American Journal of Anal yt ical Chemistry, 2011, 2, 783-794
doi:10.4236/ajac.2011.27090 Published Online November 2011 (
Copyright © 2011 SciRes. AJAC
Multifactorial Optimization Approach for Determination
of Polycyclic Aromatic Hydrocarbons in Sea Sediments of
Turkish Mediterranean Coast
Semra G. Tuncel*, Tansel Topal
Middle East Technical University, Chemistry Department, Ankara, Turkey
E-mail: *
Received April 20, 2011; revised June 17, 2011; accepted July 1, 2011
Present study aims the optimisation and validation of the extraction procedures for the determination of poly-
cyclic aromatic hydrocarbons (PAHs) in sediment samples. As analytical techniques, gas chromatography-
mass detector (GC-MS) and gas chromatography-flame ionization detector (GC-FID). The optimized meth-
ods were soxhlet extraction, ultrasonic bath extraction and solid phase micro extraction (SPME). The facto-
rial design was used in order to search out the main factors affecting extraction efficiencies. The best extrac-
tion method was chosen as ultrasonic bath extraction and optimum values for main factors were selected for
the development of this extraction method. The optimised methods were validated to confirm the reliability.
Percent errors were in between 1.40% - 25.0%. Relative standard deviation was less than 1% and limit of
detection changed from 0.006 to 0.028 mg/L. The accuracy of the method was verified by analyzing National
Institute of Standards and Technology, Standard Reference Material 1597a (NIST SRM 1597a). The selec-
tivity, accuracy and precision were quite adequate for the determination of PAHs in sediment samples. The
best extraction and analysis methods were then applied to determine 16 PAHs in sea sediments from
Ölüdeniz Lagoon, of Turkish Meditreanean coast and marine sediment, SRM 1941b. The highest observed
cocentration for PAHs was 0.620 mg·kg –1 for Acenaphtene.Total sum of the concentrations for all the ob-
served PAHs was1.854 mg·kg–1.
Keywords: PAH, Sediment, Response Surface Methodology, Pareto Chart, Chemometrics
1. Introduction
Polycyclic aromatic hydrocarbons (PAHs) which are
commonly found in the different compartment of the
environment including drinking water, seawater, aerosols,
soil and sediments. Since they are carcinogenic and
commonly found around us, they attracted more attention
around the world. They enter to aquatic environments by
oil spills, industrial discharges, municipal and urban
runoff and atmospheric precipitation. As PAHs are
hydrophobic, they partition to sediments through absorp-
tion by organic matter in the sediments. Therefore it is
important to investigate sediment in order to monitor
PAH concentration in aquatic environment, and it is also
important to develop, validate and optimize methodologies
for the determination of PAH level in sediments [1].
There are a lot of studies related with PAH analysis
and method validation in sediment samples [2-7]. It is
very well known that extraction is the first and important
step in the determination of PAHs especially in solid
matrices like sediments. Extraction methods for deter-
mination of PAHs from sediments were validated and
modified in some studies [3]. The mostly used methods
for extracting of PAHs from sediments and other solid
environmental materials are mechanical shaking, soxhlet
extraction, ultrasonic extraction, solid phase extraction
(SPE) and solid phase micro extraction (SPME). There
are several important factors affecting efficiency of the
extraction of PAHs from sediment samples, like solvent,
sorbent and clean up procedures. It is important to say
that there is no single methodology to provide very good
results for at least EPA’s priority 16 PAHs analysis.
Some investigators obtained lower detection limits but
only four or five selected PAHs [3]; therefore there is a
need to investigate different extraction methodologies
and their optimization for obtaining better extraction
efficiencies of PAHs in sediment samples. In most
optimization studies generally only one or two sediment
extraction methods were tried to be optimized without
using a chemometric approach [8-10] In this study, three
extraction methods were compared and one of them
(solid phase micro extraction) is very novel. Apart from
the other studies; both analysis methods (GC-FID, GC-
MS) and extraction methods were optimized before
starting PAH analysis in sediment samples. Most studies
optimized either extraction method or analysis method
alone [3,4] not both. For the optimization of extraction
methods, chemometrics [5]. “A new approach” was utili-
zed and instead of using established methods. The most
important advantage of chemometrics, is the minimi-
sation of the number of required experiments which in
turn results in the reduction of reagent consumption and
labor. Therefore chemometrics is faster and more cost
effective than traditional one variable at a time technique.
By using this method it is possible to study several con-
trol factors simultaneously. It is also possible to develop
mathematical models providing determination of the
relevance, statistical significance of factors and the esti-
mation of interaction effects between factors [6]. As an
analytical technique, capillary gas chromatography (GC)
coupled to either flame ionization (FID) or mass spec-
trometry (MS) detectors were used. By this way both
analysis and extraction methods were optimised and
validated for the PAH analysis in sediment samples. The
developed methods were tested on SRM 1941b and
applied to real samples.
The aims of this study were (i) to develop analysis
methodologies for the analysis of PAHs in sea sediments
collected in Ölüdeniz Lagoon by using GC-FID and
GC-MS (ii) to optimize extraction methods including
soxhlet, ultrasonic and solid phase micro extraction
(SPME) (iii) to evaluate the significance of the factors
and interactions affecting extraction efficiency.
2. Experimental
2.1. Reagents and Materials
All chemicals used during the study were of analytical
reagent grade. Two different PAH standards; Restek
(ISO 9001 registered in methylene chloride, 2000 ug/mL)
and Environmental Protection Agency (EPA) 610 (in
methanol: methylene chloride having a range of 10-1000
ppm, Supelco) were used. Standards (except for SPME)
for the calibration curve were prepared diluting them
with dichloromethane (DCM, Merck, HPLC-grade).
Standard reference materials which were SRM 1941b
(marine sediment) and 1597a (complex mixture of poly-
cyclic aromatic hydrocarbons from coal tar prepared in
DCM), were bought from National Institute of Standards
and Technology. The surrogate solution (Surrogate Stand-
ard Base/Neutrals Mix 19) including three surrogates;
nitrobenzene D5, p-terphenyl D14, 2-fluorobiphenyl was
bought from Dr. Ehrenstorfer. The concentration of
surrogate was 1000 g·ml–1 in DCM solution. The GC-
MS surrogate solution mix which includes Acenaphtene-
D10 (Ace-D10), Phenanthrene-D10 (Phe-D10), Chry-
sene-D12 (Chr-D12) and Perylene-D12 (Per-D12) (1000
mg/L, in Acetone) was bought also from Dr. Ehrenstorfer.
High purity nitrogen (99.999), helium (99.999), hydro-
gen (99.999) and dry air (99.999) were used for the
analyses. The sodium sulphate (Na2SO4, Merck, Darm-
stadt, Germany) was used for the removal of the water
from extracted samples. High purity glass wool was used
during the extraction of sediment samples. Cellulose
extraction thimbles (Schleicher & Schuell Micro science
603, 33 × 80 mm, ref. no. 10350240) were used for sox-
hlet apparatus. Sodium sulphate (anhydrous extra pure,
Merck) and glass-wool were pre-cleaned by transferring
them to a large glass column (1 L capacity or larger) and
washing sequentially with hexane and DCM before use.
Sodium sulphate was activated in an oven at 400˚C for 4 h.
Glass-wool was conditioned overnight at 225˚C in an
oven. Cellulose extraction thimbles were cleaned with
proper solvents prior to the use with soxhlet extraction
system. Alconox detergent powder (Supelco Cat. No.
1104) was used for the cleaning of all glassware used in
laboratory. All glasswares were cleaned with detergent in
hotwater and rinsed with deionized water. After rinsed
with DCM/acetone, they were dried in an oven overnight.
2.2. Sample Collection and Instrumentation
Samples (totally 68 sediment samples together with the
blank) were collected in 20-22 March, 2003 from Olud-
eniz Lagoon which is located on the intersection of
coastal lines of Mediterranean Sea and Aegean Sea in
Turkey. Among the analysed sediment samples 3 of them
was collected from the inside and 2 of them was col-
lected from outside of the Lagoon.
Before the extraction, sediment samples were homo-
genized and used for the optimization of the extraction
methods. For doing this 100 gram portions were taken
from each sample, put in a bottle having a capacity of 5.0
L and mixed. The prepared sample was kept in refrig-
erator at +4˚C. The samples were mixed with sodium
sulphate prior to extraction with soxhlet or ultrasonic
bath methods. Following the homogenization water con-
tent of the samples was determined. Moisture content of
the samples was ranged from 3% - 4%.
All the PAH analysis was performed with Hewlett-
Packard (HP) 6890 GC system equipped with a flame
Copyright © 2011 SciRes. AJAC
ionization detector. The system has also 5973 mass se-
lective detector. Heidolph rotary evaporator (laborota
4000 efficient) and mini-vap evaporator with 6 ports
(Catalog No: 22970, Cat. No: 22971) were used for the
evaporation of solvents. Bransonic ultrasonic bath
(Model B-2200 E4, 205 W, 220 V) was used for the sam-
ple extractions. Deionised water (ultra filtered type 1
water) was supplied by Barnstead nanopure ultrapure
water system.
3. Results and Discussion
Before starting optimization of extraction methods the
most proper solvent was selected. In addition some
preliminary experiments were done for SPME and best
fiber was selected for both direct SPME and HSSPME.
Both studies are discussed below.
3.1. Solvent Selection
The solvents were selected by considering some factors
such as suitability to analytes polarity, suitability to
standards solvent, toxicity and cost. In this study, three
solvents; DCM (solvents of the standards), acetone (least
toxic solvent) and hexane (suitable to polarity of analytes)
were selected under the EPA guidance. The most proper
(which extracts better than the others) three solvents/
solvent pairs were chosen as hexane/acetone, DCM/
acetone and DCM. Soxhlet extraction was chosen as a
method for the solvent selection and weight of sediment,
time and solvent volume were 80 g, 24 h and 500 ml
respectively. The number of replicate was three. Surrogate
was added to sediment sample prior to extraction in order
to decide which solvents recovery was the highest after
an extraction. According to average surrogate recovery
values: DCM/acetone was chosen as best solvent and it
was used in both soxhlet and ultrasonic bath extractions
(Figure 1).
Figure 1. Solvent selection.
Except nitrobenzene D5, other two surrogates, p-ter-
phenyl D14 and 2-fluorobiphenyl gave best result with
DCM/acetone. Especially with 2-fluorobiphenyl, DCM/
acetone resulted very good recovery value; 69%. Besides
having good recovery value, DCM/acetone was also
compatible with the standards solvent (DCM) and it was
less toxic than DCM because of the dilution (1:1) with
3.2. Fiber Selection for SPME and HSSPME
Fiber selection is one of the most important step in SPME
and HSSPME extractions. Three kinds of SPME fibers (7,
100 µm polydimethylsiloxane (PDMS) and 85 µm poly-
acrylate (PA)) were compared for the extraction effi-
ciency of the 16 EPA priority PAHs by GC-FID. Poly-
acrylate fiber provided highest extraction efficiency both
in direct SPME application and the headspace analysis
(HSSPME). In all experiments, acetone was used as a
solvent and SRM 1941b (Marine Sediment) was used as
a soil matrix.
Peak areas were used for the comparison of three fibers
and the selection of the best fiber. Peak areas showed
that, 85 µm PA had the best efficiency. Extraction effi-
ciencies of the fibers were in the order of; 85 µm PA >
100 µm PDMS > 7 µm PDMS.
3.3. Optimization of the Instrumentation and
Extraction Methodologies
After the selection of the solvent and the type of the fibers,
extraction methodologies were optimised. A chemometric
approach was used to reduce the number of experiments.
The name of the approach used was the screening design.
These type of experiments involve selection of the factors
which are important for the success of a process.
Screening analysis (23) which contains 8 runs, were
designed with the help of statgraphics plus 3.1 program
(statistical software) and used for the optimization of ex-
traction methods. Time, amount of sediment and solvent
(for ultrasonic bath and soxhlet), adsorption temperature,
desorption temperature, adsorption time, desorption time
and inlet temperature (for SPME) were evaluated among
the factors that may influence the extraction efficiency.
For SPME analysis, representative samples were
prepared in 4 ml vials; 8 standard water (prepared from
deionized water) solutions containing 1 mg/L PAH and 2
mg/L surrogate and 8 standard soil samples spiked with
0.5 mg/kg PAH and 1 mg/kg surrogate. These analyses
were done with only using the best fiber; 85 µm PA.
Sediment saples were analyzed with HSSPME and water
samples were analyzed with direct SPME. There were
Copyright © 2011 SciRes. AJAC
Copyright © 2011 SciRes. AJAC
mainly three factors at two levels (min-max) in direct
SPME and HSSPME designs as can be seen from Table
Factor names and levels for soxhlet and ultrasonic
bath extractions were also given in Table 1. There were
also three factors (A, B and C) and two levels (low and
high) in their factorial designs.
As can be seen from Table 1, for soxhlet extraction
the most important three factors were; extraction time,
solvent volume and amount of sediment. They were in-
vestigated at two levels which are low and high. Magni-
tude of the low and high values were determined by con-
sidering EPA procedures and literature studies [11].
Boiling temperature was chosen above the solvents boil-
ing points and it was kept constant during the whole ex-
For ultrasonic bath extraction, the factors were the same
with soxhlet extraction. Extractions were made in room
temperature. Low and high values (Table 1) were based
on the information given in EPA standards [12,13].
The important three factors; adsorption temperature,
inlet temperature and desorption time, for direct SPME,
their low and high values were given in Table 2. Ad-
sorption time was kept at maximum level; 60 min and
stayed constant during the optimization. The SPME
manual and fibers properties were considered while de-
termining factor levels. For example for proper desorp-
tion, inlet temperature cannot be too low or too high.
Since equilibrium is reached after a while and values
(concentrations, peak areas etc.) stays constant, longer
desorption and adsorption times were unnecessary. After
an adsorption step, SPME syringe was immediately
transferred to the inlet for desorption. Therefore moisture
in the needle was neglected.
Table 1 also shows the factors; adsorption temperature,
inlet temperature, adsorption time and their levels for
HSSPME. Instead of adsorption time in case of SPME
optimization, desorption time was kept constant (20 min)
during the optimization of HSSPME.
Using screening analyses results which were found in
terms of surrogate recoveries (or peak area) pareto charts
were drawn. They were used in order to decide the
signifance of the main effects. Response surface charts
were also prepared by using the same surrogate recovery
values. These charts were used to determine optimum
values for the selected factors.
The analysis of factorial design results produces main
effect and the two-factor interactions pareto charts (P =
95). In these charts bar lengths are proportional to value
of the effect, prividing comparison of the relative
importance of effects. Pareto charts also include, the
minimum t-values (at the 95% confidence interval) as a
vertical line. When the value of parameter is higher then
±t, it was assumed as a significant [14].
Table 1. Factor levels in factorial design (23) for extraction methods.
Factor Key Low (–) High (+)
Soxhlet extraction
Time (min) A 6 24
Solvent volume (ml) B 200 500
Amount of sediment (g) C 10 50
Ultrasonic bath extraction
Time (min) A 15 45
Solvent volume (ml) B 20 50
Amount of sediment (g) C 1 10
Direct SPME
Adsorption temperature (˚C) A 20 80
Inlet temperature (˚C) B 220 280
Desorption time (min) C 5 20
Adsorption temperature (˚C) A 20 80
Inlet temperature (˚C) B 220 280
Adsorption time (min) C 30 60
The results of factorial design and pareto charts for
each extraction techniques are discussed separately be-
low and given in Figure 2.
3.4. Pareto Chart Analysis
Pareto chart for sediment HSSPME analysis was shown
in Figure 2(a). As can be seen from the Figure 2(a) no
factor was significantly important (below the ±t line at
~13, P = 95). However factor A (Adsorption temperature)
was relatively more important than the others. The order
of the importance of the factors were; adsorption tem-
perature > inlet temperature > adsorption time. Interac-
tions of factors were also shown in the chart. For exam-
ple the interaction of inlet temperature & adsorption time
was more important (nearly twice as much as important)
than the interaction of the adsorption temperature & inlet
temperature and the interaction of the adsorption tem-
perature & adsorption time. In addition interaction of the
adsorption temperature & inlet temperature and the in-
teraction of the adsorption temperature & adsorption
time had nearly the same importance.
Figure 2. (a) Pareto chart for sediment HSSPME analysis; (b) Pareto chart for direct SPME water analysis; (c) Pareto chart
for soxhlet; (d) Pareto chart for ultrasonic bath.
Copyright © 2011 SciRes. AJAC
Figure 2(b) shows the pareto chart for direct SPME
analysis. This graph shows the importance of the factors
and their relationships. According to the graph all the
factors and interactions were below the line showing the
degree of the significance (line at 13 in standardized ef-
fects axis). Therefore there was no significantly impor-
tant factor affecting analysis performance. However the
most important factor was adsorption time which was
slightly more important than adsorption temperature. The
inlet temperature had the least importance. Order of the
importance of the factors were; adsorption time > ad-
sorption temperature > interaction of adsorption tem-
perature and adsorption time > inlet temperature >inter-
action of inlet temperature and adsorption time > interac-
tion of adsorption temperature and inlet temperature.
According to previous soxhlet analysis results DCM/
acetone was found to be best and used during the screen-
ing analysis experiments. For the extraction recovery
calculations, 3rd surrogates (p-terphenyl D14) responses
were used.
According to the standardized pareto chart (Figure
2(c)), sediment amount was the most important. The least
important factor was the solvent amount. The importance
of the factors and interactions were found as sediment
amount > interaction of extraction time and sediment
amount > extraction time > interaction of solvent volume
and sediment amount > solvent volume > interaction of
extraction time and solvent volume. No one of the fac-
tors or interactions was significantly important.
As in the case of soxhlet analysis, for ultrasonic bath
extraction DCM/acetone (best solvent) was also used as
a solvent. Calculated surrogate efficiency (3rd surrogate;
p-terphenyl D14) values were used for drawing pareto
According to pareto analysis in Figure 2(d) sediment
and time relationship could be investigated and solvent
had the less importance. There was no significantly im-
portant factor or interaction. The order of the importance
was found to be interaction of time and sediment >
sediment amount > solvent amount > extraction time >
interaction of time and solvent > interaction of solvent
and sediment.
3.5. Response Surface Methodology
Response surface methodology is a method combining
mathematical and statistical techniques based on fitting
of a polynomial equation to the experimental data ob-
tained. It describes the behaviour of a data set. In order to
use this method several variables that influence the re-
sponse must exist. The objective is to optimize the levels
of these variables simultaneously to obtain the best per-
formance. Firstly experimental design, defining which
experiments should be carried out, was chosen and then
the response surface methodology was applied [15]. Here,
the data obtained from factorial design of each extraction
techniques were discussed.
3.5.1. HSSPME
Response surface graph for the HSSPME analyses was
given in Figure 3(a).
As can be seen from the estimated response surface
graph maximum surrogate peak area (average of the peak
areas of two surrogates; nitrobenzene D5 and 2-fluoro-
biphenyl) was observed when factor A; Tadsorption (˚C), B;
Tinlet (˚C) and C; tdesorption (hour) were at their maximum
values (+1) which were 80, 280 and 1 respectively. The
lowest recovery value was observed when all factors are
(–1); that is when they were at their minimum values.
3.5.2. Direct SPME
Surface response analysis for SPME analyses was given in
Figure 3 (b).
As can be seen from the estimated response surface
graph maximum surrogate peak area (average of the peak
areas of 2 surrogates; nitrobenzene D5 and 2-fluorobi-
phenyl) was observed when Factor A; Tadsorption (˚C) and
B; Tinlet (˚C) were at their lowest levels (–1); room tem-
perature and 220 respectively and Factor C; tdesorption (min)
was in its maximum level (+1); 20. The lowest peak area
was found when factor A and factor B (+1) and factor C
is (–1).
3.5.3. So xh l et Extraction
Response surface analysis result for soxhlet was shown
in Figure 3(c).
For the surrogate recovery calculations, surrogate 3 (2-
fluorobiphenyl) was used. The maximum surrogate re-
covery was observed when factor A; time (min), B;
solvent (ml) and C; sediment (g) were at their maximum
levels (+1); 24, 500, 50 respectively. The minimum sur-
rogate recovery was observed when all factors were at
their lowest levels (–1).
3.5.4. Ultrasonic Bath Extraction
Estimated response surface analysis for ultrasonic bath
extraction was given in Figure 3(d).
For the surrogate recovery; surrogate 3 (2-fluorobi-
phenyl) was used. The maximum was observed when
factor A: time (min) and B: solvent (ml) were at their
low level (–1) and factor C: sediment amount (g) was at
its maximum level (+1): which were 15 min, 20 ml and 9
g respectively. The minimum was observed when factor
A and B were at their maximum values (+1) and factor C
was in its low value (–1).
Copyright © 2011 SciRes. AJAC
Estimated Response Surface Graph
Factor A
Factor B
0.2 0.2 0.6 1
0.2 1
Surrogate Peak Area
Factor C
Estimated Response Surface Graph
0.2 0.2 0.6 1
Factor B
Surrogate Peak Area
Factor C
Estimated Response Surface Graph
0.2 0.2 0.6 1
0.20.6 1
Surrogate Recovery
Factor C
0.2 0.2 0.6 1
0.2 0.6
Surrogate Recovery
Factor B
Factor A
Estimated Response Surface Graph
Figure 3. (a) Response surface for HSSPME sediment analysis; (b) Response surface for direct-SPME; (c) Response surface
for soxhlet extraction; (d) Response surface for ultrasonic bath.
3.6. Influences of the Selected Parameters
To see the influences of the selected parameters on ex-
traction efficiency, it should be looked at the models
created by statistical analysis model created according to
design matrix using any software package like SPSS.
Positive or negative contribution of the factors and their
amounts were easily observed from the models. By using
created model it is possible to calculate theoretical ex-
traction efficiency only by putting the value of the factor.
The statistical model created according to the design
matrix was given in the following Equation (1):
Y = b0 + b1x1 + b2x2 + b3x3
+ b12x1x2 + b13x1x3 + b23x2x3 (1)
where Y is the response variable, b0 is the general mean,
b1, b2, b3, b12, b13, b23 are the factors effect and x1, x2, x3
are the factors. Student test with a p-value was used to
test the significance of the factors. When p-values of the
factors were less than 0.05, they were considered as sta-
tistically significant factors [5].
Discussion related for this model discussed below.
3.6.1. HSSPME
For this technique the model formulation is given in
Equation 2. As can be seen from the Equation (3), factor
A, B and C had positive contribution on the extraction
efficiency. Among the factors, the greatest contribution
was belong to factor A and smallest contribution was
from factor C. Effect of the interactions was also under-
stood from the model. For example interaction of factor
A and B negatively contributed to the extraction effi-
ciency however interaction of factor A and C is posi-
tively contributed and the amount of their contributions
were nearly the same.
Extraction Efficiency (peak area)
= 257.875 + 124.125 × Factor A + 100.875 × Factor B
+ 22.375 × Factor C – 13.875 × Factor A × Factor B
+ 13.625 × Factor A × Factor C. (2)
3.6.2. SP M E
With the same approach the formulation is given in
Equation (3). As can be seen from Equation (3) factor A
and B were affecting extraction efficiency negatively. On
the other hand factor C had positive effect on the extrac-
Copyright © 2011 SciRes. AJAC
tion efficiency. According to the model; factor C > factor
A > factor B in terms of contribution to the extraction
efficiency. Effects of the interactions were also observed
from the model. Interaction of A and B was the least
Extraction Efficiency (peak area)
= 443.75 – 70.5 × Factor A – 24.5 × Factor B
+ 74.0 × Factor C + 9.25 × Factor A × Factor B
– 32.25 × Factor A × Factor C
+14.75 × Factor B × Factor C. (3)
3.6.3. So xh l et Extraction
Likewise for soxhlet extraction formulation is in the
Equation (4), all the factors and the interactions contrib-
ute positively and the order of importance was following;
factor C > factor A > factor B. Interaction of the A and B
was the least important.
Extraction Efficiency (surrogate recovery)
= 20.955 + 8.655 × Factor A + 3.1125 × Factor B
+ 12.285 × Factor C + 1.9025 × Factor A × Factor B
+ 10.27 × Factor A × Factor C
+ 3.1125 × Factor B × Factor C. (4)
3.6.4. Ultrasonic Bath Extraction
As can be understood from the Equation (5) all factors
contributed negatively and the order of the contribution
of the factors were found as; factor C > factor B > factor
A. Interaction of A and C was the most important among
Extraction Efficiency (surrogate recovery)
= 71.9375 – 1.7625 × Factor A – 4.7875 × Factor B
– 7.1125 × Factor C + 0.9625 × Factor A × Factor B
– 8.2125 × Factor A × Factor C
– 0.7375 × Factor B × Factor C. (5)
3.7. Comparison of Extraction Methods
By using surface response graphs best optimum values
were chosen for each extraction methods and methods
were compared according to surrogate recovery results.
The optimum surrogate recovery value for each method
was determined by using the optimized values of factors
in the Equations (3)-(6). Two µg surrogate was used for
both HSSPME and Direct-SPME. In order to calculate %
surrogate recovery of HSSPME, optimum peak area
determined from HSSPME, was divided into the opti-
mum peak area determined from direct-SPME and multi-
plied with 100 as given in the formula below.
% Surrogate Recovery for HSSPME
= Found Optimum Peak Area of Added Surrogate
(HSSPME) *100/Found Optimum Peak
Area of Added Surrogate (Direct-SPME) (6)
In order to calculate % surrogate recovery of ultrasonic
and soxhlet extraction, each concentration (found as a
result of each run), was divided into 10 and multiplied
with 100 as given below.
% Surrogate Recovery for UB and Soxhlet
= Found Concentration of Surrogate *100/Added
Concentration of Surrogate (10 mg/L) (7)
The optimization studies of extraction methods showed
that ultrasonic bath extraction was the best in terms of
surrogate recovery (81%) but HSSPME extraction (79%
recovery) was also applicable to extraction of PAHs
from sea sediments and it was a good alternative to the
ultrasonic bath for the moderately or less polluted sam-
ples. Soxhlet had the poorest extraction efficiency (60%
recovery). Therefore ultrasonic bath extraction method
with optimized parameters was used for the analysis of
Ölüdeniz sediment samples. The optimized soxhlet ex-
traction system might also be used for the analysis of
sediment samples for the less polluted samples.
3.8. Validation of the Method
As instrumental analyses, two methodologies namely
GC-MS and GC-FID were used in this study and their
features were discussed below.
3.8.1. GC - FID Analysi s
Before starting analysis with GC-FID, some important
parameters like injection volume, inlet temperature, car-
rier gas (He) flow rates, oven temperature program pa-
rameters (initial oven temperature, ramp rate and initial
time), make-up gas flow rate and flame ionisation detec-
tor (FID) temperature were optimised. Peak areas were
considered for the comparisons. GC conditions for de-
termination of PAHs were as follows.
After trying different columns, HP-5 5% Phenyl Methyl
Siloxane column (30 m length and 0.32 mm ID 0.25 µm
film thickness) having best resolution, higher peak areas
and lower background was chosen as best and used dur-
ing PAH analysis. Helium was used both carrier and
make up gas. Splitless mode was used. Injection volume
was 2 µl and inlet temperature was 300˚C. He flow was
1.5 ml·min–1 similar to the GC-MS conditions. Hydrogen
and air ratios were optimized and their values were cho-
sen as 40 ml·min–1 and 400 ml·min–1 respectively. It was
necessary to optimize these values to create more stable
flame which provides better ionization and lower back-
ground. Linear velocity was 27 cm·s–1 and FID tempera-
ture was 325˚C. It was necessary to keep this tempera-
ture higher than the oven temperature not to give any
damage to detector therefore 300˚C and 325˚C were tried
Copyright © 2011 SciRes. AJAC
Copyright © 2011 SciRes. AJAC
and according to the optimization studies 325˚C was
chosen as the best. Oven temperature program was 50˚C
at 2 min, 8˚C/min 280˚C at 10 min (total run = 40.75
The method detection limits (MDL) of the each target
analytes for GC-FID were determined and the results
were given in Table 2.
Each target compound was identified by its retention
time of the analyte. The width of the retention time win-
dow was 0.01 min to identify the target compounds. The
peak areas were integrated by chemstation and manually
adjusted if necessary. For controlling the retention times
of the analytes, GC-MS was used.
The standard curves were obtained by using 10, 50,
100, 500 and 1000 g·l–1 PAH standard and surrogates.
The response curves for 16 PAHs were linear with cor-
relation coefficients around 0.99. The method detection
limits ranged from 0.006 - 0.028 mg/L as was shown in
Table 2.
3.8.2. GC -MS Analysi s
The GC-MS conditions were the same with GC-FID. In
addition to them MS source temperature was 230˚C and
MS quadrupole temperature was 150˚C.
The standard curves were obtained 10, 50, 100, 500
and 1000 g l–1 PAH standards and the response curves
for 16 PAHs were linear with correlation coefficients
around 0.99. The range of r values lies between 0.997 -
1.000. The method detection limits ranged from 0.000012
- 0.000390 mg/L as can be seen from Table 2. The
method detection limits were determined based on the
concentration (or amount) of an analyte which gave a
signal three times the background noise.
3.8.3. Accuracy Test
The NIST SRM 1597a reference material was analyzed
in order to assess the accuracy of the developed methods
(GC-FID and GC-MS). The SRM 1597 offers certified
contents for 9 PAHs. Table 3 shows the concentrations
and recoveries of PAHs in SRM 1597a (mg·l–1) for
GC-MS. As can be seen, % errors of individual PAH are
lying between 5.74% - 46.6%. The concentrations and
recoveries of PAHs for GC-FID were also shown in Ta-
ble 3. It can be seen that the percent errors were in the
Table 2. Method detection limits (MDL) of PAHs for GC-FID and GC-MS.
No Name GC-FID MDL (mg·L1) GC-MS MDL (mg·L1)
1 Naphthalene 0.0110 0.0000980
2 Acenaphtylene 0.00700 0.0000560
3 Acenaphtene 0.0160 0.0000250
4 Fluorene 0.00900 0.0000370
5 Phenanthrene 0.0140 0.0000390
6 Anthracene 0.0100 0.0000250
7 Fluoranthene 0.00600 0.0000240
8 Pyrene 0.0130 0.0000120
9 Benzo(a)anthracene 0.0140 0.0000490
10 Chrysene 0.0140 0.000215
11 Benzo(b)fluoranthene 0.00900 0.000124
12 Benzo(k)fluoranthene 0.0160 0.0000720
13 Benzo(a)pyrene 0.0190 0.0000740
14 Indeno(1,2,3-cd)pyrene 0.0280 0.000193
15 Dibenzo(a,h)anthracene 0.0140 0.000390
16 Benzo(g,h,i)perylene 0.0140 0.000164
Table 3. Certified concentrations and % errors of PAHs in SRM 1597a (mg·l–1) for GC-MS (n = 4).
SRM 1597a Certified Conc. (mg/L) GC-FID % Error GC-MS % Error
Nap 1000 ± 50.0 1.70 –46.6
Phe 400 ± 4.00 25.0 –17.8
Fla 278 ± 4.00 2.10 –9.67
Pyr 204 ± 3.00 13.0 –5.74
BaA 85.3 ± 3.40 4.80 23.5
Chr 62.0 ± 1.00 20.0 32.7
BaP 82.9 ± 5.30 –1.40 –46.9
IcP 52.1 ± 4.00 8.20 –13.5
BgP 46.5 ± 6.70 –20.0 –23.7
range of 1.40% - 25.0%, demonstrating a good quantita-
tive agreement and a satisfactory performance of the
method. It can be understood from the Table 3 that;
GC-FID was better than GC-MS in terms of accuracy.
3.8.4. Application of Me thod to SRM 1941b
Since the comparison of extraction methods showed that
best extraction method was ultrasonic bath extraction,
SRM 1941b was dissolved by using this method. The best
optimum conditions for ultrasonic bath extraction ob-
tained from response from chart (Figure 3(d)). Table 4
shows the recovery values and percent errors for the
PAHs in SRM 1941b. As can be seen recovery values are
very good and changing between 67% - 101%.
3.9. Application of the Method to Sediment
Since analysis of the Ölüdeniz sediment samples were
done before the optimization of the extraction methods,
ultrasonic bath method based on EPA 3550 B [13] was
used for the extraction of 5 sediment samples. Solvents
were also selected according to the EPA 3550 B method
and hexane-DCM (1:1) pair was chosen as a solvent.
The extraction efficiency was calculated by using sur-
rogate spike containing 2-Fluorobiphenyl in DCM and it
was found as 91%. The procedure for the extraction of
sediment was summarized below.
Five gram (air dried) sediment was weighted into an
extraction vessel. After spiking with surrogate standard,
1ml of 6.5 g/ml 2-fluorobiphenyl in methylene chloride,
the sample was extracted in a 50 ml beaker for about 10
minutes with an 10 ml of DCM and hexane mixture (1:1,
v/v). The same procedure was repeated with a mixture of
hexane (5 mL) and DCM (5 mL) twice more for a total
volume of 30 ml.
After getting the extract, it was first transferred to the
glass wool column to get rid of the impurities in the so-
lution. Secondly, the extract was eluted from sodium
sulphate column with the use of pasteur pipette. Sodium
sulphate column was 20 cm in length, plugged with glass
wool and pre-washed twice with hexane and dichloro-
After drying, extracts were preconcentrated under the
stream of nitrogen and volume was reduced to 1 ml and
the final solution was kept in a 2 ml amber glass vials
tapped with a teflon-lined cap and preserved in a refrig-
erator at 4˚C until the analysis with GC-FID. The quanti-
fication was made against an external standard analytical
curve, constructed with solutions prepared from the Su-
pelco PAH standard in DCM: methanol (1:1).
In order to make a precision comparison between GC-
MS and GC-FID, three selected sediments among five
extracted sediments, were analyzed with both GC-FID
and GC-MS at the same time, with similar conditions. By
considering accuracy and precision values of the meth-
ods, best analysis method was chosen.
Table 5 shows the comparison of mean concentrations
of 5 analyzed sediment samples and their precisions.
3.10. Literature Comparison
Different extraction methods (soxhlet, pressurized fluid
extraction, shaking, microwave and SPME) and instru-
ments (GC-MS, GC-FID, HPLC-DAD) were compared
in terms of % recovery in Table 6 [14]. Optimized ul-
trasonic bath extraction method was used in this study.
This method includes extraction of 9 g wet sediment
sample with 20 ml DCM-Acetone (1:1 v/v) for 15 min-
utes. These optimum values were obtained from response
surface chart. Analysis with GC-FID resulted 81% re-
covery with surrogate; 2-Fluorobiphenyl. Recoveries
obtained for GC-MS; 41% for Acenaphtene D10, 60%
for Phenanthrene D10, 85% for Chrysene D12 and 86%
for Perylene D12. As can be seen from % recoveries in
Table 6, recovery values for this study are quite high or
comparable with the other values.
Table 4. SRM 1941b analysis results.
Certified Conc. (ug/kg)Found Conc. (ug/kg) % Error % Recovery
Naphthalene 848 ± 95.0 859 ± 33.0 1.00 101
Fluorene 85.0 ± 15.0 73.0 ± 13.0 –14.0 86.0
Phenanthrene 406 ± 44.0 333 ± 23.0 –18.0 82.0
Pyrene 581 ± 39.0 386 ± 27.0 –33.0 67.0
Benzo(a)anthracene 335 ± 25.0 225 ± 24.0 –33.0 67.0
Benzo(b)fluoranthene 453 ± 21.0 344 ± 48.0 –24.0 76.0
Benzo(e)pyrene 325 ± 25.0 351 ± 52.0 8.00 108
Benzo(a)pyrene 358 ± 17.0 330 ± 22.0 –8.00 92.0
Benzo(g,h,i) perylene 307 ± 45.0 267 ± 32.0 –13.0 87.0
Indeno (1,2,3-c,d) pyrene 341 ± 57.0 333 ± 79.0 –2.00 98.0
Copyright © 2011 SciRes. AJAC
Table 5. Comparison of the precisions of GC-MS and GC-FID measurements (n = 5).
PAHs Mean Conc. Precision Mean Conc. Precision
(mg/kg) Std dev. (mg/kg) Std dev.
Nap 0.0177 0.00410 ND ND
Acy 0.0370 0.0124 ND ND
Ace 0.0347 0.00790 0.662 0.480
Fle 0.0101 0.00440 0.0132 0.00860
Phe 0.406 0.517 0.439 0.380
Ant 0.114 0.0730 0.0405 0.0583
Fla 0.0960 0.141 0.248 0.338
Pyr 0.160 0.223 0.207 0.249
BaA 0.393 0.602 0.182 0.232
Chr 0.0830 0.115 0.291 0.250
BbF 0.104 0.131 0.0144 0.0147
BkF 0.0759 0.0898 0.0231 0.0239
BaP 0.0516 0.0679 0.178 ND
IcP 0.271 0.341 0.416 ND
DaA 0.0268 0.0242 0.0634 0.0559
BgP 0.0740 0.0744 0.0776 0.0655
Table 6. Literature comparison.
Sample Preparation Instrument % Recovery References
Soxhlet extraction with dichloromethane-16h. GC-MS 90-118 [15]
Pressurized solvent extraction with a mixture
of dichloromethane and acetone (3:1, v/v). GC-MS 75-105 [16]
Soxhlet extraction for 20 h with
Dichloromethane. GC-MS 90-105 [17]
Ultrasonication by acetonitrile15 min GC-MS
GC-MS: 53-73,
HPLC-DAD: 64-92 [18]
A. Soxhlet extraction: with a solution (1:1 v/v)
n-hexane-acetone B. Supercritical fluid
extraction: with co-solvents: n-hexane,
methanol and toluene.
GC-MS A: 73.8-89.1 B: > 90[19]
Shaking with dichloromethane for 30 min. HPLC-DAD > 80 [20]
Microwave extraction followed by solid phase
microextraction (with PDMS and PA fiber). GC-MS 58.6-112 [21]
Ultrasonication for 15 min with DCM/Acetone
(1:1 v/v).
GC-FID: > 81
GC-MS: 41-86 This study
4. Conclusions
Different extraction methodologies were optimised for
sixteen PAH compounds in sediment samples. Sixteen
PAH compounds in sediment samples were determined
with high recoveries by using HP 6890 series GC coupled
to a flame ionization detector. The optimization studies
of extraction methods showed that ultrasonic bath ex-
traction was the best in terms of surrogate recovery but
SPME extraction was also suitable for the extraction of
PAHs from sea sediments. The detection limits obtained
were comparable with the ones in EC directives.
It was also found that, optimized soxhlet extraction
system may also be used for the analysis of less polluted
sediment samples. For the analysis of naphthalene and
acenaphtylene which were below detection limit of GC-
Copyright © 2011 SciRes. AJAC
FID, GC-MS was offered as a better alternative.
Concentrations of PAHs in sediments were generally
moderate in Oludeniz sediment samples. Acenaphtene
and chrysene were dominant ones among the sixteen
PAHs analysed and their concentrations were 0.620 and
0.515 mg·kg–1 relatively.
5. References
[1] J. Xu, Y. Yu, P. Wang, W. Guo, S. Dai and H. Sun,
“Polycyclic Aromatic Hydrocarbons in the Surface Sedi-
ments from Yellow River, China,” Chemosphere, Vol. 67,
No. 7, 2007, pp. 1408-1414.
[2] L. Culotta, C. D. Stefano, A. Gianguzza, M. R. Mannino
and S. Orecchio, “The PAH Composition of Surface
Sedi- ments from Stagnone Coastal Lagoon, Marsala
(Italy),” Marine Chemistry, Vol. 99, No. 1-4, 2006, pp.
117-127. doi:10.1016/j.marchem.2005.05.010
[3] Y. F. Song, X. Jing, S. Fleischmann and B. M. Wilke,
“Comparative Study of Extraction Methods for the Deter-
mination of PAHs from Contaminated Soils and Sedi-
ments,” Chemosphere, Vol. 48, No. 9, 2002, pp. 993-
1001. doi:10.1016/S0045-6535(02)00180-7
[4] N. F. Leite, P. P. Zamora and M. T. Grassi,
“Multifactorial Optimization Approach for the Determi-
nation of Polycyclic Aromatic Hydrocarbons in River
Sediments by Gas Chromatography-Quadrupole İon Trap
Selected İon Storage Mass Spectrometry,” Journal of
Chromatography A, Vol. 1192, No. 2, 2008, pp. 273-281.
[5] M. B Deschamps, J. J. Daudin and E. Barriuso, “An
Experimental Design Approach to Optimise the Determi-
nation of Polycyclic Aromatic Hydrocarbons from
Rainfall Water Using Stir Bar Sorptive Extraction and
High Performance Liquid Chromatography-Fluorescence
Detection,” Journal of Chromatography A, Vol. 1167, No.
2, 2007, pp. 143-153. doi:10.1016/j.chroma.2007.08.025
[6] C. R. T. E. Tarley, G. Silveira, W. N. L. Santos, G. D.
Matos, E. G. P. Silva, M. A. Bezerra, M. Miró and S. L.
C. Ferreira, “Chemometric Tools in Electro Analytical
Chemistry: Methods for Optimization Based on Factorial
Design and Response Surface Methodology,” Micro
chemical Journal, Vol. 92, No. 1, 2009, pp. 58-67.
[7] J. N. Miller and J. C. Miller, “Statistics and Chemomet-
rics for Analytical Chemistry,” 4th Edition, Pearson
Education, England, 2000, Chapter 7.
[8] B. E. Berg, H. S. Lund, A. Kringstad and A. L. Kvern-
heim, “Routine Analysis of Hydrocarbons, PCB and PAH
in Marine Sediments Using Supercritical CO2 Extrac-
tion,” Chemosphere, Vol. 38, No. 3, 1999, pp. 587-599.
[9] D. R. Banjoo and P. K. Nelson, “Improved Ultrasonic
Extraction Procedure for the Determination of Polycyclic
Aromatic Hydrocarbons in Sediments,” Journal of Chro-
matography A, Vol. 1066, No. 1-2, 2005, pp. 9-18.
[10] H. Budzinski, M. Letellier, P. Garrigues and K. Le Men-
ach, “Optimisation of the Microwave-Assisted Extraction
in Open Cell of Polycyclic Aromatic Hydrocarbons from
Soils and Sediments: Study of Moisture Effect,” Journal
of Chromatography A, Vol. 837, No. 1-2, 1999, pp. 187-
200. doi:10.1016/S0021-9673(99)00067-9
[11] EPA METHOD 3540 C, 1996.
[12] EPA METHOD SW-846 3550C, 2007.
[13] EPA METHOD 3550 B, 1996.
[14] A. Nikolaou, M. Kostopoulou, G. Lofrano and S. Meric,
“Determination of PAHs in Marine Sediments: Analytical
Methods and Environmental Concerns,” Global Nest
Journal, Vol. 11, 2009, pp. 391-405.
[15] C. Anyakora, A. Ogbeche, P. Palmer, H. Coker, G. Ukpo
and C. Ogah, “GC/MS Analysis of Polynuclear Aromatic
Hydrocarbons in Sediment Samples from the Niger Delta
Region,” Chemosphere, Vol. 60, No. 7, 2005, pp. 990-
997. doi:10.1016/j.chemosphere.2004.12.073
[16] R. Boonyatumanond, G. Wattayakorn, A. Togo and H.
Takada, “Distribution and Origins of Polycyclic Aromatic
Hydrocarbons (PAHs) in Riverine, Estuarine and Marine
Sediments in Thailand”, Marine Pollution Bulletin, Vol.
52, No. 8, 2006, pp. 942-956.
[17] C. H. Koch, J. S. Khim, D. L. Villeneuve , K. Kannan
and J. P. Giesy, “Characterization of Trace Organic Con-
taminants in Marine Sediment from Yeongil Bay, Korea:
1. Instrumental Analyses,” Environmental Pollution, Vol.
142, No. 1, 2006, pp. 39-47.
[18] A. Filipkowska, L. Lubecki and G. Kowalewska, “Poly-
cyclic Aromatic Hydrocarbon Analysis in Different Ma-
tri- ces of the Marine Environment,” Analytica Chimica
Acta, Vol. 547, No. 2, 2005, pp. 243-254.
[19] V. Librando, O. Hutzinger, G. Tringali and M. Aresta,
“Supercritical Fluid Extraction of Polycyclic Aromatic
Hy- drocarbons from Marine Sediments and Soil Sam-
ples,” Chemosphere, Vol. 54, No. 8, 2004, pp. 1189-1197.
[20] R. A. Gimeno, E. Comas, R. M. Marcé, J. Ferré, F. X.
Rius and F. Borrull, “Second-Order Bilinear Calibration
for De- termining Polycyclic Aromatic Compounds in
Marine Sediments by Solvent Extraction and Liquid
Chromatography with Diode-Array Detection,” Analytica
Chimica Acta, Vol. 498, No. 1-2, 2003, pp. 47-53.
[21] V. Pino, J. H. Ayala, A. M. Afonso and V. Gonzalez,
“Micellar Microwave-Assisted Extraction Combined with
Solid-Phase Microextraction for the Determination of
Polycyclic Aromatic Hydrocarbons in a Certified Marine
Sediment,” Analytica Chimica Acta, Vol. 477, No. 1,
2003, pp. 81-91.
Copyright © 2011 SciRes. AJAC