Analytica Chimica Acta 486 (2003) 159–169

Determination of colloidally-associated polycyclic aromatic

hydrocarbons (PAHs) in fresh water using C18
solid phase extraction disks
Jeffrey N. Brown, Barrie M. Peake∗
Chemistry Department, University of Otago, P.O. Box 56, Dunedin, New Zealand
Received 9 October 2002; received in revised form 12 February 2003; accepted 9 April 2003

Abstract
A new application of reversed-phase octadecyl (C18 ) solid phase extraction disks has been developed to separate the
colloidally-associated polycyclic aromatic hydrocarbons (PAHs) from those that were truly dissolved in the samples of fresh
water. A correction for the retention of small amounts of colloidal material on the C18 disks was required, which would have
otherwise lead to minor underestimates in the degree of partitioning between the two phases. Using the humic substance
Aldrich Humic Acid (AHA) as a model colloid and the 16 PAHs on the US Enrivonmental Protection Agency priority
pollutant list, the partitioning coefficients of the PAHs between the colloidal and truly dissolved phases were shown to be
proportional to the hydrophobicity of the PAHs, as measured by their octanol water partition coefficients (Kow ). The values for
the partition coefficients obtained (cKdoc ) were similar to those previously reported in the literature using alternative methods,
confirming that the technique was producing acceptable results. The technique allows the in situ partitioning of PAHs between
the truly dissolved and colloidal phases in fresh water bodies to be determined. It will provide an invaluable cross-check of
the laboratory-based methods which often require substantial manipulation of the sample and potentially alter the partitioning
between the phases.
© 2003 Elsevier Science B.V. All rights reserved.
Keywords: C18 disk; Partition coefficients; PAHs; Humic substances; Stormwater

1. Introduction to 1 ␮m), namely macromolecular dissolved organic

matter (DOM) [3]. The presence of this third phase
Traditionally, polycyclic aromatic hydrocarbons has been shown to greatly influence the transport,
(PAHs) in natural waters have been considered to fate and toxicity of organic micropollutants in aquatic
be partitioned between the particulate and dissolved environments [3–5].
phases depending on whether they are retained or Laboratory-based approaches previously used to
pass through a filter (typically 0.7 ␮m pore size) study the partitioning of PAHs between the truly dis-
[1,2]. However, there is increasing recognition that solved phase (<1 nm) and colloidal DOM include the
a significant proportion of the “dissolved” contam- use of surrogate DOM compounds, such as AHA.
inants are adsorbed to colloidal matter (from 1 nm After spiking the solutions with PAHs, the partition-
ing is investigated by indirect techniques, such as
∗ Corresponding author. Fax: +64-3-4797906. fluorescence quenching and solubility enhancement
E-mail address: [email protected] (B.M. Peake). [6,7], or separation of the phases is undertaken via

0003-2670/03/$ – see front matter © 2003 Elsevier Science B.V. All rights reserved.
doi:10.1016/S0003-2670(03)00472-0
160 J.N. Brown, B.M. Peake / Analytica Chimica Acta 486 (2003) 159–169

solid phase extraction (SPE), dialysis or ultrafiltra- extensive clean-up and contaminant desorption steps
tion [8,9]. However, the model DOM may bind the [12,18].
micropollutants to a greater degree than the in situ C18 disks, where the C18 media is impregnated
natural DOM compounds [10,11], possibly due to into a PTFE or glass microfibre filter, were developed
changes induced in the model DOM during its extrac- to extract trace levels of dissolved organic pollutants
tion [3]. Less invasive techniques, such as ultrafiltra- from large water samples. Their advantages over other
tion, may be used to isolate natural aquatic DOM for SPE technologies include higher filtration rates, im-
partitioning studies [7] or filtered natural waters may proved blanks and the elimination of channeling ef-
be used directly [5]. Laboratory-based studies may fects [18–20]. Their extraction efficiency is impaired
not, however, fully reflect the partitioning behavior by DOM which complexes many of the more hy-
of these contaminants in the environment [5,12] as drophobic species and allows them to pass through the
the PAHs are often added at concentrations several disk [21,22]. It, therefore, seemed feasible to investi-
orders of magnitude higher than those present natu- gate the use of C18 disks to determine the in situ par-
rally [11,13]. Also, although the association of PAHs titioning of contaminants between the truly dissolved
to aquatic DOM appears to be rapid (complete within and colloidal (DOM-associated) phases in water sam-
several minutes) [14], long-term time-dependent ef- ples [21] in the same manner as C18 cartridges have
fects may exist [15]. It is, therefore, prudent to vali- been used for laboratory-based partitioning studies.
date laboratory-generated partitioning data using field This paper reports the development of a C18
observations. disk-based method to separate the colloidal and truly
When investigating the in situ partitioning of con- dissolved phases of PAHs present in urban stormwa-
taminants in environmental samples, the traditional ter, which is often highly contaminated with PAHs
techniques, such as ultrafiltration [16], used to iso- [23].
late contaminants associated with the colloidal phase,
do not generally provide an adequate phase separa-
tion for many organic micropollutants because their 2. Methods
hydrophobic nature causes them to bind to the or-
ganic carbon-rich ultrafiltration membrane [8]. Many 2.1. Materials
of the techniques used for laboratory-based partition-
ing studies, e.g. fluorescence quenching and solubility HiPerSolv grade hexane (BDH), acetone (BDH),
enhancement, require the progressive addition of the “far UV” acetonitrile (BDH), and Aldrich Humic
compounds to the sample as part of the method in or- Acid (Aldrich, batch No. 1201816) were used without
der to observe the adsorption isotherm [6]. Such tech- further purification. Dichloromethane (LR grade) was
niques are, therefore, not able to be applied to studies purified by fractional distillation. Anhydrous sodium
of the in situ partitioning of contaminants because the sulfate, glass microfibre filters (Advantec GF75,
compounds added would mask the response of those 0.7 ␮m pore size, 47 mm), filtration glassware, glass
already present in the original samples. wool, 100 ml screw top glass bottles, 125 ml glass
C18 SPE cartridges have been frequently used flasks, and aluminium foil were purified by ashing
in laboratory-based partitioning studies of PAHs at 450 ◦ C for at least 16 h. Silica gel (70–230 mesh)
[9,11,14]. The truly dissolved PAHs contained in was activated at 130 ◦ C for at least 12 h. All other
small samples (<20 ml) are adsorbed onto the C18 glassware was cleaned by successive solvent rinsing,
media, while those associated with DOM pass through soaking in an alkaline detergent bath, distilled water
[11]. Their use for in situ partitioning studies is prob- rinses, air-drying and further solvent rinses. C18 solid
lematic because of the need to use significantly larger phase extraction disks, ENVI-18 DSK (47-mm diam-
sample volumes (1–2 l) which can lead to excessive eter; Supelco, Sigma–Aldrich) and a PAH standard
extraction times (>100 min) [17]. Quantitative elu- mixture containing the 16 US Environmental Protec-
tion of the truly dissolved contaminants may also be tion Agency (EPA) priority PAHs (PM-611, ULTRA
difficult to achieve [11,13]. Other SPE media, for ex- Scientific) were used. Deionised water (Milli-Q) came
ample, XAD-2 resin and polyurethane foam require from a Millipore system.
 J.N. Brown, B.M. Peake / Analytica Chimica Acta 486 (2003) 159–169 161

2.2. Collection, preparation and spiking of water to pH 7.4 by the addition of small volumes of 0.1
samples M NaOH.

2.2.1. Natural water samples 2.2.3. Sample spiking procedure

For the determination of the method detection lim- To facilitate mixing, the PAH standard solutions
its, samples of natural water with minimal PAH con- were injected under the surface of the samples, which
tent [24] were taken from Morrisons Creek, a tributary were then shaken and equilibrated for 24 h. To avoid
of the Water of Leith, Dunedin, New Zealand by sub- the acetonitrile carrier solvent affecting the partition-
merging pre-cleaned 2500 ml amber glass bottles in ing results, the volume of spiking solution added was
the flow. To test the method on natural waters contain- always <0.03% of the volume of the sample solu-
ing PAHs, samples were collected from a stormwater tion [13]. To determine the retention efficiency of the
drain (Portobello Road) that receives the runoff from C18 disks for truly dissolved PAHs, a low (60 ng l−1 )
a large urban catchment of Dunedin City [24] using and a high spiking level (730 ng l−1 ) were used.
a pre-cleaned stainless steel bucket and then tipped The high spiking level exceeded the water solubility
into clean 2500 ml amber glass sample bottles. Sam- of three of the compounds (dibenz[a,h]anthracene,
ples were also collected from the Water of Leith at benzo[ghi]perylene and indeno[1,2,3-cd]pyrene) by
the University of Otago, and from a major stormwater a factor of two. The PAH concentrations in the
drain (St. David Street drain) that enters the Water of other spiked samples were a factor of 5 or more
Leith near this site [24]. These samples (pH ranges, lower than the lowest PAH solubility (300 ng l−1 for
7.2–7.8) were processed without pH modification. dibenz[a,h]anthracene) [25].

2.2.2. Artificial water samples 2.3. Phase separation

Solutions of AHA were prepared by diluting a stock
solution (1000 mg C l−1 ) with Milli-Q, filtering to re- The phase separation procedure (Fig. 1) firstly in-
move any particles >0.7 ␮m and their pH adjusted to volved the removal of particulate-associated PAHs
7.4 by the addition of small volumes of 0.1 M HCl. by filtration of the sample into a second pre-cleaned
The pH of sample blanks of Milli-Q was adjusted 2500 ml amber glass bottle (bottle 2) using a standard

Fig. 1. Schematic diagram of phase separation procedure.

162 J.N. Brown, B.M. Peake / Analytica Chimica Acta 486 (2003) 159–169

glass filtration system. The original 2500 ml sample ing the water-filled pores between the C18 particles
bottle (bottle 1) was rinsed twice with 45 ml of Milli-Q [19]. Three 20 ml volumes of dichloromethane ex-
water to transfer any remaining particles and colloids tracted any truly dissolved PAHs from the walls of the
into the filtration reservoir. A 50 ml sub-sample of the 2500 ml bottle (bottle 2) and continued the elution of
filtrate was taken for DOM analysis and the filter re- the disk. A 5 ml dichloromethane rinse of the filtration
moved. Any truly dissolved PAHs adsorbed onto the reservoir completed the elution procedure.
walls of the original 2500 ml sample bottle (bottle 1) Sample extracts were dried over anhydrous
were removed with two 20 ml dichloromethane rinses sodium sulfate (15 g) with two 20 ml volumes of
which were passed through the filtration apparatus, dichloromethane used to rinse the extract flask. The
together with a 5 ml dichloromethane rinse of the fil- dried extracts were reduced to 5 ml on a rotary evap-
tration reservoir, and collected in a 125 ml glass flask. orator operating at 20 ◦ C to minimize losses of the
The filtration apparatus was cleaned with acetone more volatile 2–3 ring PAHs. The extracts were
and air dried. The C18 disk was then inserted, eluted solvent swapped into hexane (8 ml) and reduced to
under low vacuum with 10 ml acetone and 10 ml 2 ml. Sample extracts were purified using dry packed
dichloromethane to remove possible contaminants silica gel columns (3 g topped with 2 g anhydrous
and then air dried under low vacuum for 5 min. sodium sulfate). The columns were cleaned with
C18 disks require pre-conditioning with a water- 10 ml dichloromethane, followed by 20 ml hexane.
miscible organic solvent [18,19] or alternatively, a The 2 ml sample extract was loaded onto the column
water miscible organic solvent must be added to the using a Pasteur pipette and two 2 ml hexane rinses
sample (typically 0.5% (v/v) methanol) [20]. The of the extract flask ensured the complete transfer of
former method was used as the addition of an or- the sample. Aliphatic hydrocarbons were eluted with
ganic co-solvent may have affected the partitioning 5 ml of hexane and then the PAHs eluted with 25 ml
of the contaminants [13] as well as invalidating any of 20% dichloromethane in hexane. The cleaned ex-
subsequent DOM measurements. Acetonitrile (10 ml) tracts were reduced to 5 ml followed by the addition
was drawn through the disk and was then completely of 3 ml of acetonitrile and further reduction to about
rinsed away by washing with 500 ml Milli-Q water. 1 ml. This extract was transferred to a 1.5 ml HPLC
Immediately prior to the disk running dry, the vac- sample vial and the volume determined by weight.
uum was reduced and approximately 20 ml of the
sample introduced into the filtration reservoir. The 2.4. Instrumental analysis
filtration apparatus was then rapidly transferred to a
clean 2500 ml amber glass sample bottle (bottle 3). Dissolved organic carbon (DOC) was used as a sur-
With the reservoir full, the vacuum was increased and rogate measurement for DOM and was determined
the sample processed. Typical times of 30–45 min following APHA Method 5310 B [26] using a Shi-
were required to extract a 2500 ml sample. Two 45 ml madzu TOC 5000 analyzer utilizing high tempera-
Milli-Q rinses of the sample bottle (bottle 2) rinsed ture catalytic oxidation (HTCO, 680 ◦ C). Optical ab-
any colloidally-associated PAHs from it into the filter sorbance at 280 nm was measured using a Cary 500
reservoir. The disk was then air dried for 30 s and Scan UV-Vis near-IR spectrophotometer.
the vacuum released. A 50 ml aliquot of the C18 disk The PAHs were separated by HPLC using a Shan-
filtrate was taken for DOM analysis and the remain- don Hypersil Green PAH column (150 × 4.6 mm,
ing filtrate extracted by liquid–liquid extraction using 5 ␮m particle size, 20 ␮l injection) with gradient elu-
three successive 60 ml volumes of dichloromethane. tion (Table 1) and detected using a Jasco UV-975
The final step involved the elution of the truly dis- UV-Vis detector and a Hitachi F-1050 fluorescence
solved PAHs from the C18 disk. The filtration ap- detector [27]. Fluorescence excitation and emission
paratus including the disk was placed on the 125 ml wavelengths of 280 and 400 nm, respectively, were
glass flask and the PAHs eluted under low vacuum chosen to maximize the average PAH response.
by a two-solvent regime beginning with 10 ml ace- Compound identification (Table 2) was based on re-
tone. The use of a water miscible solvent as the first tention time matches with a diluted external PAH stan-
eluant enhances the recovery of analytes by penetrat- dard run at the beginning, after every 10 samples and
 J.N. Brown, B.M. Peake / Analytica Chimica Acta 486 (2003) 159–169 163

Table 1 relative percent difference (R.P.D.) of <10% and an

HPLC analytical parameters average relative standard deviation (R.S.D.) of 3%.
Gradient elution programa
Time (min) 0.0 10.0 30.0 40.0 41.0 50.0 2.6. Determination of method detection limits
Water (%) 50 50 0 0 50 50
(MDLs)
Acetonitrile (%) 50 50 100 100 50 50
UV detector program
The MDLs for each of the 16 PAHs in the truly
Time (min) 0.0 10.0 14.5
Wavelength (nm) 221 262 254 dissolved and colloidal phases were determined ac-
cording to APHA Method 1030 C [26]. Five samples
a Flow rate = 1.5 ml min−1 ; column temperature = 30 ◦ C. of Morrisons Creek water were processed through the
entire phase separation procedure and then four were
the end of each analytical session. Quantification was spiked at a concentration of five times the estimated
undertaken using a series of external standards. Cali- MDL for each analyte. The fifth sample was used to
bration curves were linear (R2 > 0.997) over several correct for any existing levels of PAHs in the water.
orders of magnitude (Table 2). A selection of samples The spiking level and the mean analyte recoveries over
were also sent to an independent laboratory for anal- the two phases are given in Table 3. The MDLs deter-
ysis by GC–MS to confirm the peak identifications in mined were similar to those reported in other studies
environmental matrices. for PAHs in fresh and saline water [2,20].

2.5. Quality assurance

3. Results and discussion
Small amounts of some PAHs were sometimes de-
tected in some field and procedural blanks but were The use of C18 -based techniques for investigating
always <5% of the amount measured in the samples. the partitioning of PAHs between the truly dissolved
Good agreement was obtained between the measured and colloidal phases may result in potential method-
and certified concentrations of PAHs in the reference ological artifacts arising from: (a) incomplete retention
material SRM2260 (National Institute of Standards of truly dissolved PAHs on the C18 media; (b) reten-
and Technology, Gaithersburg, USA) with an average tion of DOC on the C18 media; and (c) the dissociation

Table 2
Analyte retention times, detection method and calibration range of the 16 PAHs identified as priority pollutants by the US Environmental
Protection Agency
Compound Abbreviation Retention time (min) Calibration range (␮g l−1 ) Detector

Naphthalene NAP 6.9 0.5–800 UV-Vis

Acenaphthylene ACY 8.9 2–800 UV-Vis
Acenaphthalene ACE 12.2 5–500 Fluorescence
Fluorene FL 13.5 2–800 UV-Vis
Phenanthrene PHEN 15.7 2–800 UV-Vis
Anthracene ANT 17.6 0.5–800 UV-Vis
Fluoranthene FLR 19.4 0.5–800 Fluorescence
Pyrene PYR 20.5 0.1–800 Fluorescence
Benzo[a]anthracene BaA 24.5 0.05–400 Fluorescence
Chrysene CHY 25.4 0.5–800 Fluorescence
Benzo[b]fluoranthene BbF 28.2 0.5–800 Fluorescence
Benzo[k]fluoranthene BkF 29.5 0.01–400 Fluorescence
Benzo[a]pyrene BaP 30.5 0.01–400 Fluorescence
Dibenz[a,h]anthracene DBA 32.3 0.05–400 Fluorescence
Benzo[ghi]perylene BPY 33.0 0.05–400 Fluorescence
Indeno[1,2,3-cd]pyrene INP 34.0 1–800 UV-Vis
164 J.N. Brown, B.M. Peake / Analytica Chimica Acta 486 (2003) 159–169

Table 3
Method detection limits for PAHs in fresh water samples
Spiking level (ng l−1 ) Mean sample concentration (S.D.a ; ng l−1 ) Mean recovery (%) MDL (ng l−1 )

Truly dissolved Colloidal Truly dissolved Colloidal

NAP 1.24 2.1 (0.3) 0.52 (0.46) 105 1.2 2.1

ACY 4.5 3.3 (0.6) 2.9 (0.4) 69 2.5 1.7
ACE 13.2 9.4 (2.4) 10.8 (1.5) 77 11.1 6.9
FL 3.4 5.8 (2.5) 4.3 (1.8) 146 11.3 8.2
PHEN 4.3 4.6 (0.9) 5.2 (1.3) 114 3.9 6.1
ANT 1.2 1.8 (0.7) 1.2 (0.2) 122 3.3 0.97
FLR 0.92 1.2 (0.1) 1.2 (0.2) 130 0.34 0.85
PYR 0.16 0.25 (0.03) 0.31 (0.13) 170 0.14 0.57
BaA 0.10 0.13 (0.01) 0.14 (0.01) 130 0.03 0.06
CHY 0.51 0.54 (0.02) 0.61 (0.08) 113 0.10 0.36
BbF 0.20 0.17 (0.01) 0.18 (0.03) 117 0.03 0.14
BkF 0.03 0.01 (0.01) 0.06 (0.01) 124 0.04 0.04
BaP 0.04 0.03 (0.01) 0.04 (0.01) 82 0.04 0.02
DBA 0.09 0.10 (0.00) 0.10 (0.02) 116 0.01 0.07
BPY 0.08 0.09 (0.02) 0.10 (0.02) 126 0.07 0.08
INP 2.2 1.5 (0.4) 2.0 (0.8) 81 1.8 3.8
a Standard deviation.

of the PAH–DOC species in the presence of the C18 3.2. Retention of DOC by the C18 disks
media (acting as a competing adsorbent) [9,13,28].
Therefore, for the present method to be successful, the If some DOC is trapped by the C18 media, and
following aspects were considered. by inference, some DOC-associated PAHs, a positive
bias in the truly dissolved phase will occur when the
3.1. Blank contamination, retention and C18 media is eluted leading to a negative bias in the
elution of truly dissolved PAHs colloidal phase and a negative error in the calculated
partitioning coefficient. The adsorption of DOC on
Analysis of procedural blanks of Milli-Q water the C18 disks was assessed by passing solutions of
showed that no contamination of either PAHs or DOC known DOC concentration through the C18 disks and
was originating from the C18 disks. The retention of measuring the DOC concentrations in the filtrate by
truly dissolved PAHs on the disks was shown to be HTCO (Table 5).
from 99 to 100% at all spiking levels and volumes The mean DOC retention for solutions of AHA in
by passing Milli-Q water spiked with PAHs through Milli-Q of 6 ± 1% (mean ± 1standard deviation; n =
the C18 disks and analyzing the disk filtrate (Table 4). 4) is similar to that previously reported using C18 car-
Mass balances determined that the solvent elution tridges [11,13,28] indicating that the passage of AHA
of the truly dissolved PAHs from the C18 disks was through the disks is largely quantitative. The fresh wa-
quantitative. ter samples exhibited much higher retention values

Table 4
Method evaluation of C18 disks—retention and elution of truly dissolved PAHs
Sample volume (ml) Spiking level (ng l−1 ) Retention on C18 disks (%) Elution from C18 disks (%)

350 (n = 3) 60 100 100

350 (n = 3) 730 100 100
2500 (n = 1) 60 99 100
2500 (n = 1) 730 99 100
 J.N. Brown, B.M. Peake / Analytica Chimica Acta 486 (2003) 159–169 165

Table 5
Method evaluation of C18 disks—retention of dissolved organic carbon (DOC)
Sample n Volume (ml) Mean DOC (mg C l−1 ) Mean (%) DOC retention

AHA in Milli-Q 3 2500 18.0 5

AHA in Milli-Q 1 2500 9.8 6
St David St. 3 350 4.6 42
Water of Leith 3 2500 2.3 31
Portobello Road 3 2500 10.5 37

with up to 42% of the DOC being retained on the C18 PAHs between the colloidal (DOC-associated) and the
disks. Although the present samples involved fresh truly dissolved phases as expressed by the apparent
water, these high results are similar to those reported 
DOC-water partition coefficient Kdoc (l kg−1 C−1 ):
for saline samples [13,22]. The enhanced retention of
 [PAH]colloidal /[DOC]
the fresh water DOC on the disks may be due to high Kdoc = (1)
levels of dissolved metal cations present [29] causing [PAH]truly dissolved
the partial aggregation of the humic substances [28].
Ozretich et al. [13] have provided equations which
Also, the freshwater DOC may have a higher degree
allow the correction of this bias leading to an equa-
of hydrophobicity than the AHA and may, therefore,
tion for the corrected partition coefficient cKdoc
be adsorbed to the C18 particles to a greater extent [9].
The retention of DOC on the C18 disks was further (l kg−1 C−1 ):
investigated by measuring the absorbance at 280 nm ([PAH]colloidal × 1/(1 − FR))/[DOC]
(Abs280) of samples prior to and after passage through cKdoc =
[PAH]total − ([PAH]colloidal × 1/(1 − FR))
the disks. Absorbance in the UV region of the spec-
(2)
trum has been used previously as a relative measure
of the retention of DOC on C18 cartridges [9,28] and where FR is fraction of DOC retained on the C18 me-
can provide information on the nature of the humic dia.
substances responsible for the binding of PAHs [7,10].
The results for 10 samples taken on two occasions [total DOC] − ([C18 passed DOC]
from the Portobello Road and Water of Leith sites had −[C18 released DOC])
an average retention of DOC of 24% with a R.S.D. of FR = (3)
[total DOC]
24% based on the Abs280 measurements. The aver-
age retention based on the HTCO measurements was To illustrate the effect of correcting for FR on the
lower at 16% but exhibited a poorer reproducibility measured colloidal and truly dissolved PAH concen-
with a R.S.D. of 51%. A paired Students t-test deter- trations, the results for fresh water samples from the
mined that there was no statistical difference between Portobello Road site that were spiked with PAHs are
the mean DOC retention values as measured by the shown in Fig. 2. The decrease in the concentration
two techniques (P > 0.05). The use of the Abs280 of each of the PAHs in the truly dissolved phase is
measurements would, therefore, appear to provide an small, as is the corresponding increase in the colloidal
equivalent and more precise measure of the DOC re- phase. Fig. 3 shows the mean values of log cKdoc plot-
tention. ted against the log Kow value for each PAH for these
samples. The log Kdoc  values have also been plotted
3.3. Effect of DOC retention on the PAH to show the effect of correcting for FR. Naphthalene
concentrations and calculated partitioning constants exhibits a higher partition coefficient than would be
expected based on its log Kow value which may be
It was likely that there were PAHs associated with due to its smaller size that enables it to penetrate hy-
the DOC being retained on the C18 disks [9,28], drophobic cavities in the DOC that are not available to
leading to an underestimate of the partitioning of the larger PAHs [24,30]. With naphthalene excluded
166 J.N. Brown, B.M. Peake / Analytica Chimica Acta 486 (2003) 159–169

Fig. 2. Colloidal and truly dissolved PAH concentrations in spiked water samples (mean, n = 3) prior to, and after correction (indicated
by the prefix c) for the fraction of the DOC retained on the C18 disks (FR). Naphthalene shows the largest change in concentration.

Fig. 3. Partitioning coefficients (mean, n = 3) to DOC for spiked fresh water samples from the Portobello Road site showing the effect
of correcting for FR. NAP is excluded from the linear regression analysis.
 J.N. Brown, B.M. Peake / Analytica Chimica Acta 486 (2003) 159–169 167

Fig. 4. Difference in log cKdoc vs. log Kdoc

 as a function of log Kow .

from the dataset, there is a significant trend of increas- tween the log cKdoc and log Kdoc values would in-
ing log cKdoc values with increasing log Kow (Students crease as the log Kow values of the PAHs increase
t-test, P < 0.01) suggesting that the strength of the (Fig. 4). With naphthalene excluded from the dataset,
PAH–DOC interactions are proportional to the Kow the slope of the regression equation is statistically sig-
values of the PAHs. nificant (P < 0.01). This confirms that the DOC being
If the DOC being retained on the C18 disks was retained on the C18 disks is involved in the partition-
participating in the binding of PAHs, it would be ex- ing of PAHs, and that the corrections to the measured
pected that the larger, more hydrophobic PAHs would truly dissolved and colloidal PAH concentrations are
bind more strongly to it and that the difference be- necessary.

Table 6
DOC-water partition coefficients for PAHs binding to Aldrich Humic Acida (spiking concentration = 70 ng l−1 )
PAH log Kow b cKdoc (x, 104 l kg−1 C−1 ) Log cKdoc (l kg−1 C−1 )
Mean (n = 3) S.E.c Mean (n = 3) S.E.c

NAP 3.37 1.94 0.19 4.28 0.04

ACY 4.00 2.22 0.32 4.34 0.07
ACE 3.92 1.55 0.31 4.17 0.09
FL 4.18 –d –d –d –d
PHEN 4.57 1.96 0.15 4.29 0.03
ANT 4.54 2.58 0.09 4.41 0.02
FLR 5.22 9.87 0.49 4.99 0.02
PYR 5.18 16.1 0.31 5.21 0.01
BaA 5.91 24.5 0.9 5.39 0.02
CHY 5.86 21.9 1.4 5.34 0.03
BbF 5.80 63.6 2.6 5.80 0.02
BkF 6.00 60.2 1.2 5.78 0.01
BaP 6.04 69.9 5.9 5.84 0.03
DBA 6.75 144 6.9 6.16 0.02
BPY 6.50 132 18 6.11 0.06
INP 7.04 356 49 6.54 0.06
a DOC concentration = 9.8 mg C l−1 .
b Values from [25].
c Standard error.
d Fluorene could not be determined in the colloidal phase due to an interfering peak.
168 J.N. Brown, B.M. Peake / Analytica Chimica Acta 486 (2003) 159–169

Fig. 5. Comparison of the regression equation determined in the present study for the PAH log cKdoc values vs. log Kow for solutions of
AHA with other equations reported in the literature. The thin dashed lines represent the 95% prediction interval for the present regression
equation; C18, C18 SPE methods and FQ, fluorescence quenching methods.

3.4. Method validation method was capable of generating partitioning coeffi-

cients for the PAHs binding to AHA that are compa-
Solutions of AHA were used to demonstrate that rable to those obtained using other methodologies.
the C18 disk technique yielded partitioning coef-
ficients comparable to those obtained using other
techniques. This DOC surrogate has been used by 4. Conclusion
many researchers for studying PAH–DOC interactions
[6,13,14,28]. The observed partition coefficients mea- The use of the C18 disk-based method developed
sured in the present study (Table 6) were similar to in the present study for the determination of the in
those reported in the literature. For example, Landrum situ partitioning of PAHs between the colloidal and
et al. [31] measured a Kdoc value of (2.00 ± 0.07) × truly dissolved phases in bodies of fresh water, has
104 l kg−1 C−1 (mean ± S.E.; n = 4) for ANT using clear advantages over the previous alternative meth-
C18 cartridges. Ozretich et al. [13] reported a log cKdoc ods. In particular, the C18 disks enable larger sample
value of 5.93 ± 0.01 l kg−1 C−1 (mean ± S.E.; n = 3) volumes to be treated and hence enable much lower
for BaP to AHA using C18 cartridges. levels of PAHs to be detected in the colloidal phase.
A more thorough comparison was made by compar- The disks also require minimal preparation, have min-
ing the linear regression equation for the present data imal or no background interferences and allow the rel-
of log cKdoc = 0.77 (±0.19) log Kow + 1.07 (±1.09) atively rapid processing of large samples. Correction
(R2 = 0.95, S.E.R. = 0.18, n = 14) with those re- can be made for the adsorption of small amounts of
ported in recent reviews (Fig. 5) [14,32]. The inclusion colloidal material and DOC-associated PAHs on to the
of these other equations within the 95% prediction in- disks. The method is likely to be able to be applied
terval for the present equation shows that the C18 disk to other types of hydrophobic organic contaminants,
 J.N. Brown, B.M. Peake / Analytica Chimica Acta 486 (2003) 159–169 169

such as polychlorinated biphenyls, other organochlo- [12] K. Naes, J. Axelman, C. Naf, D. Broman, Environ. Sci.
rine compounds and pesticides which have been shown Technol. 32 (1998) 1786.
to exhibit reduced recoveries from water samples in [13] R.J. Ozretich, L.M. Smith, F.A. Roberts, Environ. Toxicol.
Chem. 14 (1995) 1261.
the presence of DOC when extracted using C18 SPE [14] H.B. Krop, P.C.M. van Noort, H.A.J. Govers, Rev. Environ.
[13,21,22]. Contam. Toxicol. 169 (2001) 1.
[15] F.D. Kopinke, A. Georgi, K. Mackenzie, Acta Hydrochim.
Hydrobiol. 28 (2000) 385.
Acknowledgements [16] J.M. Greenamoyer, S.B. Moran, Mar. Chem. 57 (1997) 217.
[17] J.L. Zhou, T.W. Fileman, S. Evans, P. Donkin, R.F.C.
Mantoura, S.J. Rowland, Mar. Pollut. Bull. 32 (1996) 599.
The authors express their gratitude to the Dunedin [18] J. Dachs, J.M. Bayona, Chemosphere 35 (1997) 1669.
City Council for material and financial support and [19] M.L. Mayer, C.F. Poole, Anal. Chim. Acta 294 (1994) 113.
the University of Otago for a Postgraduate Scholarship [20] J.C. Means, J. AOAC Int. 81 (1998) 657.
[21] J.J. Ridal, M.E. Fox, C.A. Sullivan, R.J. Maguire, A.
and Bridging Grant.
Mazumder, D.R.S. Lean, Anal. Chem. 69 (1997) 711.
[22] B. Sturm, H.D. Knauth, N. Theobald, G. Wunsch, Fresenius
J. Anal. Chem. 361 (1998) 803.
References [23] D.K. Makepeace, D.W. Smith, S.J. Stanley, Crit. Rev. Environ.
Sci. Technol. 25 (1995) 93.
[1] I. Bouloubassi, A. Saliot, Mar. Pollut. Bull. 22 (1991) 588. [24] J.N. Brown, Partitioning of Chemical Contaminants in Urban
[2] G.D. Foster, E.C. Roberts, B. Gruessner, J. Velinsky, Appl. Stormwater, Ph.D. Thesis, University of Otago, Dunedin, New
Geochem. 15 (2000) 901. Zealand, 2002.
[3] A.C. Sigleo, J.C. Means, Rev. Environ. Contam. Toxicol. 112 [25] D. Mackay, W.Y. Shiu, K.C. Ma, Illustrated Handbook of
(1990) 123. Physical–Chemical Properties and Environmental Fate for
[4] M. Haitzer, S. Hoss, W. Traunspurger, C. Steinburg, Organic Chemicals, vol. II, Lewis Publishers, Michigan, 1992.
Chemosphere 37 (1998) 1335. [26] APHA, Standard Methods for the Examination of Water and
[5] B.J. Eadie, N.R. Morehead, P.F. Landrum, Chemosphere 20 Wastewater, 20th ed., American Public Health Association,
(1990) 161. Washington, DC, 1998.
[6] T.D. Gauthier, W.R. Seitz, C.L. Grant, Environ. Sci. Technol. [27] S.A. Wise, L.C. Sander, W.E. May, J. Chromatogr. 642 (1993)
21 (1987) 243. 329.
[7] J. Peuravuori, Anal. Chim. Acta 429 (2001) 75. [28] N. Li, H.K. Lee, Anal. Chem. 72 (2000) 5272.
[8] R.M. Burgess, R.A. McKinney, W.A. Brown, J.G. Quinn, [29] L.M. Mosley, B.M. Peake, N.Z. J. Mar. Freshwater Res. 35
Environ. Sci. Technol. 30 (1996) 1923. (2001) 615.
[9] J. Kukkonen, J. Pellinen, Sci. Total Environ. 152 (1994) 19. [30] M.A. Schlautman, J.J. Morgan, Environ. Sci. Technol. 27
[10] Y.P. Chin, G.R. Aiken, K.M. Danielsen, Environ. Sci. Technol. (1993) 961.
31 (1997) 1630. [31] P.F. Landrum, M.D. Reinhold, S.R. Nihart, B.J. Eadie,
[11] P.F. Landrum, S.R. Nihart, B.J. Eadie, W.S. Gardner, Environ. Environ. Toxicol. Chem. 4 (1985) 459.
Sci. Technol. 18 (1984) 187. [32] L.P. Burkhard, Environ. Sci. Technol. 34 (2000) 4663.