a n a l y t i c a c h i m i c a a c t a 6 0 7 ( 2 0 0 8 ) 37–44

available at www.sciencedirect.com

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

Analysis of emerging contaminants in sewage effluent and

river water: Comparison between spot and passive sampling

Zulin Zhang, Andrew Hibberd, John L. Zhou ∗

Department of Biology and Environmental Science, School of Life Sciences, University of Sussex,
Falmer, East Sussex, Brighton BN1 9QG, UK

a r t i c l e i n f o a b s t r a c t

Article history: Passive sampling is highly complimentary to spot sampling in environmental analysis. A
Received 30 August 2007 polar organic chemical integrative sampler (POCIS) was extensively tested to optimize its
Received in revised form performance under both controlled and field conditions. The passive sampler was subse-
8 November 2007 quently used for the sampling and analysis of estrone, 17␤-estradiol, 17␣-ethynylestradiol,
Accepted 15 November 2007 bisphenol A, propranolol, sulfamethoxazole, meberverine, thioridazine, carbamazepine,
Published on line 21 November 2007 tamoxifen, indomethacine, diclofenac and meclofenamic acid in sewage effluent and river
water. Under laboratory conditions, the kinetics of compound uptake by POCIS were linear
Keywords: during 10-day of exposure. POCIS sampling rates of the target compounds were signifi-
Passive sampling cantly greater by using polyethersulfone instead of polysulfone membrane, and enhanced
Pharmaceuticals with increasing sorbent exposure area. Together with spot water sampling, the optimized
Endocrine disrupting chemicals POCIS was deployed in the River Ouse, West Sussex, UK to obtain field-derived sampling
Liquid chromatography–tandem rates which, for endocrine disrupting chemicals (EDCs), were significantly higher than those
mass spectrometry from laboratory experiments. Both spot and passive sampling demonstrated that most of
Gas chromatography–mass the target chemicals were frequently detected in sewage effluent and river waters, and that
spectrometry the daily changes in the pollutant concentrations were greater for pharmaceuticals than
for EDCs. The aqueous concentrations of all compounds were elevated at a sewage outfall,
which is confirmed to be an important source of the target compounds in the river. The
validated POCIS was then successfully used to estimate the concentrations of the target
compounds in effluent and river water, which were in good agreement with those from spot
sampling for pharmaceuticals.
© 2007 Elsevier B.V. All rights reserved.

1. Introduction 17␣-ethynylestradiol (EE2) is the main component of the oral

Water pollution from emerging pollutants such as endocrine
disrupting chemicals (EDCs), pharmaceuticals and personal
care products (PPCPs) is one of the important aspects of cur-
rent environmental research due to their interference with
the endocrine system of wildlife and human [1,2]. Most
EDCs and PPCPs are man-made organic chemicals being
introduced to the environment by anthropogenic inputs, e.g.
contraceptive pill, and carbamazepine and diclofenac are used
as anti-epileptic and anti-inflammatory drugs, respectively
[3–5]. In addition, EDCs can be naturally occurring in the envi-
ronment, e.g. female hormones estrone (E1) and 17␤-estradiol
(E2) are both excreted by women and hence ubiquitous in the
aquatic environment. Such compounds may not be removed
by sewage treatment works (STW) and are reactivated during
those processes.

∗
Corresponding author. Tel.: +44 1273 877318; fax: +44 1273 678937.
E-mail address: j.zhou@sussex.ac.uk (J.L. Zhou).
0003-2670/$ – see front matter © 2007 Elsevier B.V. All rights reserved.
doi:10.1016/j.aca.2007.11.024
38 a n a l y t i c a c h i m i c a a c t a 6 0 7 ( 2 0 0 8 ) 37–44
Considering the potential impacts of EDCs and PPCPs, they
should be routinely monitored in aquatic systems. The most
widely used technique for performing such monitoring is spot
sampling followed by laboratory-based extraction and analy-
sis [6–8]. This approach, however, yields only an instantaneous
measurement of pollutant levels and suffers from the uncer-
tainty of short- and long-term concentration variations. An
increase in sampling frequency or the use of flow- and time-
weighted automatic samplers may reduce such uncertainty;
however, the associated increase in costs may prove unfeasi-
ble [9,10].
There has been rapid development in the use of passive
sampling devices that allow continuous monitoring of aque-
ous pollutants without the disadvantages of using organisms
[9–12]. Of the various passive sampling devices, the most
widely used is the semi-permeable membrane device (SPMD)
consisting of a tubular polyethylene membrane containing
a film of lipids such as triolein [13,14]. In comparison to
traditional water sampling, SPMD can be standardized and
deployed over a long period, can detect low levels of organic
contaminants, and mimics the bioconcentration of pollutants
in aquatic organisms [15]. However, SPMD is not suitable
for polar organics, as either the membrane is impermeable
to such compounds or accumulation is thermodynamically
unfavorable due to the low affinity of the receiving phase for
such analytes [16]. As it is the polar compounds (e.g. EE2) that
are primarily responsible for the estrogenic activities, recent
development has been focused on their passive sampling,
e.g. polar organic chemical integrative sampler (POCIS) and
Chemcatcher [17,18]. POCIS comprises a solid receiving phase
(sorbent) sandwiched between two microporous polyethersul-
fone (PES) membranes. POCIS samples from water and thereby
enables the chemical concentration to be estimated as follows
[12,17]:
Ms = Cw Rs t
where Ms is the mass of analytes in the receiving phase at
time t, Cw represents time-weighted average concentration in
water during the deployment period. Rs is the sampling rate of
the system, which may be interpreted as the volume of water
cleared of analyte per unit of exposure time by the device [19].
Chemcatcher uses a polytetrafluoroethylene (PTFE) sup-
port device to protect a layer of membrane which covers a solid
receiving phase (e.g. C18 Empore disk). Accumulation rates
and selectivity are regulated by the choice of both the mem-
brane and the solid receiving material; both are supported
and sealed in place by an inert plastic housing [12,18]. Both of
these samplers have advantages over traditional methods of
sampling; however, further improvement and validation are
needed. For example, existing calibration experiments were
only performed in small systems, e.g. 1 L microcosms [17].
Certain factors (e.g. pollutant concentration, salinity) which
may affect the sampling rate were not reported [17,18]. Fur-
thermore, in order to estimate water concentration more
accurately, it is essential to validate laboratory-derived with
in situ sampling rates.
This paper focused on the validation and application of
POCIS for emerging contaminants in water, including evalua-
tion of spot and passive sampling for studying the occurrence
and behavior of emerging contaminants in sewage effluent
and river waters. Robust extraction and analytical methods
have been developed based on solid-phase extraction (SPE)
followed by gas chromatography–mass spectrometry (GC–MS)
for EDCs [20,21] and liquid chromatography–tandem mass
spectrometry (LC–MS–MS) for pharmaceuticals [22]. Labora-
tory calibration was performed to evaluate the sampling rate
of the modified POCIS for the target compounds. The effects
of different conditions (e.g. size of sampler) on POCIS sam-
pling rates were evaluated. The POCIS sampling rates derived
from laboratory conditions were compared with those from
field experiments. Estimation of the ambient aqueous concen-
tration of target chemicals was achieved by using field-based
sampling rates.
2. Experimental
2.1. Materials
The solvents used including methanol and ethyl acetate were
of distilled-in-glass grade from Rathburns Chemicals Ltd.
(Scotland). Compounds E1, E2, EE2, E2-d2 , propranolol (pro),
sulfamethoxazole (sul), meberverine, thioridazine, carba-
mazepine (carb), tamoxifen, indomethacine (indo), diclofenac
(diclo) and meclofenamic acid were purchased from Sigma,
UK, and bisphenol A (BPA) and BPA-d16 were supplied by
Aldrich (Dorset, UK). Diuron-d6 and 13 C-phenacetin were
purchased from Cambridge Isotope Laboratories, USA. Sep-
arate stock solutions of individual compounds (1000 mg L−1 )
were prepared by dissolving an appropriate amount of each
substance in methanol. From these standards, a mixture
of working standards or internal standards (BPA-d16 , E2-
d2 , diuron-d6 , 13 C-phenacetin) containing each compound
at 10 mg L−1 was prepared by diluting the stock solution in
(1) methanol. All the standard solutions were stored at −18 ◦ C
prior to use. Ultrapure water was supplied by a Maxima Unit
from USF Elga, UK.
The passive sampling device was similar to POCIS [17]
except the holder which supports both the diffusion-limiting
membrane and sorbent and seals them in place, was made of
PTFE rather than stainless steel (Fig. 1). In addition, the device
of different exposure diameters (27, 38, 54 mm) was tested.
The Oasis HLB sorbent (100 mg) from Waters Ltd., UK was
chosen because it sorbs a wider range and polar compounds
better than C18 [20,22]. The Oasis sorbent was spread evenly
between the membranes. The PES membrane (0.1 ␮m pore
size) and polysulfone (PS) membrane (0.2 ␮m pore size) were
Fig. 1 – The schematic diagram of the POCIS device (1) PTEE
holder, (2) screw, (3) membrane, (4) sorbent and (5) hole.
 a n a l y t i c a c h i m i c a a c t a 6 0 7 ( 2 0 0 8 ) 37–44
provided by Pall Gelman Sciences (VMR International, UK).
PTFE used to construct the sampler was from Aquarius Plastics
Ltd., UK. Peristaltic pumps (Watson Marlow 401U/DM2) and
tubes (0.5 mm × 1.6 mm and 3.2 mm × 1.6 mm) for controlling
the flow-through system were from Fisher, UK.
2.2. Oasis sorbing capability for target chemicals
Before the sorbents were used as the receiving phase of POCIS,
it was necessary to determine their sorbing capacity for the
target chemicals. Samples of 50 mL of water spiked with 5 ␮g of
each compound were passed through glass columns contain-
ing 100 mg sorbent. The concentrations of target compounds
were selected at levels above typical sequestered chemical
residues from passive samplers to ensure future applications
of POCIS. Analysis of permeate from the sorbent beds showed
concentrations of the chemicals below limits of detection
(LOD), indicating that target chemicals had been retained by
the sorbents. Recovery of the target compounds from the sor-
bents was achieved by eluting the sorbent beds with 20 mL
of methanol, followed by concentration under N2 , and GC-
MS analysis [20,21]. Good recovery (80–87%) of the target
compounds was achieved from replicate sampling and anal-
ysis, confirming that the sorbent has sufficient capacity for
trapping target compounds and can be used in POCIS for mon-
itoring purpose.
2.3. Exposure in a flow-through system
To determine Rs , a flow-through tank allowing the continu-
ous addition of water and target compound solutions was
used throughout the experiments. It is hypothesized that
laboratory-derived Rs will be different from field-derived Rs ,
due primarily to different hydrodynamic regimes. Deionized
water and a water solution of the target compounds were
pumped simultaneously into the exposure tank by the peri-
staltic pumps. The tank was always filled with water (30 L) at
approximately 15 ◦ C. The rates of addition of distilled water
and target compound solution were controlled at approxi-
mately 20 and 1 mL min−1 , respectively. The concentration of
analytes in the exposure tank was checked daily (100 mL, n = 3),
followed by SPE (Oasis HLB), and instrumental analysis [20,21].
The results show a relatively constant chemical concentration
(R.S.D. = 2.1–11.1%) in the exposure tank and a replacement
time for the system of 23.8 h. Following exposure, the POCIS
(n = 3) was removed and dismantled. The sorbent in POCIS
was removed, spiked with 100 ng of internal standards, and
then extracted with 10 mL of methanol 3 times. The extracts
were concentrated to 0.3 mL under nitrogen, and analyzed by
GC-MS for EDCs [20,21]. In addition, a minimum of 3 blanks
for water, membrane and sorbent were analyzed during cal-
ibration experiments to determine the initial levels of target
compounds in various matrices, the results show that none of
the blanks contained the target compounds above LOD.
2.4. Sampler validation and application in the field
Two sites in the River Ouse, West Sussex, England, includ-
ing the outfall of Sheffield Park STW and its downstream
(700 m), were chosen for field validation of POCIS for 2 weeks
39
(29 January to 9 February 2007). The samplers were suspended
vertically at a water depth of 50–100 cm by attaching to a float-
ing ball. POCIS fitted with PES membrane and Oasis sorbent
as the receiving phase with an exposure diameter of 54 mm,
were deployed at each site. Three spot water samples (1 L)
and POCIS were sampled each day. Water samples were fil-
tered through pre-combusted GF/F filters (0.7 ␮m), spiked with
100 ng of internal standards and extracted by SPE. The ana-
lytes were eluted from the cartridges with 10 mL methanol,
which were concentrated to 0.3 mL under N2 , and analyzed
by LC-tandem-MS for pharmaceuticals and GC-MS for EDCs
[20–22]. The results were used to derive field-based sampling
rates for POCIS. Based on the field POCIS blanks, the limits of
quantification (LOQ) for selected target compounds in water,
as calculated from Eq. (1), were from 6 to 487 pg L−1 .
The validated POCIS was deployed in the River Ouse (out-
fall, upstream, downstream) during 23–27 October 2006, for
comparing performance with spot sampling. Again water
samples were taken daily for monitoring the target com-
pounds, while POCIS was sampled at the end of the week-long
experiment. In addition, water properties such as tempera-
ture, pH and dissolved oxygen at each site were recorded at
each visit.
3. Results and discussion
3.1. Flow-through exposure under controlled
conditions
The performance of POCIS was tested by exposure to a
relatively constant concentration of target compounds in
the exposure tank (R.S.D. = 2.1–11.1%). Characteristic analyte
uptake curves for the sampler are shown in Fig. 2, demon-
strating that the uptake of selected target compounds in POCIS
follows a linear pattern. Based on uptake kinetics, the mean
sampling rate varied from 0.036 L day−1 for BPA to 0.069 L day−1
for E1. Alvarez et al. reported that laboratory-derived Rs val-
ues for some other compounds such as diuron, isoproturon,
azithromycin and omeprazole varied from 0.005 to 0.12 L day−1
[17]. Matthiessen et al. [23] estimated a sampling rate from
0.09 to 0.129 L day−1 for E2 when calibrating in a glass beaker
with stirring. Such Rs values are comparable to ours for target
chemicals although they are for different compounds.
The concentrations of organic micropollutants in aquatic
environments are highly variable, both temporally and spa-
tially. To investigate the effect of compound concentrations on
the sampling rate of POCIS, the experiments were performed
with chemical concentrations varying from 10 to 1000 ng L−1 .
The results (Table 1) show that Rs for each compound was not
significantly influenced by compound concentrations, which
are consistent with other reports that the sampling rates for
SPMD are independent of environmental concentration [24].
The findings suggest that POCIS can be applied for monitoring
the target compounds of different concentrations.
3.2. Membrane evaluation
The microporous membrane in POCIS acts as a semi-
permeable barrier between the sorbent and the surrounding
40 a n a l y t i c a c h i m i c a a c t a 6 0 7 ( 2 0 0 8 ) 37–44

Fig. 2 – Accumulation of BPA (1.63 ␮g L−1 ), E1 (1.41 ␮g L−1 ), E2 (2.14 ␮g L−1 ) and EE2 (1.37 ␮g L−1 ) from water by the passive
sampler under laboratory conditions.

aquatic environment. It allows polar organic chemicals to
pass through to the sorbent, while particulates with cross-
sectional diameters greater than the membrane pore size will
be excluded selectively. Without the membrane, direct contact
of these excluded materials with the sorbent can result in a
site-specific bias of apparent contaminant concentrations in
the sorbent, reduce uptake due to enhanced biofouling in the
membrane and sorbent, and potentially interfere with sam-
ple processing and analysis. Kingston et al. [18] suggested
that PS membrane was suitable for polar organic compounds.
Alvarez et al. [17] however concluded that PES membrane
was preferred to sample polar organic contaminants, after
studying many types of membrane and considering fac-
tors such as the biodegradability, pore size, strength and
durability.
In this study, the performance of POCIS made by PES and
PS membrane was compared, by calculating the sampling
rates of POCIS following exposure. As shown in Fig. 3, the
sampling rate of the target compounds varied from 0.033 to
0.048 L day−1 (R.S.D. = 12.5–19.1%) for PES membrane, and from
0.004 to 0.017 L day−1 (R.S.D. = 6.8–13.4%) for PS membrane. As
the PES membrane showed significantly higher Rs than PS

Table 1 – Sampling rate (Rs (L day−1 )) of POCIS with an

exposure diameter of 27 mm at different EDC
concentrations
Concentration (ng L−1 ) BPA E1 E2 EE2

10 0.047 0.034 0.041 0.047

20 0.040 0.039 0.037 0.065
50 0.053 0.040 0.028 0.049
100 0.028 0.026 0.029 0.046
250 0.038 0.040 0.039 0.053
500 0.042 0.039 0.042 0.051
1000 0.033 0.064 0.045 0.046

Average 0.040 0.040 0.037 0.051

Fig. 3 – Comparison of the sampling rates for POCIS
S.D. 0.008 0.012 0.007 0.007
R.S.D. (%) 21.1 28.4 17.6 13.0
(diameter = 27 mm) using PES and PS membranes, for
compounds BPA, E1, E2 and EE2.
 a n a l y t i c a c h i m i c a a c t a 6 0 7 ( 2 0 0 8 ) 37–44
Fig. 4 – Relationship between the sampling rates and the
exposure surface area of POCIS using PES membrane.
membrane, it was therefore selected as the membrane for
POCIS for further experiments.
3.3. Sampler size optimization
The kinetics of trapping the target compounds in passive sam-
plers are expected to be dependent on the size, and perhaps
more importantly, the exposure surface area of such devices.
POCIS with three different exposure areas of 5.72, 11.33 and
22.89 cm2 , respectively, was examined for the uptake of tar-
get compounds. The relationship between the sampling rate
and sampler exposure surface area (Fig. 4) shows that Rs
had a positive relationship with the exposure surface area
of the sampler, with the correlation coefficient (r2 ) of the
regression from 0.82 (EE2) to 1.0 (BPA). The trend is similar
to what was described by Huckins et al. [24] and Vrana et al.
[25] that Rs was proportional to the surface area of the sam-
pling device. As a result, POCIS with a sorbent diameter of
54 mm and a sampling surface area to sorbent mass ratio of
229 cm2 g−1 was chosen as the optimum for further exposure
experiments.
3.4. Effects of environmental condition on sampling
rate
Generally, the speciation of weakly acidic compounds in
aqueous solutions depends on the solution pH. A series of
experiments for investigation of pH effect on sampling rate
were performed, and the results show that Rs for target com-
pounds remains relatively similar among pH 4–10, with R.S.D.
less than 5%. The pKa values of the chemicals are all higher
than 10, varying from 10.2 for BPA to 10.5 for EE2. At pH
value less than 10, the test chemicals will stay predominantly
as neutral molecules. At pH 10, between 32 and 63% of the
compounds will be in dissociated form. The findings there-
fore suggest that both neutral and ionized forms of the target
chemicals can be accumulated in POCIS.
Natural waters can have different salinity. It is well
known that the aqueous solubility of many organic com-
pounds decreases with increasing salt concentration, thus
their absorption efficiency in the sorbent of POCIS may
increase. The effect of water salinity (0, 18‰, 35‰) on POCIS
41
sampling rate was determined following 3-day exposure
in the flow-through system. It was found that Rs did not
vary significantly with changing salinity, with R.S.D. below
12%. Hence the sampler can be applied to many types
of natural waters, such as fresh water, estuarine water or
seawater.
3.5. Occurrence of emerging contaminants in water
measured by spot sampling
The distribution of the target compounds in sewage effluent
and receiving river water was monitored daily, to under-
stand their temporal changes. As shown in Fig. 5, overall,
the mean daily concentrations of BPA, E1 and E2 were
relatively invariable from Monday to Friday, whilst for phar-
maceuticals (propranolol, sulfamethoxazole, carbamazepine,
indomethacine and diclofenac) their concentrations were
more variable, in particular for carbamazepine.
For EDCs BPA, E1 and E2, their concentrations were clearly
elevated at the sewage outfall, suggesting that it is the source
of EDCs in the area. The mean concentrations for E1, E2 and
BPA in the upstream were 9.9, 11.8 and 19.5 ng L−1 , respectively,
which are lower than those (up to 158 ng L−1 of BPA) found in
New Orleans, LA, USA [26]. Similarly Kolpin et al. [27] reported
E2 to be present at average concentrations of 9 ng L−1 in US
surface water.
For pharmaceuticals, propranolol, sulfamethoxazole, car-
bamazepine, indomethacine and diclofenac were regularly
detected in water samples, while meberverine, thioridazine,
tamoxifen and meclofenamic acid were lower than their LOD
in all water samples. Carbamazepine is one of the most fre-
quently detected compounds with the highest concentration
of 652 ng L−1 from sewage effluent. Similarly up to 1075 ng L−1
of carbamazepine was found in Berlin waters [28], and even
higher concentrations of up to 7100 ng L−1 were detected in the
River Elbe and its tributaries [29]. Wiegel et al. [29] suggested
that such levels were due to the high persistency of carba-
mazepine, which was therefore chosen as an tracer substance
for pharmaceutical agents. For the other compounds, the lev-
els of propranolol (1.7–85.2 ng L−1 ) in this study were similar
to those (10–93 ng L−1 ) in Canada [30]. The concentrations of
diclofenac (1.5–91 ng L−1 ) in the Ouse were lower than those
(<20–599 ng L−1 ) detected by Ashton et al. [3] in southeast UK.
For sulfamethoxazole, its concentrations (12.4–25.1 ng L−1 ) in
the STW effluent were also lower than those (243–871 ng L−1 )
in Canadian cites [30].
3.6. Validation of POCIS by field trials
As field environment is very different from laboratory con-
ditions, POCIS performance should be validated in situ. To
understand such differences, POCIS were deployed in the field
to determine Rs values in situ, where target compounds were
measured simultaneously in water and POCIS. As shown in
Fig. 6a, field-derived sampling rates for EDCs were signifi-
cantly greater than those from laboratory experiments, which
is due to significantly high water flow and associated water
turbulence in the field, hence increasing the mass transfer of
EDCs from water to POCIS. During laboratory-based experi-
ments, relatively small flow rates were used with no mixing,
42 a n a l y t i c a c h i m i c a a c t a 6 0 7 ( 2 0 0 8 ) 37–44
Fig. 5 – Daily variations of emerging contaminant
concentrations in sewage effluent and receiving river water
in the River Ouse, West Sussex, UK between 23–27 October
2006.
which explain the lower sampling rates than those from the
field, where factors such as fluctuations in water temperature
and in particular water turbulence would have increased the
uptake kinetics of the compounds in POCIS [17,18,31]. It has
been reported that Rs from turbulent environments will be
much higher than that from quiescent conditions; for some
Fig. 6 – (a) Comparison of the POCIS (diameter = 54 mm)
sampling rates for BPA, E1, E2 between field and
laboratory-derived data; (b) sampling rates for
pharmaceuticals derived from field data. DW: downstream
river water; EW: STW effluent; Lab: laboratory-derived
sampling rates.
compound this difference could be several-fold between the
two conditions [17,19,32,33]. The results therefore suggest the
need for appropriate field validation of such sampling device.
Under field conditions, the sampling rates for BPA, E1 and E2
were higher at downstream site than those from STW efflu-
ent; whilst for pharmaceuticals, the difference between sites
was less clear-cut (Fig. 6b).
3.7. Comparison between spot and passive sampling
The POCIS was deployed and sampled daily at three sites
in River Ouse for a period of 5 days. Following sam-
pling, the target compounds in POCIS and water were
determined. The field-derived sampling rates for the tar-
get chemicals in POCIS were used in the calculation of
the predicted ambient water concentrations by Eq. (1). The
sampling rates derived from downstream river water were
applied to estimate pollutant concentration in water at
upstream and downstream sites, while Rs derived from
effluent was used to predict the compound concentration
in the effluent. Then the predicted compound concentra-
 a n a l y t i c a c h i m i c a a c t a 6 0 7 ( 2 0 0 8 ) 37–44
Fig. 7 – Comparison of the mean (n = 15, 5-day) emerging
contaminant concentrations from spot water sampling with
those predicted by POCIS (diameter = 54 mm) using
field-derived Rs values at three sites in the River Ouse, West
Sussex, UK.
tions in water were compared with those measured by spot
sampling.
As shown in Fig. 7, the concentrations of BPA, E1 and E2
predicted from POCIS are lower than those by spot sampling
except for E1 in the effluent site where the two values are much
43
Fig. 8 – Correlation between spot sampling and POCIS
sampling for all compounds from all sites including
upstream (UW), STW effluent (EW) and downstream (DW).
closer. For the pharmaceuticals their predicted concentrations
from POCIS are similar to those by spot sampling for most
measurements. The mean aqueous concentrations measured
by spot sampling for propranolol, sulfamethoxazole, carba-
mazepine, indomethacine and diclofenac, varied between 3.0
and 45.6 ng L−1 , <LOD and 17.6 ng L−1 , 16.6 and 539 ng L−1 ,
0.4 and 7.2 ng L−1 and 2.4 and 65.2 ng L−1 , respectively; while
their concentrations predicted by POCIS were between 2.8 and
40.5 ng L−1 , <LOD and 18.2 ng L−1 , 26.3 and 427 ng L−1 , 0.5 and
11.9 ng L−1 and 4.4 and 165 ng L−1 , respectively. It is appar-
ent (Fig. 8) that overall, for all compounds from all sites, the
predicted concentrations from POCIS are lower than those
by spot sampling. For carbamazepine in the upstream and
diclofenac in the effluent, the predicted and measured values
differ significantly. When using POCIS in the River Ravens-
bourne and the Thames Tideway in England, Alvarez et al.
[17] showed that the predicted ambient water concentrations
from POCIS for diuron were much higher than those from
spot sampling while for isoproturon an opposite trend was
observed. They suggested that field validation of such sam-
pling device was needed, which has been achieved in this
study.
4. Conclusions
The monitoring of emerging pollutants in the aquatic envi-
ronment has been realized by two approaches: spot sampling
and time-weighted average passive sampling. As demon-
strated in this study, passive sampling device should be
thoroughly assessed both in the laboratory and in the field,
as its performance can differ between the two environments.
There is a good agreement between pharmaceutical con-
44 a n a l y t i c a c h i m i c a a c t a 6 0 7 ( 2 0 0 8 ) 37–44
centrations obtained using spot sampling and those from
passive sampling which has been validated in situ, high-
lighting the potential benefits of using passive sampling for
water quality monitoring (e.g. low maintenance, high repro-
ducibility). However, the agreement between the two sets
of values for EDCs is not satisfactory, and further work is
needed to improve its performance as a complimentary tech-
nique to spot sampling for the integrative analysis of emerging
pollutants.
Acknowledgements
The project was funded by the 6th Framework Programme of
the European Commission (MIF1-CT-2004-510012) and an EU
Interreg grant (162/039/096).
Appendix A. Supplementary data
Supplementary data associated with this article can be found,
in the online version, at doi:10.1016/j.aca.2007.11.024.
references
[1] J.P. Sumpter, Acta Hydrochim. Hydrobiol. 33 (2005) 9–16.
[2] B.H. Sorensen, S.N. Nielsen, P.F. Lanzky, F. Ingerslev, H.C.H.
Lutzhoft, S.E. Jorgensen, Chemosphere 36 (1998) 357–393.
[3] D. Ashton, M. Hilton, K.V. Thomas, Sci. Total Environ. 333
(2004) 167–184.
[4] S.D. Richardson, Anal. Chem. 76 (2004) 3337–3364.
[5] M. Petrović, M.D. Hernando, M.S. Diaz-Cruz, D. Barceló, J.
Chromatogr. A 1067 (2005) 1–14.
[6] R. Jeannot, H. Sabik, E. Sauvard, T. Dagnac, K. Dohrendorf, J.
Chromatogr. A 974 (2002) 143–159.
[7] S.D. Richardson, T.A. Ternes, Anal. Chem. 77 (2005)
3807–3838.
[8] T.A. Ternes, M. Stumpf, J. Mueller, K. Haberer, R.D. Wilken,
M. Servos, Sci. Total Environ. 225 (1999) 81–90.
[9] A. Kot, B. Zabiegala, J. Namiesnik, Trends Anal. Chem. 19
(2000) 446–459.
[10] T. Gorecki, J. Namiesnik, Trends Anal. Chem. 21 (2002)
276–291.
[11] B. Vrana, P. Popp, A. Paschke, G. Schuurmann, Anal. Chem.
73 (2001) 5191–5200.
[12] B. Vrana, G.A. Mills, I.J. Allan, E. Dominiak, K. Svensson, J.
Knutsson, G. Morrison, R. Greenwood, Trends Anal. Chem.
24 (2005) 845–868.
[13] J.N. Huckins, G.K. Manuweera, J.D. Petty, D. Mackay, J.A. Lebo,
Environ. Sci. Technol. 27 (1993) 2489–2496.
[14] J.D. Petty, J.N. Huckins, C.E. Orazio, J.A. Lebo, B.C. Poulton,
R.W. Gale, C.S. Charbonneau, E.M. Kaiser, Environ. Sci.
Technol. 29 (1995) 2561–2566.
[15] J.D. Petty, B.C. Poulton, C.S. Charbonneau, J.N. Huckins, S.B.
Jones, J.T. Cameron, H.F. Prest, Environ. Sci. Technol. 32
(1998) 837–842.
[16] J.D. Petty, J.N. Huckins, D.A. Alvarez, W.G. Brumbaugh, W.L.
Granor, R.W. Gale, A.C. Rastall, T.L. Jones-Lepp, T.J. Leiker,
C.E. Rostad, E.T. Furlong, Chemosphere 54 (2004) 695–705.
[17] D.A. Alvarez, J.D. Petty, J.N. Huckins, T.L. Jones-Lepp, Environ.
Toxicol. Chem. 23 (2004) 1640–1648.
[18] J.K. Kingston, R. Greenwood, G.A. Mills, G.M. Morrison, L.B.J.
Persson, J. Environ. Monit. 2 (2000) 487–495.
[19] B. Vrana, G.A. Mills, E. Dominiak, R. Greenwood, Environ.
Pollut. 142 (2006) 333–343.
[20] R. Liu, J.L. Zhou, A. Wilding, J. Chromatogr. A 1022 (2004)
179–189.
[21] Z.L. Zhang, A. Hibberd, J.L. Zhou, Anal. Chim. Acta 577 (2006)
52–61.
[22] Z.L. Zhang, J.L. Zhou, J. Chromatogr. A 1154 (2007) 205–213.
[23] P. Matthiessen, D. Arnold, A.C. Johnson, T.J. Pepper, T.G.
Pottinger, K.G.T. Pulman, Sci. Total Environ. 367 (2006)
616–630.
[24] J.N. Huckins, J. Petty, H. Prest, R. Clark, D. Alvarez, C. Orazio,
J. Lebo, W. Cranor, B. Johnson, A Guide for the Use of
Semipermeable Membrane Devices (SPMDs) as Samplers of
Waterborne Hydrophobic Organic Contaminants. API
Publication 4690, American Petroleum Institute,
Washington, DC, 2002.
[25] B. Vrana, G.A. Mills, R. Greenwood, J. Knutsson, K. Svensson,
G. Morrison, J. Environ. Monit. 7 (2005) 612–620.
[26] G.R. Boyd, J.M. Palmeri, S.Y. Zhang, D.A. Grimm, Sci. Total
Environ. 333 (2004) 137–148.
[27] D.W. Kolpin, E.T. Furlong, M.T. Meyer, E.M. Thurman, S.D.
Zaug, L.B. Barber, H.T. Buxton, Environ. Sci. Technol. 36
(2002) 1202–1211.
[28] T. Herberer, Toxicol. Lett. 131 (2002) 5–17.
[29] S. Wiegel, A. Aulinger, R. Brockmeyer, H. Harms, J. Loffler, H.
Reincke, R. Schmidt, B. Stachel, W.V. Tumpling, A. Wanke,
Chemosphere 57 (2004) 107–126.
[30] X.S. Miao, F. Bishay, M. Chen, C.D. Metcalfe, Environ. Sci.
Technol. 38 (2004) 3533–3541.
[31] G.S. Ellis, J.N. Huckins, C.E. Rostad, C.J. Schmitt, J.D. Petty, P.
Maccarthy, Environ. Sci. Technol. 14 (1995) 1875–1884.
[32] K. Booij, H.M. Sleiderink, F. Smedes, Environ. Toxicol. Chem.
17 (1998) 1236–1245.
[33] B. Vrana, G. Schuurmann, Environ. Sci. Technol. 36 (2002)
290–296.