Analytica Chimica Acta 589 (2007) 125–132

Microwave-assisted solvent extraction of solid matrices and subsequent

detection of pharmaceuticals and personal care products
(PPCPs) using gas chromatography–mass spectrometry
Stacie L. Rice 1 , Siddhartha Mitra ∗
Department of Geological Sciences and Environmental Studies, Binghamton University, Binghamton, NY 13902-6000, United States
Received 3 November 2006; received in revised form 20 February 2007; accepted 21 February 2007
Available online 24 February 2007

Abstract
Concentrations of pharmaceuticals and personal care products (PPCPs) in natural solids remain largely unknown. Contributing to this, is a lack
of methods permitting the simultaneous detection of the diverse, low-level contaminants present in these complex matrices. We have developed a
microwave-assisted solvent extraction (MASE)-based method targeting seven diverse PPCPs (caffeine, 17␤-estradiol, ibuprofen, ketoprofen, musk
ketone, naproxen, and triclosan) and a molecular marker for fecal waste (epicoprostanol). The method consisted of optimizing the following vari-
ables: derivatization of the polar target analytes, silica gel open column clean-up, and gas chromatographic–mass spectrometric (GC–MS) analysis
of sample extracts for analysis and detection of the compounds noted above. Testing of the method on spiked soil allowed for 89.6 ± 2.89% recovery
of three target compounds and 25.0 ± 1.93% recovery of five of the compounds. Although the latter recoveries were low, the precision across all
recoveries was high, suggesting good reproducibility in application of the method. Furthermore, we suspect that matrix effects are likely responsible
for the lower recoveries. Techniques with the exclusive incorporation of organic solvents were found inapplicable in the study of a pharmaceutical
salt, diphenhydramine HCl. Application of the developed method to sediment collected directly downstream of the effluent pipe of a wastewater
treatment plant allowed detection of ibuprofen, naproxen, ketoprofen, and epicoprostanol at ng–␮g per gram dry weight concentrations. The obser-
vation of acidic pharmaceuticals, previously believed to exhibit insignificant sorption to solid matrices, in the tested sediment samples, coupled with
application of biosolids for agricultural purposes, demonstrates the need for expanded investigation of PPCP contamination of natural solid matrices.
© 2007 Elsevier B.V. All rights reserved.

Keywords: PPCPs; MASE; Biosolids

1. Introduction The extent of exposure from contaminated matrices remains

Pharmaceuticals and personal care products (PPCPs) have
been reported in a variety of natural matrices from numerous
locations [1–5]. PPCPs include medications ranging from anal-
gesics and antibiotics to contraceptives and lipid regulators, in
addition to the active ingredients in soaps, detergents, perfumes,
and skin, hair, and dental care products [6–8].
Continuous introduction of PPCPs into the environment, mul-
tiple dispersal mechanisms, and their pharmacological activities
may result in detrimental impacts on wildlife and humans [8,9].
∗ Corresponding author. Tel.: +1 607 777 3404.
E-mail address: smitra@binghamton.edu (S. Mitra).
1 Present address: Department of Environmental and Aquatic Animal Health,
Virginia Institute of Marine Science, The College of William and Mary, P.O.
Box 1346, Gloucester Point, VA 23062-1346, United States.
largely unknown, however, studies have reported the bioaccu-
mulation of some PPCPs in lobster, clams, and human breast
milk [10]. Increased hermaphroditism in organisms exposed
to female reproductive hormones has also been observed [11].
In addition, increased bacterial resistance among colonies sub-
jected to widely used antibacterial agents has been documented
[12]. The discovery of multiple classes of PPCPs coincident in
environmental samples further necessitates the consideration of
potential interactive effects [5,8]. For instance, a mixture of 13
pharmaceuticals resulted in a 10–30% reduction in growth of
human embryonic kidney cells after 2 days of exposure in vitro,
while no effects were observed when one of the chemicals was
presented individually [13].
Wastewater treatment plant (WWTP) effluents are considered
a primary source for PPCP introduction to the environment [9].
Pharmaceuticals may enter wastewater via excretion or disposal

0003-2670/$ – see front matter © 2007 Elsevier B.V. All rights reserved.
doi:10.1016/j.aca.2007.02.051
126 S.L. Rice, S. Mitra / Analytica Chimica Acta 589 (2007) 125–132
of unused medications. Personal care products are incorporated
through washing and bathing practices [6]. During wastewater
treatment, some removal of PPCPs occurs through sorption to
sludge [6]. Over one-half of the sewage sludge generated annu-
ally in the U.S. is further stabilized, then referred to as biosolids,
and applied to agricultural fields, golf courses, and residen-
tial lawns as fertilizer/soil conditioner [8,14]. The remainder is
placed in landfills or incinerated [11]. Following biosolid appli-
cation onto land, incorporated contaminants may enter soil or be
translocated via leaching, volatilization, or transport on eroded
particles [8,10]. Some contaminants in biosolids are likely to
be available for uptake by plants, microorganisms and animals
which inhabit or feed on soil and sediment [10,11,14]. Biosolids
have high nutrient and organic carbon content; however, the
identities and levels of organic contaminants therein are largely
unknown and unregulated.
Diverse chemicals constitute the compound class known as
PPCPs and, hence, such chemicals will exhibit a variety of fates
in WWTPs and the natural environment. The incorporation of
polar functional groups in many suggests considerable water
solubility, while others are more hydrophobic or possess posi-
tively charged moieties which may lead to significant interaction
with solids. Many PPCPs contain combinations of these struc-
tural properties, complicating prediction of their behavior and
necessitating their quantification in multiple matrices [6]. The
synthetic steroid ethinylestradiol has been detected in sewage
effluent, surface waters, activated and digested sludge, and
river sediment [2,15], while the over-the-counter antihistamine
diphenhydramine was found in sediment at concentrations that
are believed to exceed those in aqueous matrices by three orders
of magnitude [3].
Currently, our understanding of the fate of PPCPs is hindered
by a lack of quantitative data concerning PPCP presence in the
environment [9,10]. Only recently has attention been given to
the potential contamination of biosolids and amended soils with
WWTP-derived organic compounds. Moreover, data concerning
mixtures of PPCPs are limited, although individual subclasses
(e.g. antibacterial agents, synthetic musks, and synthetic hor-
mones) have been detected in solid samples [1,2,8,11,16].
Collection of these data is impeded by the scarcity of analytical
methods capable of reliably detecting diverse PPCPs at trace
concentrations in complex matrices [8,17]. Published methods
are typically specific to a single contaminant or PPCP class
[2,3,7,12,15,16]. For efficient and timely determination of the
variety of PPCPs present in the environment and subsequent
study of interactive effects on exposed organisms, multi-residue
analytical methods must be established.
The main objective of this research was to develop a time-
and cost-effective analytical method for the simultaneous detec-
tion and quantification of eight structurally diverse PPCPs and
a representative molecular marker for fecal waste in solid envi-
ronmental matrices, i.e. soil and sediment. Microwave-assisted
solvent extraction (MASE) was chosen for the separation of the
target PPCPs from solid matrices, as it has an extraction effi-
ciency comparable to traditional Soxhlet methods, but requires
less time and solvent volume [7,18]. Gas chromotagraphy–mass
spectrometry (GC–MS) was used as the quantification method.
The final multi-residue PPCP method was applied to both
standard-amended soil samples and to natural sediment samples
collected directly outside a WWTP effluent pipe.
2. Experimental
2.1. Chemicals
Standards for the target compounds were purchased from
Sigma Chemical Company and Sigma–Aldrich, Inc. (St. Louis,
MO) or Aldrich Chemical Company (Milwaukee, WI). The
chemical standards and their purity levels (where available)
are: caffeine (anhydrous), diphenhydramine HCl, 17␤-estradiol
(>98%), epicoprostanol (>95%), ibuprofen (>98%), ketoprofen,
musk ketone (>98%), naproxen, and triclosan (>97%). Standard
solutions were prepared in acetone (GC grade, EMD Chemicals,
Inc., Gibbstown, NJ). Calibration standards were prepared by
diluting mixtures of the nine individual standards with acetone.
Additional solvents included methylene chloride (LC-GC grade,
EMD Chemicals, Inc.), methanol (GC grade, Burdick and Jack-
son, Muskegon, MI), hexane (GC grade, VWR International,
West Chester, PA), and ethyl acetate (GC grade, Omni Solv,
EMD Scientific, San Diego, CA).
All glassware used during standard preparation and experi-
mentation was cleaned by ashing at 450 ◦ C for 4 h prior to use.
The wearing of fragrance materials was avoided by laboratory
members during experimentation.
Derivatizing agents were purchased from Pierce (Rock-
ford, IL): N-methyl-N trifluoroacetamine (MSTFA)+1%
trimethylchlorosilane (TMCS), N-tremethylsilyimidazole
(TMSI), Tri-Sil TBT (TMSI: BSA (N,O-bis[trimethylsilyl]
acetamide): TMCS 3:3:2) and N,O-bis[trimethylsilyl]tri-
fluoroacetamide (BSTFA) + 1% TMCS. Pyridine (EMD
Chemicals, Inc.) was used as the derivatization solvent.
2.2. Detection using GC–MS
Target PPCPs were detected and quantified using a Shimadzu
QP5050A GC–MS system equipped with a Shimadzu AOC-20i
autosampler and a Restek (Columbia, MD) XTI-5 (bonded 5%
phenyl-extended temperature and inertness) GC column (30 m,
0.25 mm i.d., 0.25 ␮m film thickness). Helium was used as the
carrier gas, with a column flow rate of 1 mL min−1 . All analyses
incorporated splitless injection and electron impact ionization.
The interface temperature between the GC and the MS was
maintained at 310 ◦ C.
The PPCPs chosen for method development were based on
popular consumption and diversity in molecular structure. Epi-
coprostanol was added as a molecular marker for fecal waste
(Table 1; Fig. 1). The target compounds containing hydroxyl or
carboxyl functional groups were derivatized prior to GC injec-
tion.
Derivatization involves the optimization of five variables: the
derivatizing agent, derivatization solvent, reaction temperature,
duration of reaction, and the conditions under which the deriva-
tization solution is dried subsequent to the reaction (prior to
GC–MS analysis). All of these variables were addressed to arrive
 S.L. Rice, S. Mitra / Analytica Chimica Acta 589 (2007) 125–132
Table 1
Target compounds selected for development of Mixed PPCP Method
Target compound Function
Caffeine Stimulant
Diphenhydramine hydrochloride Anti-histamine
Epicoprostanol Molecular marker for fecal waste (steroid)
17␤-Estradiol Female reproductive hormone
Ibuprofen Anti-inflammatory; analgesic
Ketoprofen Anti-inflammatory; analgesic
Musk ketone Synthetic fragrance
Naproxen Anti-inflammatory; analgesic
Triclosan Anti-bacterial agent
at a derivatization method which: (1) permitted detection of the
six compounds containing polar function groups with adequate
signal to noise ratios (S/N), (2) encouraged complete derivati-
zation (<1% contribution of the underivatized or incompletely
derivatized form of a compound to its total chromatographic
peak area), and (3) was time-efficient. Four different derivatiz-
ing agents were tested on a mixed standard solution of the target
PPCPs (1:1, v/v) with various reaction times and temperatures:
MSTFA + 1%TMCS for 15 and 30 min at 23 and 60 ◦ C; TMSI
for 20, 30, and 60 min at 23 and 70 ◦ C; Tri-Sil TBT for 180 and
360 min at 60 ◦ C; BSTFA + 1%TMCS for 10 and 20 min at 60
and 70 ◦ C.
Following selection of a derivatizing agent, a more exten-
sive procedure was developed to permit the removal of excess
BSTFA prior to GC–MS injection, as derivatizing agents may
damage GC columns. Aliquots of a mixed standard solution
were dried under N2 in conical- or round-bottom test tubes in
a heated water bath (45–50 ◦ C) or sand pit (45–55 ◦ C). Residue
was recovered in pyridine, ethyl acetate, or methylene chloride
(0.4 or 0.5 mL) and derivatizing agent (0.2 or 0.25 mL). The
tubes were then capped, shaken, and heated for 20 or 45 min at
70 ◦ C. After cooling for 5 min, the solutions were again dried
Fig. 1. Molecular structures of target compounds chosen for development of
Mixed PPCP Method.
127
as above. Residues were recovered in small aliquots of hexane
and transferred to 1.5 mL amber glass vials for final volumes of
∼1 mL. Those PPCPs not having polar functional groups were
also exposed to derivatization conditions in order to determine
any effect on their detection and quantification.
The GC temperature program was varied to achieve adequate
resolution of all nine target compounds within a reasonable anal-
ysis time. Detection by the MS was optimized by developing a
selected ion monitoring (SIM) program using the three most
prominent mass:charge ratios (m/z values) associated with each
target compound, as determined by running the GC–MS in scan
mode.
As isotopically-labeled PPCP standards are rare and costly,
external calibration was used for quantification of injected
masses. Four mixed standard solutions of increasing concen-
tration levels were injected before and after derivatization
for average injected masses of 2.53 ± 0.148, 17.6 ± 1.29,
46.3 ± 3.87, and 65.5 ± 5.14 ng for each of the target com-
pounds. Chromatographic peak areas were quantified using
Shimadzu GCMS SolutionsTM Software. The reproducibility of
instrument response was assessed by injecting three aliquots of
each PPCP calibration standard and calculating the relative stan-
dard deviation of the resultant peak areas for each compound at
each calibration level.
2.3. Extraction of PPCPs from spiked natural soil
Preliminary testing was completed to ascertain which extrac-
tion solvent(s)/cocktail(s) resulted in greatest recovery of the
target compounds from a natural solid matrix. For this purpose, a
soil sample was collected from a residential yard, dried in a 50 ◦ C
oven, and homogenized using a laboratory blender. Soil subsam-
ples (∼3 g) were added to test tubes and spiked with 0.5 mL of
mixed PPCP standard such that the final spiked concentration
was ∼15 ␮g of target analyte per gram dry weight (gdw) of
sediment. The tubes were then set vertically on a rocking plat-
form for 22 h. This was done to distribute the spiked standards
evenly across the soil as the carrier solvent evaporated. Soil
samples were then extracted by adding 10 mL aliquots of sol-
vent and replacing the tubes on the rocking platform for 1 h, after
which samples were centrifuged and supernatant was collected.
Two subsequent extractions were carried out, each time with
an additional 10 mL aliquot of solvent. For each sample, super-
natants from all successive extraction steps were combined for
further processing. Four extraction solvent regimens were tested
in order to determine which resulted in maximum recovery of
the spiked contaminants. In three cases, two extractions were
completed with methylene chloride and a third with methanol,
acetone, or hexane. In the final test, one extraction was done with
each of hexane, methylene chloride, and acetone. Final details
of these spiked recovery experiments are provided in Section 3
below.
Extracts were concentrated to ∼1 mL using a rotary evapora-
tor (Buchi R-200, Brinkmann–Eppendorf Scientific, Westbury,
NY) and then split into two, approximately equal volumes. One
fraction was subjected to the optimized derivatization procedure
noted above for detection of the six polar target compounds.
128 S.L. Rice, S. Mitra / Analytica Chimica Acta 589 (2007) 125–132
The second fraction was reserved for detection of the three non-
polar PPCPs not requiring derivatization. This latter extract was
dried under N2 , and recovered in ∼1 mL of hexane, and trans-
ferred to an amber glass GC vial. All fractions were subsequently
subjected to GC-MS analysis.
Upon selection of the optimal solvent cocktail regime, the
extraction process was transferred to MASE in order to enhance
recoveries of the target compounds. An Ethos E Touch Control
Microwave Solvent Extraction Labstation (Milestone; Monroe,
CT) was used in our experiments. Power was held constant at
800 W.
In order to efficiently extract target PPCPs from solid samples
using MASE, the ideal extraction solvent should be selected and
extraction temperature and duration should be optimized [19].
In the case of our method development, the solvent regime was
selected based on high solvent polarity (i.e. dipole moment) (see
Section 3) and the significant correlation between spiked recov-
eries of our target compounds via MASE and via shake flask
extraction (r2 = 0.8), the latter suggesting that the solvent cock-
tail was optimal for both techniques. We selected an extraction
temperature of 115 ◦ C as suggested in the United States Envi-
ronmental Protection Agency’s Standard Operating Procedure
for Microwave-Assisted Extraction of Soils and Sediments for
trace organic compounds [20]. Another recent study considering
MASE of endocrine disrupting compounds in river sediments
noted that there was not a statistically significant difference
in extraction efficiency for many compounds when comparing
microwave temperatures between 90 and 130 ◦ C [19]. Thus, the
use of 115 ◦ C as the optimal temperature for extraction, was
quite justifiable.
In the case of our experiments, the microwave temperature
was ramped from room temperature to 115 ◦ C in 8 min, with
the final temperature held for 15 min. This extraction duration
(15 min) was also suggested to be optimal in another study
[19]. The microwave extractor available to us was not equipped
with a pressure sensor, thus preventing manipulations of extrac-
tion pressure. For organic solvent extraction, pressure buildup
occurs as a result of the solvent system vapor pressure at
the elevated temperature [20]. Indeed, pressure is not a criti-
cal parameter for organic solvent-based extraction techniques
[21,22]. Each soil sample was sequentially extracted in a 50 mL
Pyrex test tube with a ground glass pennyhead stopper using
three 20 mL aliquots of the previously selected solvent mix-
ture; thus, the microwave extraction used was a closed system
apparatus.
After microwave extraction, the sample tubes were cen-
trifuged at 1000 rpm for 5 min. Fresh cocktail was added and the
MASE process repeated twice more, after which supernatants
of sequential extractions were combined. Extracts were con-
centrated, fractionated, and analyzed as described previously to
determine the recovery of spiked PPCPs.
Extraction of unspiked soil was carried out alongside those
samples spiked with standard solutions in order to serve as a lab-
oratory blank. In all testing, detected quantities of PPCPs were
deemed reportable if the mass of a given compound was greater
than three times its mass found in the associated laboratory
blank.
2.4. Purification of sample extracts
An initial clean-up procedure was optimized with respect
to standard solutions of compounds in the absence of natural
organic matter. Pasteur pipettes were filled with pre-extracted
silica gel (1.11 ± 0.01 g) and further packed with anhy-
drous sodium sulfate (∼0.5 cm) and activated (2N HCl)
copper granules (∼0.5 cm) in hexane to serve as clean-up
columns.
Qualitative tests were first completed on both derivatized
and underivatized aliquots of mixed PPCP standard in order to
determine which solvent(s)/cocktail(s) most effectively eluted
the target PPCP compounds from the silica micro-columns.
These standards were added to prepared columns, followed
by three hexane rinses (∼1 mL each) of the respective con-
tainers in which the standards had been stored. Rinses were
followed by sequential elution of the columns with 4 mL each
of the following: hexane, hexane:methylene chloride (1:1, v/v),
methylene chloride, methylene chloride:acetone (1:1, v/v), ace-
tone, acetone:methanol (1:1, v/v), and methanol, with each of
the resultant fractions collected separately to yield a total of
seven fractions. Each fraction was dried under N2 , reconsti-
tuted in a final volume of ∼1 mL hexane, and analyzed using
the optimized GC–MS method. The purpose of this initial
experiment was to determine which solvent(s)/cocktail(s) frac-
tions eluted the largest percentage of the target compounds.
Once these fractions were identified, additional application
of the method involved combining those fractions during
the final analytical procedure. Solvent flow was enhanced
by the application of gentle pressure using a pipette bulb
[23].
2.5. Application of Mixed PPCP Method to spiked soil
A soil sample collected near a residential area was processed
and spiked in triplicate with mixed PPCP standard as described
above. Subsamples were analyzed using the fully developed
Mixed PPCP Method (Fig. 2).
2.6. Application of Mixed PPCP Method to natural
sediment
Surface sediment grab samples were collected on August 18,
2004 and April 25, 2005 from the top 5 to 10 cm of sediment
located along the shore of Lake Erie adjacent to the effluent
pipe of a WWTP serving a town in upstate New York (popu-
lation ∼13,000). The pipe is located in a public park, and the
discharge forms a several meter-long stream path before enter-
ing Lake Erie. In April, effluent and storm run-off filled the
channel created by historical sewage effluent pressure. During
the August field sampling, the channel leading to the lake was
visible, but not submerged.
Samples were collected using pre-rinsed (acetone) metal
spatulas and stored frozen in amber glass jars at 4 ◦ C, after which
they were thawed, dried in a 50 ◦ C oven, and homogenized using
a laboratory blender. The Mixed PPCP Method was then applied
to duplicate samples from each collection date.
 S.L. Rice, S. Mitra / Analytica Chimica Acta 589 (2007) 125–132 129

Fig. 2. Summary of Mixed PPCP Method.

3. Results and discussion
3.1. Optimization of GC–MS conditions
GC–MS permitted baseline detection of the nine target
compounds, following derivatization of the six compounds
containing polar functional groups, within 22 min (Table 2).
The oven temperature was maintained at 50 ◦ C for 1 min,
then ramped at 25 ◦ C min−1 to 150 ◦ C, 8 ◦ C min−1 to 204 ◦ C,
4 ◦ C min−1 to 212 ◦ C, 10 ◦ C min−1 to 240 ◦ C, and finally
20 ◦ C min−1 to 310 ◦ C. The final temperature was then held
for 13 min in order to ensure full elution of the sample from
the GC column, resulting in an overall, per-sample run time
of 33 min. The corresponding pressure program involved
holding at 52.8 kPa for 1 min, then ramping 9.7 kPa min−1
to 91.5 kPa, 3.0 kPa min−1 to 112 kPa, 1.5 kPa min−1 to
115 kPa, 3.6 kPa min−1 to 125.5 kPa, and 7.4 kPa min−1
to 151.4 kPa, with the final pressure maintained for
13 min.
The SIM method for MS detection involved the division of
analyses into three windows of nine m/z values each, with three
m/z values per target compound (Table 2).
In SIM mode, the S/N values for instrument detection limits
ranged from 5 to 25. Instrument detection limits were: caffeine,
0.11 ng; diphenhydramine HCl, 0.19 ng; 17␤-estradiol, 0.12 ng;
epicoprostanol, 0.37 ng; ibuprofen, 0.02 ng; ketoprofen, 2.51 ng;
musk ketone, 0.26 ng; naproxen, 0.02 ng; triclosan, 0.10 ng.
Linear external calibration curves were obtained
between mass and peak area for all target compounds
(r2 = 0.972 ± 0.026). For all derivatized PPCPs, the standard
deviation in chromatographic peak area over replicate injections
(n = 3) at all calibration levels was <10%. It was <13% for
underivatized targets, with one exception (19% for the third
highest calibration point of caffeine).

Table 2
Details of selected ion monitoring (SIM) program for target compounds
Target compound Retention time (min) Target m/z Qualifier m/z 1 Qualifier m/z 2

Ibuprofen 9.092 160.25 161.2 117.2

Caffeine 11.78 194.15 109.15 82.1
Diphenhydramine hydrochloride 12.14 58.1 165.2 152.2
Musk ketone 13.37 279.25 294.15 128.15
Naproxen 14.56 185.15 243.2 302.3
Triclosan 15.17 200.05 347.3 109.15a
Ketoprofen 16.02 282.35 105.15 311.25
17␤-Estradiol 19.54 285.35 129.15 416.5
Epicoprostanol 21.28 215.25 107.2 216.3
a 109.15 was found to be the fourth most abundant m/z specific to triclosan. It was used in place of the third most abundant value, 185.15, as this latter m/z is the

target for naproxen.

130 S.L. Rice, S. Mitra / Analytica Chimica Acta 589 (2007) 125–132
3.2. Optimization of derivatization
Comparing the four derivatizing agents, it was found that
reaction with MSTFA + 1%TMCS permitted the GC–MS detec-
tion of all target compounds, although S/N was low. Similarly,
S/N was relatively low for ibuprofen, naproxen, triclosan, and
17␤-estradiol after derivatization with Tri-Sil TBT, while epico-
prostanol was not detected at all. None of the polar compounds
were observed after reaction with TMSI, with the exception of
ibuprofen (after heating for 60 min at 23 ◦ C). Derivatization with
BSTFA + 1%TMCS proved most effective, allowing all polar
compounds to be readily detected.
The derivatization procedure was most time-efficient when
conducted in a round-bottom versus a conical-bottom test tube,
most likely due to the broader surface area exposed to the N2 .
Drying in a sand pit rather than a water bath was preferred due to
the high affinity of the BSTFA derivatizing agent for water. Pyri-
dine was found to be the most effective derivatization solvent,
as ethyl acetate and methylene chloride resulted in incomplete
derivatization and non-linear calibration curves, respectively.
A 2:1 (v/v) ratio of pyridine:BSTFA was used, with 0.5 mL:
0.25 mL resulting in complete derivatization of mixed standard
solutions across the entire calibration curve range. The optimal
heating conditions for derivatization were 70 ◦ C for 20 min.
When applying the derivatization procedure to individual and
mixed standard solutions of all nine target PPCPs, it was found
to affect two of the three non-polar PPCP compounds. Injecting
standards of musk ketone following exposure to derivatization
conditions yielded a second, yet unidentified peak (spectra not
shown). As musk ketone does not contain a readily derivatiz-
able polar functional group, silylation is not expected. The harsh
chemicals or high temperature of the derivatization procedure,
however, may result in the thermal breakdown or transformation
of the parent compound. Further tests demonstrated that linear-
ity in external calibration was compromised for musk ketone,
as well as diphenhydramine HCl, following exposure to deriva-
tization conditions. It is for these reasons that sample extracts
were split prior to GC–MS analysis, with one-half analyzed for
caffeine, diphenhydramine HCl, and musk ketone and the other
derivatized to analyze for the remaining target compounds. A
similar fractionation procedure is advised in future studies of
PPCP mixtures which incorporate neutral and acidic analytes to
be identified using GC–MS.
3.3. Extraction of natural soil spiked with PPCP mixture
In preliminary testing of extraction solvent regimens using
a simple shake-flask procedure, it was observed that 2:1 (v/v)
methylene chloride:methanol resulted in the optimum recov-
ery of ibuprofen, naproxen, ketoprofen, 17␤-estradiol, caffeine,
and diphenhydramine HCl and substantial recovery of triclosan,
epicoprostanol, and musk ketone (data not shown). Selection
and use of a cocktail of methylene chloride and methanol in
microwave extraction is justifiable for the following reasons.
Microwave extraction is based on using microwave energy
to heat solvents that are in contact with solid samples and to
partition compounds of interest into the solvent. The microwave
extraction process becomes increasingly efficient with a sol-
vent’s dipole moment [24], as a significant dipole moment
ensures that there will be sufficient heating to extract compounds
from a solid matrix. At 25 ◦ C, methanol has a dipole moment
of 2.87 and methylene chloride has a dipole moment of 1.14
[24], rendering a cocktail of these to be heated considerably by
microwave energy.
The application of MASE with this solvent cocktail to spiked
samples without a subsequent clean-up step prior to GC–MS
injection resulted in an average recovery of 62.1 ± 22.5% for
seven of the target compounds. Caffeine recovery was 2.73%
and diphenhydramine HCl was not recovered. These results
may cause concern that perhaps some of the PPCP compounds
are unstable and breakdown during MASE. To address this
concern, we spiked solvent-rinsed, clean sand with a mixed
standard solution of the target compounds and conducted a
comparison of Soxhlet extraction (refluxed for 24 h) versus
our MASE extraction using methylene chloride:methanol (2:1
v/v). This comparison yielded no significant difference in PPCP
peak areas recovered by the two procedures (t-test; p = 0.35).
Ibuprofen, musk ketone, naproxen, triclosan, ketoprofen, and
epicoprostanol were all recovered to a greater extent via MASE.
In contrast, caffeine, diphenhydramine HCl, and 17␤-estradiol
were recovered to a greater extent via Soxhlet extraction. The
lack of systematically lower recoveries using MASE compared
to Soxhlet extraction suggests that these particular PPCPs are
probably not degrading as a result of MASE.
3.4. Purification/clean-up of sample extracts
Each of the seven individual fractions of solvent/cocktail
eluted from the clean-up columns during preliminary testing
was analyzed individually using GC–MS. Those fractions which
together contained >95% of the total recovered mass of a given
target were incorporated into the finalized clean-up procedures
(Fig. 3). For PPCP compounds requiring derivatization, elution
of pipette columns with 24 mL of a hexane:methylene chlo-
ride:acetone cocktail (1:4:1, v/v/v) yielded the highest recovery.
For the compounds not requiring derivatization (e.g. caffeine
and musk ketone), the optimal elution regimen was the sequen-
tial elution with 8 mL each of methylene chloride, methylene
chloride:acetone (1:1, v/v), and acetone (these three fractions
were then combined for further analyses). Diphenhydramine
HCl did not elute as readily from the columns as caffeine and
musk ketone, thereby requiring that an additional 12 mL of
acetone:methanol cocktail (1:1, v/v) be applied. The clean-up
procedures for the derivatized and underivatized PPCP standard
solutions resulted in 69.9 ± 15.8% recovery of the spiked mass
of each of eight of the target compounds. Diphenhydramine HCl
could only be recovered to a maximum of 6% of its spiked value.
3.5. Application of Mixed PPCP Method to spiked soil
The Mixed PPCP Method was applied to triplicate, target
compound-amended soil samples to quantify how much of each
could be recovered from a matrix containing natural organic
material (Fig. 4). Highest recoveries (89.6 ± 2.89%) were
 S.L. Rice, S. Mitra / Analytica Chimica Acta 589 (2007) 125–132
Fig. 3. Solvent(s)/cocktail(s) used in the elution of silica micropipette clean-up columns which accounted for >95% of the total recovered mass of each target
compound. * Those compounds not subjected to derivatization prior to clean-up procedure.
achieved for musk ketone, triclosan, and 17␤-estradiol. Lower
percentages (25.0 ± 1.93%) of ibuprofen, caffeine, naproxen,
ketoprofen, and epicoprostanol were recovered. Lowest recov-
eries were observed for diphenhydramine HCl (1.57 ± 0.486%).
Contributions from unspiked samples, serving as blanks, were
negligible for all PPCPs.
Reduced recoveries in comparison to those achieved in the
individual clean-up step may result from the development of
the elution procedures in the absence of matrix effects. Matrix
components may interfere with analyte detection by preventing
complete elution of target compounds within the allotted elution
solvent volume. Soil and sewage sludge are rich in complex
natural organic matter. In retrospect, one approach to testing
for such matrix effects would be to conduct spiked recovery
experiments on soils or sewage sludge in which much of the
natural organic matter is extracted out. We did not attempt to
conduct such analyses to verify matrix interferences.
Other studies which investigate mixtures of diverse organic
compounds in environmental matrices similarly report broad
ranges of recovery efficiencies across target compounds [25,26].
Nonetheless, the Mixed PPCP Method presented here pro-
vides a relatively simple, time-efficient means of simultaneously
Fig. 4. Recovery of target compounds spiked into natural soil using developed
Mixed PPCP Method. Error bars represent the standard deviation across triplicate
analyses. Diph HCl, diphenhydramine hydrochloride.
131
determining whether solid matrices are contaminated by seven
diverse PPCPs and a molecular marker of fecal waste. Although
some of the compounds tested yielded low recoveries using the
complete extraction process, the precision in the recoveries was
high, indicating that our extraction procedure is reproducible and
thus practically useful, especially if one corrects for recoveries.
Those methods which have been used to individually investi-
gate estrogens [2,15], triclosan [7], and fragrance materials [4] in
solid matrices have incorporated extraction with various organic
solvent regimens, demonstrating the efficacy of organic solvents
in recovering acidic and neutral organic contaminants from solid
matrices. In this study, a PPCP formulated as a salt, diphenhy-
dramine HCl, was also studied. The binding of diphenhydramine
HCl to solid matrices seems not to be effectively interrupted
by the introduction of organic solvents. One published method
specifically targeting diphenhydramine used water/acetonitrile
as the coextraction solvent, yielding an extraction recovery of
75 ± 8% [3]. This indicates that PPCP salts may be most effec-
tively recovered by utilizing their tendency to dissociate in
water. However, conducting such an aqueous extraction would
release an overwhelming amount and variety of other non-target
polar organic compounds which would presumably interfere
with isolation and chromatographic separation of our target
PPCPs.
3.6. Application of Mixed PPCP Method to natural
sediment
Using our MASE-based Mixed PPCP Method, four of the
target compounds were detected in natural sediment samples
at ng gdw−1 to ␮g gdw−1 concentrations (Fig. 5). Reported
concentrations are conservative estimates, as application of
the Mixed PPCP Method resulted in <50% recovery of these
target compounds from spiked soil samples. Given the prelim-
inary nature of our sampling, several reasons may exist for
the observed profiles. For example, seasonal variations in the
observed PPCP concentrations in these sewage-influenced sedi-
ments may be the result of differential loading to the WWTP; or,
reduced contaminant levels in sediment collected in April may
be the result of sediment scouring and dilution due to increased
local storm run-off during the spring sampling time.
132 S.L. Rice, S. Mitra / Analytica Chimica Acta 589 (2007) 125–132
Fig. 5. Target compounds detected in natural sediment using Mixed PPCP
Method. Error bars represent the standard deviation across duplicate analyses.
Diph HCl, diphenhydramine hydrochloride; ND, not detected.
The presence of epicoprostanol in the sediment indicates the
input of fecal waste into the sediment [27]. This is most likely
due to the WWTP effluent, although local wildlife may also
contribute some portion of this compound. Detectable levels
of ibuprofen, naproxen, and ketoprofen discharged from the
WWTP may result from widespread use of anti-inflammatory
and analgesic pharmaceuticals in numerous medical situations,
compounded by the over-the-counter availability of ibuprofen
and naproxen.
In a recent sorption study, ibuprofen was reported to have neg-
ligible affinity for a sediment matrix due to its acidity, and the
majority of this contaminant is said to partition into the aqueous
phase [26]. The detection of ibuprofen, as well as two additional
pharmaceuticals having carboxyl functional groups, in the sed-
iment analyzed during this study reinforces the importance of
testing multiple natural matrices in order to obtain an accurate
depiction of contaminant abundance in the environment.
In summary, we have developed a method for isolating a
number of PPCP compounds in natural solid samples. The
results shown here suggest that some of the chemicals exhibit
higher extraction efficiencies (e.g. 17␤-estradiol, musk ketone,
triclosan) than others (diphenhydramine HCl, epicoprostanol,
caffeine, ibuprofen, ketoprofen, naproxen). Based on their chem-
ical structures (Fig. 1), the differences in recoveries cannot be
attributed to polarity or compound class. Thus, we suspect that
matrix effects may indeed have limited our ability to more effi-
ciently extract some of our target compounds from natural solids.
Another example of this is based on the observation that the
trends in the yields for the spiked recovery experiments (Fig. 4)
in no way resemble the trends in the overall extraction of the sam-
ple collected immediately downstream of the sewage effluent
pipe (Fig. 5). If a physico-chemical variable were truly respon-
sible for lower extraction efficiencies, then we would expect a
greater correlation in observed trends between Figs. 4 and 5.
We did not explore the geochemistry of the sewage solids in
detail as that was not the objective of this study. However,
quantifying organic carbon abundance and composition of solid
phase samples would be the next logical step in constraining
the variables responsible for lower recoveries of certain PPCP
compounds using this procedure.
Acknowledgments
The authors are grateful for the generous support of the New
York State Great Lakes Protection Fund, the Binghamton Uni-
versity Research Foundation, the Harpur College Dean’s Office,
the American Chemical Society Petroleum Research Fund, and
the Garden Club of America.
References
[1] E.M. Golet, A. Strehler, A.C. Alder, W. Giger, Anal. Chem. 74 (2002)
5455.
[2] T.A. Ternes, H. Andersen, D. Gilberg, M. Bonerz, Anal. Chem. 74 (2002)
3498.
[3] I. Ferrer, C.E. Heine, E.M. Thurman, Anal. Chem. 76 (2004) 1437.
[4] A.M. Peck, K.C. Hornbuckle, Environ. Sci. Technol. 38 (2004) 367.
[5] D.W. Kolpin, E.T. Furlong, M.T. Meyer, E.M. Thurman, S.D. Zaugg, L.B.
Barber, H.T. Buxton, Environ. Sci. Technol. 36 (2002) 1202.
[6] T.A. Ternes, A. Joss, H. Siegrist, Environ. Sci. Technol. (2004) 392A.
[7] S. Morales, P. Canosa, I. Rodriguez, E. Rubi, R. Cela, J. Chromatogr. A
1082 (2005) 128.
[8] M.J. La Guardia, R.C. Hale, E. Harvey, E.O. Bush, T. Matteson Mainor,
M.O. Gaylor, J. Res. Sci. Technol. 1 (2004) 111.
[9] V.L. Cunningham, M. Buzby, T. Hutchinson, F. Mastrocco, N. Parke, N.
Roden, Environ. Sci. Technol. A-Pages Mag. 40 (2006) 3456.
[10] A. Difrancesco, P.C. Chiu, L.J. Standley, H.E. Allen, D.T. Salvito, Environ.
Sci. Technol. 38 (2004) 194.
[11] M. Kuster, M. Lopez de Alda, D. Barcelo, Trends Anal. Chem. 23 (2004)
790.
[12] M.D. Prat, D. Ramil, R. Compano, J.A. Hernandez-Arteseros, M. Grana-
dos, Anal. Chim. Acta 567 (2006) 229.
[13] L. Thrall, Environ. Sci. Technol.: Environ. News (2006) 2072.
[14] R. Renner, Environ. Sci. Technol.: Environ. News 34 (2000) 242A.
[15] R.L. Gomes, E. Avcioglu, M. Scrimshaw, J. Lester, Trends Anal. Chem.
23 (2004) 737.
[16] E.M. Golet, I. Xifra, H. Siegrist, A.C. Alder, W. Giger, Environ. Sci. Tech-
nol. 37 (2003) 3243.
[17] S. Morales-Munoz, J.L. Luque-Garcia, M.J. Ramos, A. Fernandez-Alba,
M.D.L. de Castro, Anal. Chim. Acta 552 (2005) 50.
[18] A.B. Prevot, M. Gulmini, V. Zelano, E. Pamauro, Anal. Chem. 73 (2001)
3790.
[19] R. Liu, J.L. Zhou, A. Wilding, J. Chromatogr. A 1038 (2004) 19.
[20] US EPA, Standard Method # 3546, Microwave Extraction, Washington DC,
2000, p. 3546.1.
[21] V. Camel, Trends Anal. Chem. 19 (2000) 229.
[22] S. Jayaraman, R.J. Pruell, R. McKinney, Chemosphere 44 (2001) 181.
[23] K.O. Buessler, O. Buesselera, J.E. Bauer, R.F. Chen, T.I. Eglinton, O.
Gustafsson, W. Landing, K. Mopper, S.B. Moran, P.H. Santsch, R. Ver-
nonClark, M. Wells, Mar. Chem. 55 (1996) 1.
[24] L. Jassie, R. Revesz, T. Kierstead, E. Hasty, S. Matz, in: H.M. Kingston, S.J.
Haswell (Eds.), Microwave-Assisted Solvent Extraction, in Microwave-
Enhanced Chemistry, American Chemical Society, Washington, DC, 1997,
p. 569.
[25] G.R. Boyd, H. Reemtsma, D.A. Grimm, S. Mitra, Sci. Total Environ. (2003)
135.
[26] D. Loffler, J. Rombke, M. Meller, T.A. Ternes, Environ. Sci. Technol. 39
(2005) 5209.
[27] L.J. Standley, L.A. Kaplan, D.M. Smith, Environ. Sci. Technol. 34 (2000)
3124.