ExposMeter AB (Tavelsjö, Sweden) and contained the following compounds as PRCs; d10-acenaphthene, d10-fluorene, d10-phenanthrene, d12-chrysene and d12-benzo[e]pyrene. Lay-flat LDPE tubing was purchased from Brentwood Plastics Inc. (St. Louis, USA), and is similar to that used for SPMDs. The tubing was reproducibly cut along the two edges resulting in 2.5 cm wide LDPE membranes (average membrane thickness of 80 lm) and mounting loops were made at either end using a heat-sealer. Once rinsed with tap and ultra pure water samplers were air dried before being soxhlet extracted at low temperature with 50:50 methanol/hexane for 8 h. Samplers were left to dry before spiking with d10-acenaphthene, d10-fluorene, d10phenanthrene, d10-fluoranthene and d12-chrysene, as PRCs. The spiking procedure involved placing each sampler in a 50:50 methanol/water solution fortified with 5 lg of the deuterated PAHs in
471
50 mL glass tubes (Booij et al., 2003). This solution was then shaken in the dark at 100 rpm for 72 h, though equilibrium is likely to be reached much earlier (Booij et al., 2002). Samplers were then rapidly dried with a clean tissue to remove solution from the surface and left to dry for 1 min before being stored in clean jars at 20 °C until deployment. AlteSil™ silicone sheets which averaged 570 lm in thickness were purchased from Altec Ltd. (Bude, UK). 2.5 cm wide strips were cut and the length was adjusted to that of SPMDs (as were LDPE membranes). These samplers were processed in a similar way to LDPE membranes, with soxhlet extraction using a mixture of 50:50 methanol/pentane. Spiking with 4 lg of deuterated PAHs was undertaken using the same procedure as for LDPE membranes. In short, each silicone strip was added to a 150 mL (50:50 water/methanol) and left on an orbital shaker for 72 h at 100 rpm. Once ready, samplers were stored at 20 °C. Purpose-made clips to allow fastening of silicone to deployment equipment, were soxhlet cleaned prior to use. The relative standard deviations of measurements of PRCs in non-exposed spiked samplers, were mostly well below 10% (n = 8) for LDPE membranes and silicone strips. 2.3. Extraction and analysis All controls, trip controls and exposed passive samplers were cleaned thoroughly by washing with ultra pure water and wiped with a clean tissue. SPMDs were placed in a pre-cleaned glass jar and dialysed with 100 mL hexane. The procedure was repeated after 24 h (Harman et al., 2008a, b). Hexane extracts were reduced by a stream of nitrogen to 1 mL. As a clean-up step, extracts were partitioned with acetonitrile (HPLC grade) and the acetonitrile portion quantitatively removed and reduced prior to analysis. Extraction of LDPE membranes and silicone strips was similar to that given above for SPMDs, except the solvent used was methanol instead of hexane. Extracts were analysed on a HP-6890 Plus gas chromatograph equipped with a HP 5973 mass selective detector, operated in single ion monitoring mode (SIM) with electron impact ionisation (70 eV). Analytes were separated on a 30 m DB-5 column (0.25 mm i.d. and 0.25 lm film thickness, Agilent JW Scientific, Santa Clara, USA) and with a helium flow of 1 mL min1. The injection was splitless and the injection volume was 1 lL. The GC oven temperature was held for 2 min at 60 °C before increasing to 250 °C at a rate of 7 °C min1. The final step was an increase to 310 °C at a rate of 15 °C min1 (held for 6 min). Injector, transfer line, ion source and quadruple temperatures were set to 300, 280, 230 and 150 °C, respectively. Quantification of individual compounds was performed by using the relative response of surrogate internal standards. 2.4. Exposure in the Drammen River The present study was mainly conducted on the Drammen River (Norway) at a site (59°450 1300 N; 10°00 2200 E) approximately 8 km upstream of where the river flows into the fjord. This ensured relatively well-mixed river water and unidirectional flow. The water temperature during the exposure gradually decreased from 10 °C to 5 °C from September to November 2008. The level of suspended particulate matter (SPM) was low (1 mg L1). Total organic carbon (TOC) indicated that much of the organic carbon in the water was present as dissolved organic carbon rather than particulate. TOC was measured on two occasions with values between 3 and 4 mg L1. The water flow was approximately 300 m3 s1. Five replicates of each type of passive sampling device were exposed in the river for periods of 24 and 51 d (with the exception of LDPE at 51 d). The two exposures were overlapping with the 24 d exposure initiated 2 weeks after the start of the 51 d deployment.
472
I.J. Allan et al. / Chemosphere 79 (2010) 470–475
Five replicate control samplers were prepared and stored at 20 °C in the freezer to check for contamination during construction and extraction procedures. A further three samplers were used as trip controls, brought to the deployment site and opened to the air to check for contamination by target analytes during deployment and retrieval operations. Since all types of passive samplers were purposely prepared with the same surface area and configuration, a further requirement here was to deploy these samplers under similar water turbulences. This was achieved by deploying samplers in commercially available SPMD cages (Environmental Sampling Technologies, St. Joseph, USA), which were attached to specifically installed moorings. Samplers were mounted onto spider holders inside the deployment cages (five per cage), distributed across three cages for the 24 d exposure and two cages for the 51 d exposure. All cages were held vertical and perpendicular to the river water flow. 2.5. Exposure in the Alna River All types of samplers were also deployed for 28 d in the Alna River, a relatively small watercourse flowing through Oslo (59°540 1600 N; 10°470 3100 ). The temperature of the water during sampler exposure in December 2008 decreased from 5 °C to just above freezing with ice forming in river in the last week of exposure. The water flow was 1.5 m3 s1. The SPM content of 23 mg L1 on average was much higher than that measured in the Drammen River. TOC was on average 6 mg L1. Triplicate SPMDs were deployed in similar cages to those used in the Drammen River, while LDPE membranes and silicone strips were attached to flat cages made of stainless steel mesh with 1 in. wide openings. Since the water depth was <1 m, flat cages were deployed parallel to the water flow. 3. Results and discussion 3.1. Performance reference compound dissipation rates for exposures in the Drammen River Dissipation of PRCs with log Kow values in the range 3.5–6.0 for all three types of samplers and both exposure periods was observed and allowed the estimation of sampling rates, Rs:
NPRC ¼ N0;PRC eke t
ð1Þ
where N0,PRC and NPRC are PRC masses in the samplers prior to and following exposure, respectively, and where the dissipation rate ke is given by:
ke ¼
kO A RS ¼ K SW V K SW V
ð2Þ
with kO is the overall mass transfer coefficient, A the surface area of the sampler (cm2), V the volume of the sampler (cm3) and RS the analyte sampling rate (L d1). Ksw values used here were from Booij et al. (2003) and Smedes et al. (2009) for LDPE membranes and silicone strips, respectively, and from Huckins et al. (2006) for SPMDs. As shown by Eq. (2), the dissipation rate, ke is proportional to A/V and this needs accounting for when comparing PRC data for samplers with differing configurations. Here, while the surface area of the samplers was kept constant, sampler volumes were dissimilar as a result of different thicknesses. Nominal volumes of LDPE membranes, silicone strips and SPMDs were 1.95, 13.05 and 4.95 cm3, respectively. Dissipation rates were normalised to the A/V ratio, log-transformed and are shown in Fig. 1 as a function of log Kow. Log Kow values used here and throughout this study were those given by Smedes et al. (2009). As expected, log keV/A values generally
appear to decrease with increasing log Kow values for all samplers. Despite dissipation for PRCs with log Kow 6.0 for LDPE membranes and SPMDs representing just over 10% of starting concentrations, these were significant (one-sided t-test at 95%). Fig. 1A shows a clear overlap of dissipation rates for PRCs with log Kow 4.0–5.5 for all samplers for the 24 d exposure. The spread of the data is mostly 0.2–0.3 of a log unit and is below values observed during the comparison of a wider range of passive sampling devices with analysis conducted in three different laboratories (Allan et al., 2009). Silicone and SPMD data for the 51 d exposure are consistent and the within-sampler variability appears extremely low (Fig. 1B). Most log keV/A values (cm h1) reported by Allan et al. (2009) for the Chemcatcher passive sampling device, LDPE, silicone strips and SPMDs are in the range 3.6 to 4.6, with values between 5 and 6 for PRCs with log Kow of 5.8 for LDPE membrane samplers. Values observed in the present study are lower, but of the same order as those reported by Rusina (2009) in a calibration experiment with AlteSil™ silicone membrane. Eq. (2) and ke values could be used to estimate sampling rates for PAH for each sampler type and exposure period. Generally these sampling rates are at the lower end of the range observed with samplers of this size (Augulyte and Bergqvist, 2007; Ellis et al., 2008; Gourlay-France et al., 2008). Despite dissipation >90% for some PRCs (after 51 d exposure), little differences in Rs estimates were observed between those from the 24 and 51 d exposures, indicating that these may remain reliable despite significant offload. Under membrane-influenced exchange kinetics, a plateau in ke values may be expected. However, when a major portion of the mass transfer resistance resides in transport across the water boundary layer, ke values may be expected to decrease with increasing log Ksw or Kow (Booij et al., 2003; Huckins et al., 2006; Rusina, 2009). A clear plateau can be seen on Fig. 1A for the 24 d exposure of LDPE membrane samplers for PRCs with log Kow < 4.2. A similar plateau is also shown for SPMDs for both exposures (Fig. 1A and B) for PRCs with log Kow < 4.2. Silicone strips do not appear to follow this trend and no obvious plateau can be observed. This is likely caused by analyte diffusion coefficients being 2–3 orders of magnitude higher in silicone than in LDPE (Rusina, 2009), significantly reducing resistance to mass transfer in the silicone. This in turn extends boundary layer-controlled conditions to lower log Ksw/Kow values. 3.2. Boundary layer-controlled PRC dissipation for exposures in the Drammen River Linear regressions of log ke (d1) vs. log Kow were undertaken for all PRCs assumed to be under boundary-layer-controlled conditions. When using all significant PRC dissipation rates, linear relationships in the range 0.69 to 1.0 could be observed for all samplers (data not shown) for both deployment periods and these were supported by low values of the standard error of the slope (se in the range 0.02–0.07) and R2 values close to 1 (data not shown). Only PRCs with dissipation between 20% and 80% over the 24 or 51 d exposures were then used to determine slopes shown in Table 1 since a lower uncertainty in PRC offload may be expected for this range than when using values close to 0% or 100%. Resulting slopes were very close to 1.0 for all samplers with se values in the range 0.03–0.17. Despite minor differences, slopes were similar, providing evidence that mass transfer resistance in the boundary layer generally dominates overall mass transfer for analytes with log Kow > 4.0. Slopes obtained in our study were steeper than those of 0.42 obtained by Booij et al. (1998) but generally less steep than those of 1.039 and 1.171 obtained by other workers (Rusina, 2009) for the uptake of PAHs and PCBs into silicone and the offload of deuterated PRCs during two calibration experiments performed in the laboratory. Slopes were closer to those obtained
473
I.J. Allan et al. / Chemosphere 79 (2010) 470–475
-4.0
-4.0
(B) 51 day exposure
-4.5
-4.5
-5.0
-5.0
-5.5
-5.5
-1
Log keV/A (cm h )
(A) 24 day exposure
-6.0
-6.0
LDPE membrane Silicone strip SPMD
-6.5 3.5
4.0
4.5
5.0
5.5
6.0
-6.5 3.5
4.0
4.5
5.0
5.5
6.0
6.5
Fig. 1. PRC dissipation rates normalised to sampler surface area to volume ratio (log keV/A in cm h1) for 24 (A) and 51 (B) day exposures of LDPE membranes, silicone strips and SPMDs in the Drammen River.
Table 1 Results of linear regressions log ke–log Kow for PRCs under boundary layer-controlled exchange for passive sampler exposures of 24 and 51 d in the Drammen River. se
R
n
Exposure (d)
0.03 0.17 0.16
0.99 0.85 0.83
15 20 10
24 24, 51 24
log ke = (slope) log Kow + b. Standard error of the slope a. Number of data points where dissipation was in the range 20–80%.
by Rusina (2009) when only using PRC with dissipation between 20% and 80%. However, our field-derived data is based, at best, on two timepoints of 24 or 51 d compared with calibration data by Rusina (2009) which is based on a time series. According to Eq. (2), the product of log ke and log Ksw indicates how Rs varies with increasing analyte hydrophobicity. With log Ksw–log Kow slopes close to unity (with a standard error, se = 0.2 of a log-unit) for silicone material (Smedes et al., 2009), sampling rates are not expected to decrease much with increasing PAH hydrophobicity. However applying the log Ksw–log Kow slope of 1.48 (se = 0.30 logunit) obtained for LDPE by Smedes et al. (2009) results in increasing sampling rates with increasing hydrophobicity. When using other published log Ksw–log Kow relationships for the same polymer material (Booij et al., 2003), this increase in Rs with increasing log Kow is not observed. Such an artefact may be the result of the uncertainties in Ksw values and ke values or of some limitations with the use of Eq. (2) to estimate sampling rates for analytes under boundary layer-controlled uptake. The uncertainty in Ksw and ke values may therefore account for a large proportion of the variability in calculated dissolved contaminant concentrations for a range of passive sampling devices (Allan et al., 2009). 3.3. PAH masses absorbed during exposures in the Drammen River The 24 d sampler exposure in the Drammen River resulted in the detection of PAHs with molecular weights no greater than that of chrysene (as shown on Fig. 2A) and benzo[e]pyrene for the 51 d deployment for both silicone strips and SPMDs (see Fig. 2B). For analytes in linear uptake, the mass absorbed into the sampler is dependent on the sampling rate and the dissolved analyte concentration in water. Boundary layer-controlled uptake for silicone strips was shown to extend to analytes with log Kow of 4.0, indi-
50 -1
–1.092 –1.004 –1.03
c
ng sampler
Slope
LDPE membrane Silicone strip SPMD
b
LDPE membrane Silicone strip SPMD
(A) 24 day exposure
40 30 20 10 0 PYR
FLUO
CHRY
140 (B) 51 day exposure
120 100 -1
c
Sampler
b
ng sampler
a b
a
60
80 60 40 20 0 PYR
FLUO
CHRY
BbjF
BeP
Fig. 2. Masses of pyrene (PYR), fluoranthene (FLUO), chrysene (CHRY), benzo[b and j]fluoranthene (BbjF) and benzo[e]pyrene (BeP) absorbed in the different passive sampling devices following 24 and 51 d exposures in the Drammen River. Note: LDPE membranes were not exposed for 51 d.
cating that mass transfer coefficients for these exposures are relatively low. When uptake is linear and boundary layer-controlled, masses of analytes absorbed by the samplers should be independent of the sampler material (as long as they have sufficient capacity). Masses of pyrene, fluoranthene and chrysene accumulated in silicone strips and SPMDs over 24 d appear consistent with such an assumption (Fig. 2). The mass of pyrene absorbed in LDPE mem-
I.J. Allan et al. / Chemosphere 79 (2010) 470–475
3.4. Exposure in the Alna River Despite the low temperature of the Alna River water, 28 d exposure was sufficient to observe significant dissipation of PRCs up to d12-chrysene for all samplers. Dissipation of the least hydrophobic PRCs was almost complete, though still quantifiable. Log keV/A values for all samplers plotted on Fig. 3A show a clear decrease for PRCs with log Kow > 4.5 while for less hydrophobic PRCs, normalised ke values appear to plateau. Although this could be caused by an increase in the uncertainty in estimates of dissipation rates when offload is almost complete, values calculated for the exposure of LDPE membranes and SPMDs in the Alna River are similar to those found for the Drammen River. While overlapping of PRC data for LDPE membranes and SPMDs may be observed on Fig. 3A, PRC dissipation rates for silicone strips are generally higher and the plateau is not as obvious. Slopes of log ke–log Kow regressions for PRCs with log Kow > 5 were 1.12, 0.76 and 0.72 for LDPE membranes, silicone strips and SPMDs, respectively and are generally close to those obtained in the Drammen River deployment. Sampler–water exchange kinetics appear higher for the exposure in the Alna compared with those from the Drammen River. However, no clear impact of using different cage designs could be observed during this Alna River deployment. Rs values across
-3
(A)
-4
Log keV/A
brane samplers did not differ from those found in the two other samplers while that of fluoranthene was 80% (Fig. 2A). Chrysene in LDPE extracts was below detection limits. The 51 d exposure resulted in the measurement of extremely similar masses of analytes from pyrene up to benzo[e]pyrene for silicone strips and SPMDs (Fig. 2B). This shows the integrative nature of the uptake over 51 d for analytes with log Kow > 5–5.2 when using these two types of samplers. In most cases, relative standard deviations were well <10% for all three types of samplers at each time interval (n = 5). Considering the very low between sampler type variability in absorbed masses shown in Fig. 2 for analytes under boundary layer-controlled uptake, it is likely that a substantial increase in the variability of calculated dissolved contaminant concentrations will be caused by variability associated with Ksw and ke values. In order to confirm the transition between analytes under integrative uptake and those close to equilibrium, ratios of masses absorbed during 51 d (N51) over those accumulated during the 24 d exposure (N24) normalised to respective deployment times (N51/ 51:N24/24) were calculated for analytes with log Kow in the range 3.9–6.0. Ratios of 1 are expected: (i) if sampling is integrative, (ii) if no changes in water turbulences near the samplers occur during the two exposures, (iii) if there are no significant influences from biofilm growth at the surface of the samplers and (iv) if the dissolved PAH concentration in water does not vary significantly during the 51 d period. A ratio of 24/51 (i.e. 0.47) should be observed if equilibrium is reached within 24 d and if no further changes in analyte concentration in the sampler are observed. Ratios close to or slightly higher than 1 were observed for analytes with log Kow > 4.5 for silicone strips and SPMDs, respectively. Ratios for anthracene and phenanthrene were very close to 1 for silicone strips while values 0.75 were found for SPMDs. This confirms that uptake into silicone strips used in this study is likely to remain linear for a longer period of time than for SPMDs as a result of their larger volume. Some PAHs with log Kow < 4.5 were found to be close to equilibrium within the 24 d exposure with ratios generally between 0.45 and 0.8. Ratios of 1 indicate that biofouling (visually observed algal growth at the surface of the samplers) appeared to have had minimal effects on the uptake of analytes with log Kow > 5. In contrast with recent work by Harman et al. (2009), the biofouling here did not appear to be affected by the type of material the sampler is made from (Booij et al. (2006)).
-5
-6
LDPE membrane Silicone strip SPMD
-7 3.5
4.0
4.5
5.0
5.5
6.0
Log KOW (B)
msampler /msilicone strip
474
1
LDPE membrane Silicone strip SPMD
0.1
3.5
4.0
4.5
5.0
5.5
6.0
6.5
7.0
Log KOW Fig. 3. PRC dissipation rates normalised to sampler surface area (A) to volume (V) ratio (keV/A in cm h1) vs. log Kow for 28 d exposure of LDPE membranes, silicone strips and SPMDs in the Alna River (filled symbols) (A) and PAH masses absorbed by the various samplers normalised to those accumulated in silicone strip samplers (B). White symbols on (A) represent data for the 24 d exposure in the Drammen River.
the array of samplers and analytes were in the range 2.7– 32 L d1 with highest values found for silicone strips. Concentrations of PAHs were higher in samplers exposed in the Alna compared with exposure in the Drammen River and this is the result of higher sampling rates and also likely higher water concentrations. Similarly to the Drammen River exposure, the variability between replicate samplers from the Alna River was low (Fig. 3B). PAH masses absorbed by the various replicate samplers were normalised to the mean mass absorbed in silicone strips. The time required for the analytes with log Kow < 5.2 to reach half of equilibrium concentration was lower than the exposure time and thus sampling was not linear over 28 d. Masses absorbed into LDPE membranes and SPMDs were lower than those found in silicone strips as a result of the lower volumes of these samplers. Less significant differences were found for analytes with log Kow > 5.8 when uptake was expected to be linear and boundary-controlled (Fig. 1B). Despite the high water velocities encountered (as apparent from the PRC data), the use of two different cage designs may have had a limited effect on sampling rates, possibly as a result of
I.J. Allan et al. / Chemosphere 79 (2010) 470–475
ice forming in the water towards the end of the exposure and affecting both cage designs differently.
4. Conclusions The present study investigated the effects of sampler material on the uptake of PAHs and the release of PRCs from samplers with the same surface area and under similar water turbulences deployed in the Drammen River. Consistent PRC dissipation data was obtained with LDPE membranes, silicone strips and SPMDs for 24 and 51 d exposures. The threshold for boundary layer-controlled exchange kinetics observed through PRC release from the samplers was confirmed by extremely similar masses of analytes with log Kow > 4.5 absorbed by silicone strips and SPMD over 24 and 51 d. Differences in masses of analytes under boundary layer-controlled uptake absorbed in samplers made of different materials are much lower than differences observed between PRC-based Rs estimates for these samplers. This implies that much of the variability in estimates of dissolved contaminant concentrations seen in passive sampler intercomparison studies is the result of the uncertainty associated with Ksw and ke values and the mode of calculation of sampling rates. The consistency of data obtained with samplers tested here combined with the low variability observed supports the use of simpler single-phase samplers such as LDPE membranes and silicone strips. Acknowledgements We acknowledge financial support from the Norwegian Climate and Pollution Agency (Klif) under Contract 5009 and from NIVA’s internal funding process (E-project GLIMPS, O-28074). Views presented here are those of the authors alone. References Allan, I.J., Booij, K., Paschke, A., Vrana, B., Greenwood, R., Mills, G.A., 2009. Field performance of seven passive sampling devices for monitoring of hydrophobic substances. Environ. Sci. Technol. 43, 5383–5390. Anderson, K.A., Sethajintanin, D., Sower, G., Quarles, L., 2008. Field trial and modeling of uptake rates of in situ lipid-free polyethylene membrane passive sampler. Environ. Sci. Technol. 42, 4486–4493. Augulyte, L., Bergqvist, P.A., 2007. Estimation of water sampling rates and concentrations of PAHs in a municipal sewage treatment plant using SPMDs with performance reference compounds. Environ. Sci. Technol. 41, 5044–5049.
475
Booij, K., Sleiderink, H.M., Smedes, F., 1998. Calibrating the uptake kinetics of semipermeable membrane devices using exposure standards. Environ. Toxicol. Chem. 17, 1236–1245. Booij, K., Smedes, F., van Weerlee, E.M., 2002. Spiking of performance reference compounds in low density polyethylene and silicone passive water samplers. Chemosphere 46, 1157–1161. Booij, K., Hofmans, H.E., Fischer, C.V., Van Weerlee, E.M., 2003. Temperaturedependent uptake rates of nonpolar organic compounds by semipermeable membrane devices and low-density polyethylene membranes. Environ. Sci. Technol. 37, 361–366. Booij, K., van Bommel, R., Mets, A., Dekker, R., 2006. Little effect of excessive biofouling on the uptake of organic contaminants by semipermeable membrane devices. Chemosphere 65, 2485–2492. Ellis, S.G., Booij, K., Kaputa, M., 2008. Comparison of semipermeable membrane device (SPMD) and large-volume solid-phase extraction techniques to measure water concentrations of 4, 40 -DDT, 4, 40 -DDE, and 4, 40 -DDD in Lake Chelan, Washington. Chemosphere 72, 1112–1117. Gourlay-France, C., Lorgeoux, C., Tusseau-Vuillemin, M.H., 2008. Polycyclic aromatic hydrocarbon sampling in wastewaters using semipermeable membrane devices: accuracy of time-weighted average concentration estimations of truly dissolved compounds. Chemosphere 73, 1194–1200. Harman, C., Boyum, O., Tollefsen, K.E., Thomas, K., Grung, M., 2008a. Uptake of some selected aquatic pollutants in semipermeable membrane devices (SPMDs) and the polar organic chemical integrative sampler (POCIS). J. Environ. Monit. 10, 239–247. Harman, C., Tollefsen, K.E., Boyum, O., Thomas, K., Grung, M., 2008b. Uptake rates of alkylphenols, PAHs and carbazoles in semipermeable membrane devices (SPMDs) and polar organic chemical integrative samplers (POCIS). Chemosphere 72, 1510–1516. Harman, C., Bøyum, O., Thomas, K.V., Grung, M., 2009. Small but different effect of fouling on the uptake rates of semipermeable membrane devices and polar organic chemical integrative samplers. Environ. Toxicol. Chem. 28, 2324–2332. Huckins, J.N., Manuweera, G.K., Petty, J.D., Mackay, D., Lebo, J.A., 1993. Lipidcontaining semipermeable membrane devices for monitoring organic contaminants in water. Environ. Sci. Technol. 27, 2489–2496. Huckins, J.N., Petty, J.D., Lebo, J.A., Almeida, F.V., Booij, K., Alvarez, D.A., Clark, R.C., Mogensen, B.B., 2002. Development of the permeability/performance reference compound approach for in situ calibration of semipermeable membrane devices. Environ. Sci. Technol. 36, 85–91. Huckins, J.N., Petty, J.D., Booij, K., 2006. Monitors of Organic Chemicals in the Environment: Semipermeable Membrane Devices. Springer, New York. Rusina, T.P., 2009. New methods of sampling and sample pre-concentration of hydrophobic compounds in aquatic systems: magnifying passive sampling aspects. RECETOX – Masaryk University, Brno. p. 100. Rusina, T.P., Smedes, F., Klanova, J., Booij, K., Holoubek, I., 2007. Polymer selection for passive sampling: a comparison of critical properties. Chemosphere 68, 1344–1351. Smedes, F., Geertsma, R., van der Zande, T., Booij, K., 2009. Polymerwater partition coefficients of hydrophobic compounds for passive sampling: application of cosolvent models for validation. Environ. Sci. Technol. 43, 7047–7054. Vrana, B., Schuurmann, G., 2002. Calibrating the uptake kinetics of semipermeable membrane devices in water: impact of hydrodynamics. Environ. Sci. Technol. 36, 290–296. Vrana, B., Mills, G.A., Allan, I.J., Dominiak, E., Svensson, K., Knutsson, J., Morrison, G., Greenwood, R., 2005. Passive sampling techniques for monitoring pollutants in water. TrAC. Trends Anal. Chem. 24, 845–868.
Contents lists available at ScienceDirect
Chemosphere journal homepage: www.elsevier.com/locate/chemosphere
Effect of sampler material on the uptake of PAHs into passive sampling devices Ian J. Allan *, Christopher Harman, Alfhild Kringstad, Erling Bratsberg Norwegian Institute for Water Research (NIVA), Gaustadalléen 21, NO-0349 Oslo, Norway
a r t i c l e
i n f o
Article history: Received 3 October 2009 Received in revised form 13 January 2010 Accepted 14 January 2010
Keywords: Passive sampling Performance reference compounds Semipermeable membrane devices Low density polyethylene membrane Silicone strip
a b s t r a c t Increasing demand for simple and reliable passive samplers for monitoring hydrophobic organic contaminants in water has led to increased frequency of use of single-phase polymeric sampling devices. In this study, we evaluate the effect of sampler material on the passive sampling of polycyclic aromatic hydrocarbons (PAHs) in two Norwegian rivers. Low density polyethylene membranes (LDPE), silicone strips and semipermeable membrane devices (SPMDs) with the exact same surface area and conformation were exposed in the Drammen River for overlapping exposures of 24 and 51 d, under identical hydrodynamic conditions. Dissipation rates of performance reference compounds (PRCs) spiked in all samplers were consistent and demonstrated no significant differences in sampler–water analyte exchange kinetics between the two exposures. The transition to fully boundary layer-controlled uptake shown by PRC dissipation rates was confirmed by investigating PAH masses absorbed by the samplers. Masses of analytes with log Kow > 4.5 absorbed into the samplers were similar and independent of the sampler material used, generally indicating for these compounds that the boundary layer dominated the resistance to mass transfer. The very low variability in analyte masses absorbed across sampler types observed here indicates that much of the overall variability in dissolved contaminant concentrations seen in passive sampler intercomparison studies is likely the result of the uncertainty associated with sampler–water partition coefficients and PRC dissipation rates. PRC dissipation rates and ratios of masses absorbed over 51 and 24 d for these compounds demonstrated integrative sampling over 51 d and no major effects of biofouling on sampling. The equivalence of data obtained using silicone strips and SPMDs supports the use of single-phase polymeric passive sampling devices. Ó 2010 Elsevier Ltd. All rights reserved.
1. Introduction The selection of sampler design and choice of material are important factors that determine the performance of passive sampling devices for the measurement of hydrophobic organic contaminants dissolved in water (Rusina et al., 2007; Allan et al., 2009; Rusina, 2009). Samplers may be single-phase devices with combined membrane and receiving phases or more complex designs such as biphasic semipermeable membrane devices (SPMDs) comprising triolein lipids inside low density polyethylene (LDPE) tubing. Contaminant accumulation into passive samplers is a diffusive process driven by a gradient in chemical activity of the analyte of interest in water and the sampler initially free of analyte (Vrana et al., 2005). Therefore analyte diffusion coefficients in polymeric membranes and the sampler–water partition coefficient, Ksw affect sampler behaviour and analyte sampling rates (Rusina et al., 2007; Rusina, 2009; Smedes et al., 2009). For example, the diffusion coefficients of polycyclic aromatic hydrocarbons (PAH) have been found to be significantly lower in LDPE than in silicone material
* Corresponding author. Tel.: +47 22 18 51 00; fax: +47 22 18 52 00. E-mail address: [email protected] (I.J. Allan). 0045-6535/$ - see front matter Ó 2010 Elsevier Ltd. All rights reserved. doi:10.1016/j.chemosphere.2010.01.021
(Rusina, 2009). Such a difference is likely to result in a divergence between observed thresholds for the transition between membrane-influenced and boundary layer-controlled analyte uptake for samplers made of different materials (Booij et al., 1998, 2003; Allan et al., 2009; Rusina, 2009). Boundary layer-controlled uptake into silicone is likely to extend to compounds with lower octanol– water partitioning coefficients (Kow) than for LDPE (Rusina et al., 2007). The determination in situ of sampling rates for analytes of interest can be undertaken through the evaluation of dissipation kinetics of performance reference compounds (PRCs) spiked into the sampler prior to exposure (Booij et al., 1998; Huckins et al., 2002). When PRCs with an appropriate range of physicochemical properties are used, it is possible to establish which phases influence and control analyte exchange kinetics between samplers and water (Allan et al., 2009). The physical conformation of the samplers and deployment procedures (Booij et al., 2006) can have drastic effects on contaminant sampling rates since these can induce large differences in water turbulences and velocity at the surface of the samplers (Vrana and Schuurmann, 2002; Booij et al., 2006; Huckins et al., 2006). For example, samplers deployed outside exposure cages have shown increased sampling rates compared to those exposed within cages (Booij et al., 2006; Allan et al., 2009).
I.J. Allan et al. / Chemosphere 79 (2010) 470–475
Semipermeable membrane devices (SPMDs) have been successfully applied to the measurement of hydrophobic contaminant concentrations in water for almost two decades (Huckins et al., 1993, 2006). However, production of these samplers is a complex process and the presence of triolein and other impurities in SPMD extracts means a more comprehensive clean-up procedure is required. Single-phase polymeric devices made of LDPE or silicone may be used as a simpler alternative to SPMDs (Booij et al., 2002, 2003; Rusina et al., 2007; Anderson et al., 2008; Rusina, 2009) since more accurate values for parameters such as Ksw are available in the literature. Studies aiming to assess or demonstrate the equivalence of data obtained with different devices are scarce (Allan et al., 2009) and it is often difficult to control the many parameters involved during such comparative studies particularly when these are based on field exposure of the samplers. The present study aimed to evaluate the performance of LDPE membranes, silicone strips and SPMDs applied to the monitoring of PAHs in two Norwegian rivers. Effects of sampler material on the passive sampling of PAHs were evaluated by minimising the impact of sampler configuration. Possible impacts of sampler material used were gauged through the comparison of PRC dissipation rates and PAH masses absorbed by the different devices during two overlapping deployments in the Drammen River. In an attempt to control some of the many parameters that can influence measurements undertaken with different sampling materials, samplers used were prepared with exactly the same dimensions and exposed under identical hydrodynamic conditions. A 28 d exposure in another river, the Alna, was also undertaken, but with the use of two different types of deployment cages. 2. Methods 2.1. Reagents and glassware Polycyclic aromatic hydrocarbons and their deuterated analogues were obtained from Chiron (Trondheim, Norway) and were of analytical standard. Purities were >99% for PAHs and >99.5% for deuterated PAHs. Solvents were from Rathburn (Walkerburn, Scotland) except for cyclohexane (J.T. Baker, Deventer, Holland) and were of HPLC grade or better. Glassware was solvent rinsed and baked in a muffle furnace at 540 °C before use. Ultra pure water (Option 3, Elga™) was used for diluting solutions and rinsing equipment and for spiking PRC into silicone strips and LDPE membranes. 2.2. Passive sampler preparation LDPE membranes and silicone strips were prepared following standard SPMD dimensions and were spiked with a series of PRCs. SPMDs (nominal length of 91.4 cm between the inner LDPE welds and width of 2.5 cm) were purchased from ExposMeter AB (Tavelsjö, Sweden) and contained the following compounds as PRCs; d10-acenaphthene, d10-fluorene, d10-phenanthrene, d12-chrysene and d12-benzo[e]pyrene. Lay-flat LDPE tubing was purchased from Brentwood Plastics Inc. (St. Louis, USA), and is similar to that used for SPMDs. The tubing was reproducibly cut along the two edges resulting in 2.5 cm wide LDPE membranes (average membrane thickness of 80 lm) and mounting loops were made at either end using a heat-sealer. Once rinsed with tap and ultra pure water samplers were air dried before being soxhlet extracted at low temperature with 50:50 methanol/hexane for 8 h. Samplers were left to dry before spiking with d10-acenaphthene, d10-fluorene, d10phenanthrene, d10-fluoranthene and d12-chrysene, as PRCs. The spiking procedure involved placing each sampler in a 50:50 methanol/water solution fortified with 5 lg of the deuterated PAHs in
471
50 mL glass tubes (Booij et al., 2003). This solution was then shaken in the dark at 100 rpm for 72 h, though equilibrium is likely to be reached much earlier (Booij et al., 2002). Samplers were then rapidly dried with a clean tissue to remove solution from the surface and left to dry for 1 min before being stored in clean jars at 20 °C until deployment. AlteSil™ silicone sheets which averaged 570 lm in thickness were purchased from Altec Ltd. (Bude, UK). 2.5 cm wide strips were cut and the length was adjusted to that of SPMDs (as were LDPE membranes). These samplers were processed in a similar way to LDPE membranes, with soxhlet extraction using a mixture of 50:50 methanol/pentane. Spiking with 4 lg of deuterated PAHs was undertaken using the same procedure as for LDPE membranes. In short, each silicone strip was added to a 150 mL (50:50 water/methanol) and left on an orbital shaker for 72 h at 100 rpm. Once ready, samplers were stored at 20 °C. Purpose-made clips to allow fastening of silicone to deployment equipment, were soxhlet cleaned prior to use. The relative standard deviations of measurements of PRCs in non-exposed spiked samplers, were mostly well below 10% (n = 8) for LDPE membranes and silicone strips. 2.3. Extraction and analysis All controls, trip controls and exposed passive samplers were cleaned thoroughly by washing with ultra pure water and wiped with a clean tissue. SPMDs were placed in a pre-cleaned glass jar and dialysed with 100 mL hexane. The procedure was repeated after 24 h (Harman et al., 2008a, b). Hexane extracts were reduced by a stream of nitrogen to 1 mL. As a clean-up step, extracts were partitioned with acetonitrile (HPLC grade) and the acetonitrile portion quantitatively removed and reduced prior to analysis. Extraction of LDPE membranes and silicone strips was similar to that given above for SPMDs, except the solvent used was methanol instead of hexane. Extracts were analysed on a HP-6890 Plus gas chromatograph equipped with a HP 5973 mass selective detector, operated in single ion monitoring mode (SIM) with electron impact ionisation (70 eV). Analytes were separated on a 30 m DB-5 column (0.25 mm i.d. and 0.25 lm film thickness, Agilent JW Scientific, Santa Clara, USA) and with a helium flow of 1 mL min1. The injection was splitless and the injection volume was 1 lL. The GC oven temperature was held for 2 min at 60 °C before increasing to 250 °C at a rate of 7 °C min1. The final step was an increase to 310 °C at a rate of 15 °C min1 (held for 6 min). Injector, transfer line, ion source and quadruple temperatures were set to 300, 280, 230 and 150 °C, respectively. Quantification of individual compounds was performed by using the relative response of surrogate internal standards. 2.4. Exposure in the Drammen River The present study was mainly conducted on the Drammen River (Norway) at a site (59°450 1300 N; 10°00 2200 E) approximately 8 km upstream of where the river flows into the fjord. This ensured relatively well-mixed river water and unidirectional flow. The water temperature during the exposure gradually decreased from 10 °C to 5 °C from September to November 2008. The level of suspended particulate matter (SPM) was low (1 mg L1). Total organic carbon (TOC) indicated that much of the organic carbon in the water was present as dissolved organic carbon rather than particulate. TOC was measured on two occasions with values between 3 and 4 mg L1. The water flow was approximately 300 m3 s1. Five replicates of each type of passive sampling device were exposed in the river for periods of 24 and 51 d (with the exception of LDPE at 51 d). The two exposures were overlapping with the 24 d exposure initiated 2 weeks after the start of the 51 d deployment.
472
I.J. Allan et al. / Chemosphere 79 (2010) 470–475
Five replicate control samplers were prepared and stored at 20 °C in the freezer to check for contamination during construction and extraction procedures. A further three samplers were used as trip controls, brought to the deployment site and opened to the air to check for contamination by target analytes during deployment and retrieval operations. Since all types of passive samplers were purposely prepared with the same surface area and configuration, a further requirement here was to deploy these samplers under similar water turbulences. This was achieved by deploying samplers in commercially available SPMD cages (Environmental Sampling Technologies, St. Joseph, USA), which were attached to specifically installed moorings. Samplers were mounted onto spider holders inside the deployment cages (five per cage), distributed across three cages for the 24 d exposure and two cages for the 51 d exposure. All cages were held vertical and perpendicular to the river water flow. 2.5. Exposure in the Alna River All types of samplers were also deployed for 28 d in the Alna River, a relatively small watercourse flowing through Oslo (59°540 1600 N; 10°470 3100 ). The temperature of the water during sampler exposure in December 2008 decreased from 5 °C to just above freezing with ice forming in river in the last week of exposure. The water flow was 1.5 m3 s1. The SPM content of 23 mg L1 on average was much higher than that measured in the Drammen River. TOC was on average 6 mg L1. Triplicate SPMDs were deployed in similar cages to those used in the Drammen River, while LDPE membranes and silicone strips were attached to flat cages made of stainless steel mesh with 1 in. wide openings. Since the water depth was <1 m, flat cages were deployed parallel to the water flow. 3. Results and discussion 3.1. Performance reference compound dissipation rates for exposures in the Drammen River Dissipation of PRCs with log Kow values in the range 3.5–6.0 for all three types of samplers and both exposure periods was observed and allowed the estimation of sampling rates, Rs:
NPRC ¼ N0;PRC eke t
ð1Þ
where N0,PRC and NPRC are PRC masses in the samplers prior to and following exposure, respectively, and where the dissipation rate ke is given by:
ke ¼
kO A RS ¼ K SW V K SW V
ð2Þ
with kO is the overall mass transfer coefficient, A the surface area of the sampler (cm2), V the volume of the sampler (cm3) and RS the analyte sampling rate (L d1). Ksw values used here were from Booij et al. (2003) and Smedes et al. (2009) for LDPE membranes and silicone strips, respectively, and from Huckins et al. (2006) for SPMDs. As shown by Eq. (2), the dissipation rate, ke is proportional to A/V and this needs accounting for when comparing PRC data for samplers with differing configurations. Here, while the surface area of the samplers was kept constant, sampler volumes were dissimilar as a result of different thicknesses. Nominal volumes of LDPE membranes, silicone strips and SPMDs were 1.95, 13.05 and 4.95 cm3, respectively. Dissipation rates were normalised to the A/V ratio, log-transformed and are shown in Fig. 1 as a function of log Kow. Log Kow values used here and throughout this study were those given by Smedes et al. (2009). As expected, log keV/A values generally
appear to decrease with increasing log Kow values for all samplers. Despite dissipation for PRCs with log Kow 6.0 for LDPE membranes and SPMDs representing just over 10% of starting concentrations, these were significant (one-sided t-test at 95%). Fig. 1A shows a clear overlap of dissipation rates for PRCs with log Kow 4.0–5.5 for all samplers for the 24 d exposure. The spread of the data is mostly 0.2–0.3 of a log unit and is below values observed during the comparison of a wider range of passive sampling devices with analysis conducted in three different laboratories (Allan et al., 2009). Silicone and SPMD data for the 51 d exposure are consistent and the within-sampler variability appears extremely low (Fig. 1B). Most log keV/A values (cm h1) reported by Allan et al. (2009) for the Chemcatcher passive sampling device, LDPE, silicone strips and SPMDs are in the range 3.6 to 4.6, with values between 5 and 6 for PRCs with log Kow of 5.8 for LDPE membrane samplers. Values observed in the present study are lower, but of the same order as those reported by Rusina (2009) in a calibration experiment with AlteSil™ silicone membrane. Eq. (2) and ke values could be used to estimate sampling rates for PAH for each sampler type and exposure period. Generally these sampling rates are at the lower end of the range observed with samplers of this size (Augulyte and Bergqvist, 2007; Ellis et al., 2008; Gourlay-France et al., 2008). Despite dissipation >90% for some PRCs (after 51 d exposure), little differences in Rs estimates were observed between those from the 24 and 51 d exposures, indicating that these may remain reliable despite significant offload. Under membrane-influenced exchange kinetics, a plateau in ke values may be expected. However, when a major portion of the mass transfer resistance resides in transport across the water boundary layer, ke values may be expected to decrease with increasing log Ksw or Kow (Booij et al., 2003; Huckins et al., 2006; Rusina, 2009). A clear plateau can be seen on Fig. 1A for the 24 d exposure of LDPE membrane samplers for PRCs with log Kow < 4.2. A similar plateau is also shown for SPMDs for both exposures (Fig. 1A and B) for PRCs with log Kow < 4.2. Silicone strips do not appear to follow this trend and no obvious plateau can be observed. This is likely caused by analyte diffusion coefficients being 2–3 orders of magnitude higher in silicone than in LDPE (Rusina, 2009), significantly reducing resistance to mass transfer in the silicone. This in turn extends boundary layer-controlled conditions to lower log Ksw/Kow values. 3.2. Boundary layer-controlled PRC dissipation for exposures in the Drammen River Linear regressions of log ke (d1) vs. log Kow were undertaken for all PRCs assumed to be under boundary-layer-controlled conditions. When using all significant PRC dissipation rates, linear relationships in the range 0.69 to 1.0 could be observed for all samplers (data not shown) for both deployment periods and these were supported by low values of the standard error of the slope (se in the range 0.02–0.07) and R2 values close to 1 (data not shown). Only PRCs with dissipation between 20% and 80% over the 24 or 51 d exposures were then used to determine slopes shown in Table 1 since a lower uncertainty in PRC offload may be expected for this range than when using values close to 0% or 100%. Resulting slopes were very close to 1.0 for all samplers with se values in the range 0.03–0.17. Despite minor differences, slopes were similar, providing evidence that mass transfer resistance in the boundary layer generally dominates overall mass transfer for analytes with log Kow > 4.0. Slopes obtained in our study were steeper than those of 0.42 obtained by Booij et al. (1998) but generally less steep than those of 1.039 and 1.171 obtained by other workers (Rusina, 2009) for the uptake of PAHs and PCBs into silicone and the offload of deuterated PRCs during two calibration experiments performed in the laboratory. Slopes were closer to those obtained
473
I.J. Allan et al. / Chemosphere 79 (2010) 470–475
-4.0
-4.0
(B) 51 day exposure
-4.5
-4.5
-5.0
-5.0
-5.5
-5.5
-1
Log keV/A (cm h )
(A) 24 day exposure
-6.0
-6.0
LDPE membrane Silicone strip SPMD
-6.5 3.5
4.0
4.5
5.0
5.5
6.0
-6.5 3.5
4.0
4.5
5.0
5.5
6.0
6.5
Fig. 1. PRC dissipation rates normalised to sampler surface area to volume ratio (log keV/A in cm h1) for 24 (A) and 51 (B) day exposures of LDPE membranes, silicone strips and SPMDs in the Drammen River.
Table 1 Results of linear regressions log ke–log Kow for PRCs under boundary layer-controlled exchange for passive sampler exposures of 24 and 51 d in the Drammen River. se
R
n
Exposure (d)
0.03 0.17 0.16
0.99 0.85 0.83
15 20 10
24 24, 51 24
log ke = (slope) log Kow + b. Standard error of the slope a. Number of data points where dissipation was in the range 20–80%.
by Rusina (2009) when only using PRC with dissipation between 20% and 80%. However, our field-derived data is based, at best, on two timepoints of 24 or 51 d compared with calibration data by Rusina (2009) which is based on a time series. According to Eq. (2), the product of log ke and log Ksw indicates how Rs varies with increasing analyte hydrophobicity. With log Ksw–log Kow slopes close to unity (with a standard error, se = 0.2 of a log-unit) for silicone material (Smedes et al., 2009), sampling rates are not expected to decrease much with increasing PAH hydrophobicity. However applying the log Ksw–log Kow slope of 1.48 (se = 0.30 logunit) obtained for LDPE by Smedes et al. (2009) results in increasing sampling rates with increasing hydrophobicity. When using other published log Ksw–log Kow relationships for the same polymer material (Booij et al., 2003), this increase in Rs with increasing log Kow is not observed. Such an artefact may be the result of the uncertainties in Ksw values and ke values or of some limitations with the use of Eq. (2) to estimate sampling rates for analytes under boundary layer-controlled uptake. The uncertainty in Ksw and ke values may therefore account for a large proportion of the variability in calculated dissolved contaminant concentrations for a range of passive sampling devices (Allan et al., 2009). 3.3. PAH masses absorbed during exposures in the Drammen River The 24 d sampler exposure in the Drammen River resulted in the detection of PAHs with molecular weights no greater than that of chrysene (as shown on Fig. 2A) and benzo[e]pyrene for the 51 d deployment for both silicone strips and SPMDs (see Fig. 2B). For analytes in linear uptake, the mass absorbed into the sampler is dependent on the sampling rate and the dissolved analyte concentration in water. Boundary layer-controlled uptake for silicone strips was shown to extend to analytes with log Kow of 4.0, indi-
50 -1
–1.092 –1.004 –1.03
c
ng sampler
Slope
LDPE membrane Silicone strip SPMD
b
LDPE membrane Silicone strip SPMD
(A) 24 day exposure
40 30 20 10 0 PYR
FLUO
CHRY
140 (B) 51 day exposure
120 100 -1
c
Sampler
b
ng sampler
a b
a
60
80 60 40 20 0 PYR
FLUO
CHRY
BbjF
BeP
Fig. 2. Masses of pyrene (PYR), fluoranthene (FLUO), chrysene (CHRY), benzo[b and j]fluoranthene (BbjF) and benzo[e]pyrene (BeP) absorbed in the different passive sampling devices following 24 and 51 d exposures in the Drammen River. Note: LDPE membranes were not exposed for 51 d.
cating that mass transfer coefficients for these exposures are relatively low. When uptake is linear and boundary layer-controlled, masses of analytes absorbed by the samplers should be independent of the sampler material (as long as they have sufficient capacity). Masses of pyrene, fluoranthene and chrysene accumulated in silicone strips and SPMDs over 24 d appear consistent with such an assumption (Fig. 2). The mass of pyrene absorbed in LDPE mem-
I.J. Allan et al. / Chemosphere 79 (2010) 470–475
3.4. Exposure in the Alna River Despite the low temperature of the Alna River water, 28 d exposure was sufficient to observe significant dissipation of PRCs up to d12-chrysene for all samplers. Dissipation of the least hydrophobic PRCs was almost complete, though still quantifiable. Log keV/A values for all samplers plotted on Fig. 3A show a clear decrease for PRCs with log Kow > 4.5 while for less hydrophobic PRCs, normalised ke values appear to plateau. Although this could be caused by an increase in the uncertainty in estimates of dissipation rates when offload is almost complete, values calculated for the exposure of LDPE membranes and SPMDs in the Alna River are similar to those found for the Drammen River. While overlapping of PRC data for LDPE membranes and SPMDs may be observed on Fig. 3A, PRC dissipation rates for silicone strips are generally higher and the plateau is not as obvious. Slopes of log ke–log Kow regressions for PRCs with log Kow > 5 were 1.12, 0.76 and 0.72 for LDPE membranes, silicone strips and SPMDs, respectively and are generally close to those obtained in the Drammen River deployment. Sampler–water exchange kinetics appear higher for the exposure in the Alna compared with those from the Drammen River. However, no clear impact of using different cage designs could be observed during this Alna River deployment. Rs values across
-3
(A)
-4
Log keV/A
brane samplers did not differ from those found in the two other samplers while that of fluoranthene was 80% (Fig. 2A). Chrysene in LDPE extracts was below detection limits. The 51 d exposure resulted in the measurement of extremely similar masses of analytes from pyrene up to benzo[e]pyrene for silicone strips and SPMDs (Fig. 2B). This shows the integrative nature of the uptake over 51 d for analytes with log Kow > 5–5.2 when using these two types of samplers. In most cases, relative standard deviations were well <10% for all three types of samplers at each time interval (n = 5). Considering the very low between sampler type variability in absorbed masses shown in Fig. 2 for analytes under boundary layer-controlled uptake, it is likely that a substantial increase in the variability of calculated dissolved contaminant concentrations will be caused by variability associated with Ksw and ke values. In order to confirm the transition between analytes under integrative uptake and those close to equilibrium, ratios of masses absorbed during 51 d (N51) over those accumulated during the 24 d exposure (N24) normalised to respective deployment times (N51/ 51:N24/24) were calculated for analytes with log Kow in the range 3.9–6.0. Ratios of 1 are expected: (i) if sampling is integrative, (ii) if no changes in water turbulences near the samplers occur during the two exposures, (iii) if there are no significant influences from biofilm growth at the surface of the samplers and (iv) if the dissolved PAH concentration in water does not vary significantly during the 51 d period. A ratio of 24/51 (i.e. 0.47) should be observed if equilibrium is reached within 24 d and if no further changes in analyte concentration in the sampler are observed. Ratios close to or slightly higher than 1 were observed for analytes with log Kow > 4.5 for silicone strips and SPMDs, respectively. Ratios for anthracene and phenanthrene were very close to 1 for silicone strips while values 0.75 were found for SPMDs. This confirms that uptake into silicone strips used in this study is likely to remain linear for a longer period of time than for SPMDs as a result of their larger volume. Some PAHs with log Kow < 4.5 were found to be close to equilibrium within the 24 d exposure with ratios generally between 0.45 and 0.8. Ratios of 1 indicate that biofouling (visually observed algal growth at the surface of the samplers) appeared to have had minimal effects on the uptake of analytes with log Kow > 5. In contrast with recent work by Harman et al. (2009), the biofouling here did not appear to be affected by the type of material the sampler is made from (Booij et al. (2006)).
-5
-6
LDPE membrane Silicone strip SPMD
-7 3.5
4.0
4.5
5.0
5.5
6.0
Log KOW (B)
msampler /msilicone strip
474
1
LDPE membrane Silicone strip SPMD
0.1
3.5
4.0
4.5
5.0
5.5
6.0
6.5
7.0
Log KOW Fig. 3. PRC dissipation rates normalised to sampler surface area (A) to volume (V) ratio (keV/A in cm h1) vs. log Kow for 28 d exposure of LDPE membranes, silicone strips and SPMDs in the Alna River (filled symbols) (A) and PAH masses absorbed by the various samplers normalised to those accumulated in silicone strip samplers (B). White symbols on (A) represent data for the 24 d exposure in the Drammen River.
the array of samplers and analytes were in the range 2.7– 32 L d1 with highest values found for silicone strips. Concentrations of PAHs were higher in samplers exposed in the Alna compared with exposure in the Drammen River and this is the result of higher sampling rates and also likely higher water concentrations. Similarly to the Drammen River exposure, the variability between replicate samplers from the Alna River was low (Fig. 3B). PAH masses absorbed by the various replicate samplers were normalised to the mean mass absorbed in silicone strips. The time required for the analytes with log Kow < 5.2 to reach half of equilibrium concentration was lower than the exposure time and thus sampling was not linear over 28 d. Masses absorbed into LDPE membranes and SPMDs were lower than those found in silicone strips as a result of the lower volumes of these samplers. Less significant differences were found for analytes with log Kow > 5.8 when uptake was expected to be linear and boundary-controlled (Fig. 1B). Despite the high water velocities encountered (as apparent from the PRC data), the use of two different cage designs may have had a limited effect on sampling rates, possibly as a result of
I.J. Allan et al. / Chemosphere 79 (2010) 470–475
ice forming in the water towards the end of the exposure and affecting both cage designs differently.
4. Conclusions The present study investigated the effects of sampler material on the uptake of PAHs and the release of PRCs from samplers with the same surface area and under similar water turbulences deployed in the Drammen River. Consistent PRC dissipation data was obtained with LDPE membranes, silicone strips and SPMDs for 24 and 51 d exposures. The threshold for boundary layer-controlled exchange kinetics observed through PRC release from the samplers was confirmed by extremely similar masses of analytes with log Kow > 4.5 absorbed by silicone strips and SPMD over 24 and 51 d. Differences in masses of analytes under boundary layer-controlled uptake absorbed in samplers made of different materials are much lower than differences observed between PRC-based Rs estimates for these samplers. This implies that much of the variability in estimates of dissolved contaminant concentrations seen in passive sampler intercomparison studies is the result of the uncertainty associated with Ksw and ke values and the mode of calculation of sampling rates. The consistency of data obtained with samplers tested here combined with the low variability observed supports the use of simpler single-phase samplers such as LDPE membranes and silicone strips. Acknowledgements We acknowledge financial support from the Norwegian Climate and Pollution Agency (Klif) under Contract 5009 and from NIVA’s internal funding process (E-project GLIMPS, O-28074). Views presented here are those of the authors alone. References Allan, I.J., Booij, K., Paschke, A., Vrana, B., Greenwood, R., Mills, G.A., 2009. Field performance of seven passive sampling devices for monitoring of hydrophobic substances. Environ. Sci. Technol. 43, 5383–5390. Anderson, K.A., Sethajintanin, D., Sower, G., Quarles, L., 2008. Field trial and modeling of uptake rates of in situ lipid-free polyethylene membrane passive sampler. Environ. Sci. Technol. 42, 4486–4493. Augulyte, L., Bergqvist, P.A., 2007. Estimation of water sampling rates and concentrations of PAHs in a municipal sewage treatment plant using SPMDs with performance reference compounds. Environ. Sci. Technol. 41, 5044–5049.
475
Booij, K., Sleiderink, H.M., Smedes, F., 1998. Calibrating the uptake kinetics of semipermeable membrane devices using exposure standards. Environ. Toxicol. Chem. 17, 1236–1245. Booij, K., Smedes, F., van Weerlee, E.M., 2002. Spiking of performance reference compounds in low density polyethylene and silicone passive water samplers. Chemosphere 46, 1157–1161. Booij, K., Hofmans, H.E., Fischer, C.V., Van Weerlee, E.M., 2003. Temperaturedependent uptake rates of nonpolar organic compounds by semipermeable membrane devices and low-density polyethylene membranes. Environ. Sci. Technol. 37, 361–366. Booij, K., van Bommel, R., Mets, A., Dekker, R., 2006. Little effect of excessive biofouling on the uptake of organic contaminants by semipermeable membrane devices. Chemosphere 65, 2485–2492. Ellis, S.G., Booij, K., Kaputa, M., 2008. Comparison of semipermeable membrane device (SPMD) and large-volume solid-phase extraction techniques to measure water concentrations of 4, 40 -DDT, 4, 40 -DDE, and 4, 40 -DDD in Lake Chelan, Washington. Chemosphere 72, 1112–1117. Gourlay-France, C., Lorgeoux, C., Tusseau-Vuillemin, M.H., 2008. Polycyclic aromatic hydrocarbon sampling in wastewaters using semipermeable membrane devices: accuracy of time-weighted average concentration estimations of truly dissolved compounds. Chemosphere 73, 1194–1200. Harman, C., Boyum, O., Tollefsen, K.E., Thomas, K., Grung, M., 2008a. Uptake of some selected aquatic pollutants in semipermeable membrane devices (SPMDs) and the polar organic chemical integrative sampler (POCIS). J. Environ. Monit. 10, 239–247. Harman, C., Tollefsen, K.E., Boyum, O., Thomas, K., Grung, M., 2008b. Uptake rates of alkylphenols, PAHs and carbazoles in semipermeable membrane devices (SPMDs) and polar organic chemical integrative samplers (POCIS). Chemosphere 72, 1510–1516. Harman, C., Bøyum, O., Thomas, K.V., Grung, M., 2009. Small but different effect of fouling on the uptake rates of semipermeable membrane devices and polar organic chemical integrative samplers. Environ. Toxicol. Chem. 28, 2324–2332. Huckins, J.N., Manuweera, G.K., Petty, J.D., Mackay, D., Lebo, J.A., 1993. Lipidcontaining semipermeable membrane devices for monitoring organic contaminants in water. Environ. Sci. Technol. 27, 2489–2496. Huckins, J.N., Petty, J.D., Lebo, J.A., Almeida, F.V., Booij, K., Alvarez, D.A., Clark, R.C., Mogensen, B.B., 2002. Development of the permeability/performance reference compound approach for in situ calibration of semipermeable membrane devices. Environ. Sci. Technol. 36, 85–91. Huckins, J.N., Petty, J.D., Booij, K., 2006. Monitors of Organic Chemicals in the Environment: Semipermeable Membrane Devices. Springer, New York. Rusina, T.P., 2009. New methods of sampling and sample pre-concentration of hydrophobic compounds in aquatic systems: magnifying passive sampling aspects. RECETOX – Masaryk University, Brno. p. 100. Rusina, T.P., Smedes, F., Klanova, J., Booij, K., Holoubek, I., 2007. Polymer selection for passive sampling: a comparison of critical properties. Chemosphere 68, 1344–1351. Smedes, F., Geertsma, R., van der Zande, T., Booij, K., 2009. Polymerwater partition coefficients of hydrophobic compounds for passive sampling: application of cosolvent models for validation. Environ. Sci. Technol. 43, 7047–7054. Vrana, B., Schuurmann, G., 2002. Calibrating the uptake kinetics of semipermeable membrane devices in water: impact of hydrodynamics. Environ. Sci. Technol. 36, 290–296. Vrana, B., Mills, G.A., Allan, I.J., Dominiak, E., Svensson, K., Knutsson, J., Morrison, G., Greenwood, R., 2005. Passive sampling techniques for monitoring pollutants in water. TrAC. Trends Anal. Chem. 24, 845–868.