Journal of Chromatography A, 1377 (2015) 46–54

Contents lists available at ScienceDirect

Journal of Chromatography A
journal homepage: www.elsevier.com/locate/chroma

A validated liquid chromatographic method for the determination of

polycyclic aromatic hydrocarbons in honey after homogeneous
liquid–liquid extraction using hydrophilic acetonitrile and sodium
chloride as mass separating agent
Anastasia Koltsakidou a , Constantinos K. Zacharis b,c,∗ , Konstantinos Fytianos a
a
Environmental Pollution Control Laboratory, Chemistry Department, Aristotle University of Thessaloniki, Thessaloniki 54124, Greece
b
Research Laboratory for the Physical and Chemical Testing of Foods, Department of Food Technology, School of Food Technology and Nutrition, Alexander
Technological Educational Institute (ATEI) of Thessaloniki, 57400, Thessaloniki, Greece
c
Analytical Development Laboratory, R&D API Operations, Pharmathen SA, Building Thermi 1, 9th klm Thessaloniki-Thermi, Thessaloniki, 57001, Greece

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

Article history: In the present report, a simple and cost-effective method for the determination of twelve US EPA priority
Received 11 October 2014 polycyclic aromatic hydrocarbons (PAHs) in honey samples after salting-assisted liquid–liquid extraction
Received in revised form 9 December 2014 and UHPLC with fluorescence detection is proposed. The sample treatment is based on the usage of
Accepted 11 December 2014
hydrophilic acetonitrile as extraction solvent and its phase separation under high salinity conditions.
Available online 18 December 2014
Due to the high sugar content of the samples the phase separation is promoted effortlessly. Several
parameters affecting the extraction efficiency and method sensitivity including the concentration of the
Keywords:
honey samples, the type and volume of the extraction solvent, the type and quantity of the inorganic
Homogeneous liquid–liquid
microextraction
salt, extraction time and centrifugation time was systematically investigated. The method was validated
Polycyclic aromatic hydrocarbons in-house according to the Commission Decision 2002/657/EC guidelines. The limit of detection (LOD) of
Honey the method lay between 0.02 and 0.04 ng mL−1 (corresponding to 0.08 and 0.16 ng g−1 ) which are close to
Ultra high pressure liquid chromatography the quality criteria established by European Regulation (EC) 836/2011 concerning the PAHs in foodstuffs.
Fluorescence detection The mean analytical bias (expressed as relative recoveries) in all spiking levels was acceptable being in
the range of 54–118% while the relative standard deviation (RSD) was lower than 19%. The proposed
method has been satisfactorily applied for the analysis of the selected PAHs residues in various honey
samples obtained from Greek region.
© 2014 Elsevier B.V. All rights reserved.

1. Introduction
Polycyclic aromatic hydrocarbons (PAHs) are compounds that
contain 2-6 fused aromatic rings and normally are produced
through partial pyrolysis of organic matter and various geological
processes [1]. Due to their mutagenic and carcinogenic properties
these compounds have been characterized as priority pollutants by
European Union and US Environmental Protection Agency [2]. The
occurrence of PAHs in foodstuffs is mainly attributed to the environ-
mental contamination or technological food processing (grilling,
smoking, etc.). In honey, the source of PAHs mainly comes from the
environment (e.g. air, soil, stubble burning, burned forest, place-
ment of beehives near to industrial areas, etc.) through the bees
∗ Corresponding author. Tel./fax: +30 2310 791581.
E-mail addresses: czach@food.teithe.gr, czacharis@gmail.com (C.K. Zacharis).
as they transfer the nectar from flowering plants. Another possible
contamination source can be through the blowing smoke into the
beehives during handling by the beekeepers [3]. Since it is a food
product with world-wide consumption especially among children
it must be free of chemical contaminants particularly from PAHs.
In order to protect the public health, the scientific panel
of Contaminants in the Food Chain (CONTAM panel) of
EFSA concluded that four specific PAHs, commonly called
as PAH4, (benzo[a]pyrene, B[a]P; benzo[a]anthracene, B[a]A;
benzo[b]fluoranthene, B[b]F; chrysene, Chry) or eight substances
(called as PAH8) including the above four and benzo[k]fluoranthene
(B[k]F), benzo[g,h,i]perylene (B[g,h,i]P), dibenzo[a,h]anthracene
(Db[a,h]A), indeno[1,2,3-c,d]pyrene (I(1,2,3-cd)P) are currently the
most appropriate indicators of the carcinogenic potency and effect
of PAHs in foodstuffs [4]. Furthermore, EFSA also deduced that the
set of eight PAHs do not provide much added value compared to
the group of four PAHs.

http://dx.doi.org/10.1016/j.chroma.2014.12.039
0021-9673/© 2014 Elsevier B.V. All rights reserved.
 A. Koltsakidou et al. / J. Chromatogr. A 1377 (2015) 46–54
The analysis of PAHs in honey samples is mostly performed by
conventional sample pretreatment techniques such as liquid-liquid
extraction (LLE) [5,6] solid-phase extraction [3] or matrix solid-
phase dispersion [7] in combination with separation techniques
(GC or HPLC) coupled to optical (fluorescence) or mass spectro-
metric detectors. LLE is generally laborious and time-consuming
procedure while large volumes of organic toxic solvents are usu-
ally required (e.g. CH2 Cl2 [5]). Although that SPE is considered as
the technique of choice in many laboratories, it requires additional
hardware and consumables increasing the entire operational costs
(typically 2–5 D per cartridge). The availability of rapid and low
cost analytical schemes for the determination of persistent organic
pollutants (pesticides, PAHs, etc) in honey is highly required for
quality control purposes.
A promising approach called as salting-out assisted
liquid–liquid extraction (SALLE) has been introduced by Matkovich
and Christian aiming on the extraction of metal chelates using
a binary homogenous system of acetone and water [8,9]. This
technique is based on the usage of a low dielectric constant organic
solvent such as acetonitrile, methanol, acetone etc, which is soluble
with water in any portion (homogeneous liquid-liquid extraction,
HLLE). The contact area between the two “phases” is practically
infinitive compared to the LLE or dispersive liquid–liquid microex-
traction (DLLME) facilitating fast mass transfer. Furthermore,
no vigorous shaking of the sample or even the application of
ultrasound is required [10,11]. The phase separation is based on
the (i) temperature, (ii) pH of perfluorinated surfactants, (iii) for-
mation of ion-pair and (iv) addition of an electrolyte (salting-out).
A detailed investigation of salting-out phenomenon in water-
acetonitrile homogenous system has been recently carried out by
Valente et al. [12]. The main advantages of this technique include
simplicity in operation, low cost, reduction the extraction time,
usage of non-halogenated or aromatic solvents compared to the
conventional LLE [13] or DLLME [14] and have attracted much
attention by many researchers. It has been successfully applied
for the determination of various analytes in biological matrixes
[15], foods [16] and environmental samples [17]. Recent literature
findings revealed that there are only three methods using the
above sample preparation technique in honey analysis aiming to
the determination of sulfonamides [18], fluoroquinolones [19] and
sulfanilamide [20].
The aim of this study was therefore to develop and validate
a simple analytical tool for the determination of twelve PAHs in
honey using the SALLE approach. To the best of our knowledge, it is
the first SALLE method for the determination of PAHs in honey. The
proposed analytical scheme was validated according to Commis-
sion Decision 2002/657/EC guidelines [21] and successfully applied
to the quantification of the selected pollutants in commercially
available honey samples of different floral origins.
2. Experimental
2.1. Reagents and solutions
All reagents used throughout this study were of analytical
grade or highest. A certified mixture of 13 PAHs (EPA 525 PAH
Mix A) containing 500 ␮g mL−1 each in dichloromethane was pro-
vided by Supelco Analytical. Acetonitrile (ACN), acetone, ethanol
and isopropanol were of HPLC grade and supplied by Panreac
(Barcelona, Spain) while methanol was purchased by Fluka. Salts
including MgSO4 , Na2 SO4 , (NH4 )2 SO4 , ZnSO4 were provided by
(Merck, Darmstadt, Germany) and NaCl from Chem-Lab NV (Zedel-
gem, Belgium). Ultrapure water (18 M cm resistivity) was used
throughout this study (Millipore Direct-Q UV, Millipore S.A.S., Mol-
sheim, France).
47
A standard PAHs stock mixture (50 mg L−1 ) was prepared in ACN
stored at −18 ◦ C and protected from the light. Working standard
solutions were prepared daily by appropriate dilutions of aliquots
obtained by the stock solutions in ACN. It should be mentioned
that during SALLE experiments, the volume fraction of ACN in the
spiked solutions was less than 0.5% v/v in all cases. To avoid carry
over effects all glassware used in this study was previously washed
with water and acetone.
2.2. Instrumentation and chromatographic conditions
All separations were performed on an Acquity UPLC binary sol-
vent system (Waters) equipped with a fluorescence and/or PDA
detector. The analytical column was a reversed phase Acquity
UPLC C18 BEH (100 × 2.1 mm, 1.7 ␮m) and was protected by guard
column (VanGuard 5 × 2.1 mm, 1.7 ␮m, Waters). The instrument
control and the data acquisition were carried out via the Empower
2 Pro software. Samples were filtrated through syringe filters
(0.22 ␮m PVDF, Millex® HV, Millipore) prior to the analysis. Glass
tight syringes with volumes of 100 ␮L and 1000 ␮L (Hamilton,
Nevada, USA) were employed for measurements of the volumes
of the extraction organic solvents and for the DLLME processes.
An EBA 20 (Hettich, Germany) centrifuge and an ultrapure water
(18 M cm resistivity) purification system were used throughout
this study (Direct-Q 3UV, Millipore S.A.S., Molsheim, France).
The analytes of interest were separated by a binary gradient
elution program using water (A) and ACN (B). The initial ratio was
50% v/v of B and linearly increased to 60% in 3 min and then to
70% in 6 min. The next 1 min was kept constant and increased to
90% in 12 min and finally to 100% in 13 min. At time of 13.1 min the
gradient elution is changed to the initial ratio and stated for a period
of 5 min to maintain a stable and reproducible separation. The flow
rate was set to 0.35 mL min−1 while the injection volume was 10 ␮L.
The column was thermostated to 30 ◦ C. The analytes were detected
spectrofluorimetrically using a two-channel program as described
in Table 1. Between injections the autosampler was sequentially
washed with 1500 ␮L water and 500 ␮L ACN to remove any sample
residuals.
2.3. Sampling and SALLE protocol
All honey samples were collected from local markets including
different botanical origins. They were kept in their original con-
tainers in dark place at ambient temperature as in everyday use.
Prior to sampling each specimen was mechanically homogenized
by stirring for 3 min [22]. PAH-free honey samples were used dur-
ing the optimization of the extraction parameters and validation
of the method. A pooled sample (n = 5) was prepared and homoge-
nized by mixing 50 ± 0.1 g of five honeys of different floral origins.
A 250 g L−1 honey solution was prepared in water, spiked with
the analytes and used throughout this study. Aqueous solutions
(250 g L−1 ) of individual, commercially available honey samples
were spiked with PAHs and left to “equilibrate” for at least for
15 min prior to the extraction.
An aliquot of 5 mL of the pooled or individual honey aqueous
samples (blank or spiked) was transferred in a 10 mL screw cap
glass tube with conical bottom. A volume of 1700 ␮L ACN was added
in the extraction vessel and the mixture was vortexed for 30 s.
Afterwards an amount of 1 g of NaCl was added in the aqueous-
acetonitrile homogeneous phase and the mixture was manually
shaken for 5 min in order to dissolve the solid salt. Then the mixture
was centrifuged at 4000 rpm for 5 min to enhance the phase separa-
tion. A portion of ca 500 ␮L of the resulted PAH-riched acetonitrile
phase (upper phase) was retrieved using a 1 mL glass microsyringe.
48 A. Koltsakidou et al. / J. Chromatogr. A 1377 (2015) 46–54
Table 1
Program of detection wavelength (ext /em ) settings on different channels (Ch1 & Ch2).
PAH (abbreviation) Start time (min)
Fluorene (Flu) 0.0
Phenanthrene (Phe) 5.1
Anthracene (Ant) –
Pyrene (Pyr) 6.3
Chrysene–Benzo[a]anthracene (Chry–B[a]A) 7.5
Benzo[b]fluoranthene (B[b]F) 9.3
Benzo[k]fluoranthene (B[k]F) –
Benzo[a]pyrene (B[a]P) –
Dibenzo[a,h]anthracene (Db[a,h]A) –
Benzo[g,h,i]perylene (B[g,h,i]P) 11.4
Indeno[1,2,3-cd]pyrene (I[1,2,3-cd]P) –
Then the sample was filtered to remove any undissolved particles
and transferred to an autosampler vial for analysis.
3. Results and discussion
3.1. Optimization of the chromatographic conditions
In order to establish the optimal separation conditions various
gradient programs were tried on an Acquity BEH UPLC C18 column
using water and acetonitrile as mobile phases. Using a multi-step
gradient program a resolution higher than 1.5 was obtained for
the majority of the analytes except from the pairs of B[b]F–B[k]F
and Chry–B[a]A where Rs were 1.2 and 0 (co-elution), respectively.
Aiming to optimize further the separation several trials were car-
ried out using a Phenyl stationary column. Although an improved
resolution between B[b]F & B[k]F was obtained; however the lately
eluted PAHs (namely Db[a,h]A, B[g,h,i]P, I[1,2,3-cd]P) were partially
overlapped. Finally, the gradient elution program as it described in
Section 2.2 on the C18 column was selected since the majority of
the compounds are well resolved. The compounds Chry–B[a]A were
unable to separate and were eluted at a single peak. For this reason
these compounds were measured as total “Chry + B[a]A”. Column
temperature was also tested in the range of 25-35 ◦ C. Less baseline
noise with reasonable backpressure was recorded at 30 ◦ C and this
value was selected as optimum.
3.2. Optimization of the SALLE conditions
As discussed in the introduction the implementation of SALLE
is based on the salting-out theory [8,9,23]. Two analogous theories
are currently presumed to describe the salting-out phenomenon.
One of them is applied only to ionic compounds and it claims that
the more polar solvent favorably surrounds the electrolyte (salt)
due to the Coulombic forces [8]. Another theory admits that one
of the solvents preferentially solvates the electrolyte (ionic or not
ionic) making it unavailable to dissolve the other solvent [8]. Also,
it is possible the above phenomena can act synergistically.
The above processes and therefore the effectiveness of the SALLE
are depended on many variables including e.g. the solvent type
and volume, salt type and amount, sample concentration which
are needed to be examined and optimized in order to establish
the optimum working conditions. Since honey is sugar rich–as it
contains monosaccharides, oligosaccharides, in total of ca 77% [22]
– its matrix may interfere the performance of SALLE due to the
sugaring-out effect [24]. On this basis to avoid any instability on the
extraction procedure we decided to optimize the SALLE parameters
using a PAH-free pooled honey sample (n = 5) spiked at concen-
tration level of 10 ␮g L−1 of the analytes. The extraction recovery
(ER, %) and the enrichment factor (EF) were calculated by the well-
known equations mentioned in [14].
Analyte detection ext /em (nm/nm)
Ch1 Ch2 Ch1 Ch2
√
290 330
√
250/350 250/380
√
√
305/375
√
260/375
√
260/450 295/390
√
√
√
√
300/480 295/400
√
Since PAHs tend to absorb on the wall of the containers, we
examined their potential retention on the syringe filter during the
filtration process of samples using PAH standard solutions in ace-
tonitrile at level of 10 ␮g L−1 [7]. The peak areas obtained were
no statistically different (at 99% confidence interval) using the t-
student test.
3.2.1. Study of honey concentration
The concentration of the honey sample solution was studied
in order to achieve maximum sensitivity and satisfactory repro-
ducibility without interferences from the sample matrix. To address
this task, SALLE extractions were performed at increased sample
concentration in the range of 100–500 g L−1 . During this study we
used ACN and NaCl as hydrophilic solvent and mass-separating
agent–at fixed amounts of 2500 ␮L and 1 g respectively. Accord-
ing to the literature these agents have been extensively used in
several SALLE methodologies [12,16,20]. As it illustrated in Fig. 1
the maximum EFs was achieved at sample concentration of 200
and 250 g L−1 with acceptable repeatability (RSD < 7%, n = 6) for
all examined compounds. Lower preconcentration factors were
obtained at elevated sample concentration (>300 g L−1 ). This may
be attributed to the higher sample solution viscosity preventing the
mass transference of the analyte to the organic layer. At the level
of 500 g L−1 the resulted solution was highly viscous hindering the
penetration of the added ACN in the sample while phase separa-
tion was achieved without the addition of salt (sugaring out effect
[24]). Additionally, the ACN layer was colored yellowish indicating
that some constituents were extracted from the honey matrix that
could potentially interfere the chromatographic analysis. Based on
these findings, the sample concentration of 250 g L−1 was finally
selected for the subsequent experiments. It should be noted that
during SALLE extractions a thin and white middle layer (containing
polysaccharides, proteins) was observed at the interface between
the aqueous and ACN layers that did not influence the collection
of the PAH-rich phase. Analogous observations have been made by
other researchers [25,26].
3.2.2. Study of the type and volume of the extraction solvent
Generally, the type of the extraction solvent plays an important
role in the overall efficiency of the liquid phase microextraction
methodologies [27]. In SALLE the solvent has to meet some general
criteria: (i) high extraction efficiency, (ii) good chromatographic
behaviour and (iii) to form a consistent clear layer after phase sep-
aration preferably at minimum salt amount. Based on the above
requirements, some organic solvents with different polarities (logP,
indicated in parenthesis) namely methanol (5.1), ethanol (5.2), ace-
tonitrile (5.8), acetone (5.1) and isopropanol (3.9) were examined.
Initially, a series of experiments has been carried out with different
added organic solvent volume to obtain a final upper layer of 500
(±25%) ␮L. Thus, a volume of 2400, 1540 and 1300 ␮L of acetone,
ACN and isopropanol were added, respectively while methanol and
 A. Koltsakidou et al. / J. Chromatogr. A 1377 (2015) 46–54 49

Fig. 1. Effect of the honey concentration on the EFs of the target analytes. For experimental details see Section 3.2.1.

ethanol had poor phase separation under these conditions. It should
be noted that during these experiments a hydrophobic compound
(e.g. Sudan III) in the absence of PAHs has been employed to visu-
alize clearly the organic layer. The obtained results showed that
higher extraction recoveries (> 60%) were achieved by using ACN
than those of the other solvents (Fig. 2). This might be attributed to
its closer polarity to that of water compared to the other solvents
[29]. Based on these considerations ACN was selected throughout
this work.
The volume of the extraction solvent is a key parameter that
affects the extraction efficiency. Lower added volumes typically
enhance the enrichment factors but inaccurate results might be
obtained especially when internal standard is not employed. Taking
into account these issues, a series of experiments was performed
altering the added ACN volume in the range of 1500–2500 ␮L since
lower volumes resulted to difficulty on the retrieval of the ACN
volume due to the formation of polysaccharides’ layer. The experi-
ments showed that the upper recovered layer was almost linearly
augmented by the added ACN volume. By increasing the volume the
average ER (Fig. 3) were increased up to the volume of 1600 ␮L and
then remained practically unaffected revealing sufficient extrac-
tion of the PAHs from honey matrix. Although higher ERs were
recorded at volume of1600 ␮L however poor precision and diffi-
culties in the collection of reproducible ACN phase were observed.

Fig. 2. Influence of the type of the extraction solvent on the ERs of the analytes. For experimental details see Section 3.2.2.
50 A. Koltsakidou et al. / J. Chromatogr. A 1377 (2015) 46–54
Fig. 3. Effect of the ACN volume on the ERs of the target compounds. For experimental details see Section 3.2.2.
Based on these findings, an added volume of 1700 ␮L (resulted to
650 ␮L upper layer) was finally chosen which provide sufficient
extraction recovery (60–85%), high EFs and easiness on phase col-
lection.
3.2.3. Study of the mass-separating agent and its amount
The type of salt has different impact on the phase separation on
homogenous liquid extraction systems. Three parameters should
be considered when selecting an electrolyte for the salting-out pro-
cesses: (i) the salt should be negligibly soluble in the water-miscible
organic solvent (ii) freely soluble in water in order to have the
maximum interaction with water molecules and (iii) the salting-
out capacity follows the Hofmeister series [8,12]. Additionally, the
presence of salt mainly leads to an increase of partition coefficients
of PAHs to a hydrophobic material [30].
The effect of various salts including NaCl, MgSO4 , Na2 SO4 ,
(NH4 )2 SO4 and ZnSO4 on the performance of the proposed method
were studied. According to literature the salt’s anion is predomi-
nantly responsible for the efficient phase separation and it follows
the Hofmeister series (SO4 2− > Cl− ) [12,31]. However, the obtained
results indicated that both ions in the examined salts had simi-
lar effect in terms of phase separation. Looking to the extraction
efficiency, the NaCl resulted to higher preconcentration factors for
all analytes (data not shown) and this is in agreement with the
majority of the salt-induced LLE approaches reported in the litera-
ture [13,18]. Thus NaCl was finally selected for subsequent studies
due to its strong salting-out ability and high solubility in aqueous
samples.
The effect of the added NaCl amount on the SALLE performance
was investigated in a range of 0.5–2.0 g in the 5 mL of honey sam-
ple solution. Higher amounts were not investigated because NaCl
would be oversaturated in the solution. Fig. 4 illustrates that the
EFs decreased by increasing of amounts of NaCl for all analytes. This
trend can be explained by the fact that elevated salt concentration
resulted to an increased ACN volumes and therefore dilution. At low
amount of salt (ca. 0.5 g) poor phase separation was observed while
the retrieval of ACN layer was difficult leading to high uncertainties.
Thus, 1 g of NaCl was finally chosen for subsequent experiments
that it provides adequate sensitivity and reproducibility.
3.2.4. Study of the extraction time and centrifugation speed
Initially, the efficiency of the proposed SALLE method using
ultrasonic treatment and manual shaking was examined. Poor
precision (>16% RSD) was obtained using ultrasonic irradiation
probably due to the incomplete salt dissolution. This issue was
independent from the orientation of tube in the ultrasonic bath
(vertical, horizontal) and analogous results have been concluded by
Wei et al. [11]. Based on these findings the manual shaking-based
extraction was adopted and the effect of the extraction time was
investigated from 1 to 10 min. For all analytes the EFs were sharply
increased from 1 to 5 min and then remain almost unaffected. This
designates that the diffusion of PAHs from honey sample to the
ACN medium is relative fast despite its high viscosity. Comparing
to DLLME protocols in the certain type of samples the SALLE pro-
vide lower mass transfer processes [26]. Thus the extraction time
of 5 min was selected.
Finally, a centrifugation step is used to enhance phase sepa-
ration. The effect of centrifugation speed on the extractability of
the PAHs was examined in the range of 3500–4500 rpm at con-
stant centrifugation time of 5 min. The averages of peak areas were
slightly improved up to 4000 rpm and being unaffected at higher
rates. Therefore, a centrifuge speed of 4000 rpm was chosen and
adopted.
3.3. Method validation
The proposed approach was in-house validated according to
Commission Decision 2002/657/EC and the requirements estab-
lished by Commission Regulation EC No 836/2011 oriented for the
analysis of B[a]P, B[a]A, B[b]F and Chry in foods [21,32]. The param-
eters evaluated including specificity, sensitivity, linearity, LOD,
LOQ, precision (repeatability and intermediate precision), trueness
(in terms of recovery) and uncertainty.
3.3.1. Selectivity, linearity, LOD & LOQ
The selectivity of the proposed method was assessed by ana-
lyzing twenty blank honey samples from different floral regions to
investigate the presence of potential interferences. No interfering
peaks present in the elution region of the analytes of interest were
 A. Koltsakidou et al. / J. Chromatogr. A 1377 (2015) 46–54
Fig. 4. Effect of NaCl amount on the EFs of the selected PAHs. For experimental details see Section 3.2.3.
observed considering the sufficient selectivity for the trace analysis
of PAHs.
The assessment of the linearity of the method was performed
using matrix-matched calibration solutions. Eight point calibration
curves were constructed in the range of 0.1 to 20 ␮g L−1 (cor-
responding to 0.4–80 ng g−1 of the target analytes in the honey
samples) by plotting the peak area of the each analyte against to the
concentration. For each calibration level five independent sample
extractions were carried out. Homoscedasticity test (expressed as
F-value) revealed that there was a significant difference between
the variances obtained at the lowest and at highest calibration level
(Fexp > F4,4,0.99 = 15.98)) at the confidence level of 99% for all analytes
studied [33]. Therefore, a weighted factor of 1/x2 was selected to
minimize the deviation of back-calculated values (residuals) from
theoretical concentrations especially at the lowest concentration
level as stated in the SANCO method validation guidelines [34].
For all compounds, the coefficient of determination was greater
than 0.99 demonstrating satisfactory linearity in the concentration
range studied.
The limit of detection (LOD) and limit of quantitation (LOQ) of
the proposed method were estimated from the matrix-matched
regression equations as 3 sb /a and 10 sb /a, respectively. The param-
eters of the linear regression equations: slope (a), intercept (b),
standard deviations of the slope (sa ) and the intercept (sb ), the cor-
relation coefficient (r), the LOD and LOQ were given in Table 2.
The LOD achieved for the target PAHs were in the range of
0.02–0.04 ␮g L−1 (equivalent to 0.08–0.16 ng g−1 in the honey sam-
ple) which are lower than the maximum residue limits for B[a]P,
B[a]A, B[b]F and Chry (0.3 ng g−1 each) as established by European
Commission EC No 836/2011 [32].
3.3.2. Accuracy and precision
The accuracy/trueness of the method was assessed by recovery
experiments using a pooled blank honey sample fortified with the
target analytes. Five individual extractions (m = 5) were performed
per each concentration level (0.4, 2, 4 and 40 ng g−1 ) (p = 4) at five
consequent working days (q = 5). The percent recovery was calcu-
lated using the expression R% = (Cspiked /CSTD ) × 100, where Cspiked
represents the experimentally found concentration of the analytes
51
and CSTD the theoretical spiking concentration level. The Cspiked was
calculated from the matrix-matched calibration curves using the
respective chromatographic peak areas.
One-way analysis of variance (ANOVA) was employed to deter-
mine the mean recovery at each spiking concentration level (5 × 5
spiked samples), the repeatability and between-day reproducibil-
ity of the method. As it can be seen on Table 3 the mean
recovery at the all studied levels was varied from 74 to 113%
which fully complied with the recommended ranges for true-
ness based on Commission Decision 2002/657/EC. The method
repeatability and intermediate precision were also investigated
via ANOVA calculating the F-values. In this test the between
days variance over the within-day one was assessed. In all cases
the experimental Fexp values were lower than theoretical Ftheor
(F4,12 , 0.05) = 3.26). As it shown in Table 3 the mean repeata-
bility (expressed as RSDr %) and between-days reproducibility
(expressed as RSDR ) were satisfactory being in the range of
6–17% and met the criteria of Commission Decision 2002/657/EC
[21].
The Horwitz ratio (H) or “HorRat” value is a normalized per-
formance parameter representing the acceptability of methods of
analysis with respect to among-laboratory precision (reproducibil-
ity). This factor is defined as H = RSDR /RSDH , where RSDR is the
experimental relative standard deviation of the within-laboratory
reproducibility and RSDH is the “target” relative standard devi-
ation. For concentrations lower than 120 ␮g Kg−1 , the “target”
standard deviation (sH ) and therefore the “target” relative standard
deviation RSDH is given as sH = 0.22 C (where C is the concen-
tration of the respective compound in ␮g Kg−1 ) as proposed
by Thompson [35]. For single-laboratory validation, the Horwitz
ratio should be ranged between 0.2 and 1 [36]. The “Horrat” val-
ues obtained for all studied concentration level were lower than
unity.
3.3.3. Uncertainty
The uncertainty of the method was calculated on the basis of
the data obtained from the validation procedure according to the
LGC/VAM/1998/088 protocol [37]. Taking into consideration that
the type B contributions to uncertainty are of minor significance
52 A. Koltsakidou et al. / J. Chromatogr. A 1377 (2015) 46–54

Table 2
Analytical figures of merit of the proposed method in matrix-matched calibration solutions.

PAH Linear range(␮g L−1 ) Regression equation y = (a ± sa )x + (b ± sb ) r LOD(␮g L−1 )/(ng g−1 ) LOQ(␮g L−1 )/(ng g−1 )

Flu 0.1–10 y = (303420 ± 17715)x + (9286 ± 3166) 0.9927 0.03/0.12 0.10/0.40

Phe 0.1–10 y = (483920 ± 22690)x + (41089 ± 4649) 0.9924 0.03/0.12 0.10/0.40
Ant 0.1–10 y = (1442412 ± 38704)x − (3278 ± 9637) 0.9943 0.02/0.08 0.07/0.28
Pyr 0.1–10 y = (1087343 ± 57462)x + (65617 ± 11307) 0.9939 0.03/0.12 0.10/0.40
Chry–B[a]Aa 0.2–20a y = (510122 ± 14820)x − (5334 ± 3690) 0.9957 0.02/0.08 a 0.07/0.28 a
B[b]F 0.1–10 y = (147378 ± 5646)x − (1610 ± 1406) 0.9919 0.03/0.12 0.10/0.40
B[k]F 0.1–10 y = (276922 ± 9001)x − (1690 ± 2241) 0.9964 0.02/0.08 0.07/0.28
B[a]P 0.1–10 y = (236116 ± 7845)x − (2734 ± 1953) 0.9922 0.02/0.08 0.07/0.28
Db[a,h]A 0.1–10 y = (896557 ± 24361)x − (17333 ± 6033) 0.9951 0.02/0.08 0.06/0.24
B[g,h,i]P 0.1–10 y = (182278 ± 11563)x − (9667 ± 2279) 0.9924 0.04/0.16 0.11/0.44
I[1,2,3-cd]P 0.1–10 y = (598614 ± 16472)x − (7572 ± 4101) 0.9958 0.02/0.08 0.07/0.28
a
Calculated based on “chrysene + benzo[a]anthracene” co-elution peak.

the standard uncertainty, u(Y), can be expressed using the following A t-test using the equation:
formula:
1 − Rm
 2 t=  
2 2
u(Y ) = u(P) + u Rm u Rm

  was employed to investigate whether Rm is significantly differ than

where u(P) and u Rm are uncertainties associated with the
1. In the circumstances that t values were lower than the coverage
method precision and recovery, respectively. Relative uncertainty
factor (k = 2.33, [39]) the Rm values were not statistically differ-
can be calculated using the following equation:
ent from the unity (at 99% confidence interval) and therefore no
 correction of the results for bias was required. On the other hand,
  
2 when t values were higher than the coverage factor k = 2.33 recov-
 
u (Y )  u (P) 2 u Rm ery correction was applied. The % expanded uncertainties, Uexp %,
= +
Y P Rm was given from the formula:
u 
(Y )
  Uexp % = k × 100
where u(P)/P and u Rm /Rm are the contributions of the within- Y
laboratory reproducibility, RSDR (obtained from the ANOVA test
at the respective spiking concentration level) and bias (calculated The % expanded uncertainties of the target PAH compounds are
from the RSD of the recoveries) in the total uncertainty. tabulated in Table 3 and the values were ranged between 16 to 40%.

Table 3
Method validation data at spiking levels of 0.4, 2, 4 and 40 ng g−1 : % mean recovery (R %), repeatability (RSDr %), within-laboratory reproducibility (RSDR %), expanded
uncertainty (Uexp %, at 99% confidence level) and “Horrat” values (H).

Analyte 0.4 ng g−1 2 ng g−1

R %a RSDr %b RSDR %c H Uexp % R %a RSDr %b RSDR %c H Uexp %

Flu 91 14 15 0.68 37 95 7 11 0.50 26

Phe 88 15 16 0.73 38 83 8 10 0.46 24
Ant 106 16 14 0.64 35 86 6 8 0.36 19
Pyr 74 13 15 0.68 39 76 10 12 0.55 29
Chry–B[a]Ad 97 10 11 0.50 26 79 8 9 0.41 22
B[b]F 82 13 14 0.64 36 76 13 15 0.68 35
B[k]F 91 8 10 0.46 25 79 9 10 0.45 24
B[a]P 113 13 15 0.68 36 89 8 9 0.41 23
Db[a,h]A 92 12 14 0.64 33 87 6 7 0.32 16
B[g,h,i]P 83 13 15 0.68 38 77 9 10 0.46 24
I[1,2,3-cd]P 86 16 17 0.77 40 80 7 10 0.46 25

4 ng g−1 40 ng g−1
R %a RSDr %b RSDR %c H Uexp % R %a RSDr %b RSDR %c H Uexp %

Flu 101 13 14 0.63 34 97 11 12 0.55 29

Phe 85 10 12 0.55 29 81 6 8 0.36 20
Ant 87 12 11 0.50 26 88 11 12 0.55 28
Pyr 87 14 16 0.70 38 85 10 11 0.50 27
Chry–B[a]Ad 81 7 9 0.41 22 80 10 12 0.55 29
B[b]F 80 7 9 0.41 23 84 8 10 0.45 26
B[k]F 84 8 9 0.41 21 87 9 11 0.50 26
B[a]P 91 9 11 0.50 26 92 7 9 0.41 23
Db[a,h]A 86 8 10 0.45 23 88 8 10 0.45 24
B[g,h,i]P 88 10 13 0.60 31 92 12 13 0.60 33
I[1,2,3-cd]P 85 10 12 0.55 29 89 8 9 0.41 23
a
Mean % recovery during 5 consecutive days (n = 5 × 5 = 25 determinations per spiking level).
b
% relative standard deviation under repeatability conditions (n = 5 × 5 = 25 determinations per spiking level).
c
% relative standard deviation under within-laboratory reproducibility.
d
Calculated based on “chrysene + benzo[a]anthracene” co-elution peak.
 A. Koltsakidou et al. / J. Chromatogr. A 1377 (2015) 46–54 53

Fig. 5. Representative chromatogram of PAHs spiked at 0.1 ␮g L−1 (corresponding to 0.4 ng g−1 ) in honey sample after SALLE. Peak identification: 1: Flu; 2: Phe; 3: Ant; 4:
Pyr; 5: Chry-B[a]A; 6: B[b]F; 7: B[b]F; 8: B[a]P; 9: Db[a,h]A; 10: B[g,h,i]P; 11: B[1,2,3-cd]P.

Table 4
Percent recovery obtained from the analysis of individual fortified honey samples using the proposed method.

Percent recoveriesa (RSD %)

PAH Sample Honey A Sample Honey B Sample Honey C Sample Honey D Sample Honey E

Fortification level (ng g−1 ) Fortification level (ng g−1 ) Fortification level (ng g−1 ) Fortification level (ng g−1 ) Fortification level (ng g−1 )
0.4 2 4 0.4 2 4 0.4 2 4 0.4 2 4 0.4 2 4

Flu 93 (15) 99 (13) 94 (5) 107 (18) 100 (10) 99 (10) 51 (12) 66 (9) 92 (5) 84 (7) 100 (4) 100 (4) 85 (7) 85 (7) 97 (8)
Phe 75 (15) 64 (10) 67 (6) 91 (11) 77 (9) 74 (5) 85 (11) 77 (9) 74 (5) 106 (17) 67 (16) 74 (10) 80 (15) 79 (15) 82 (4)
Ant 83 (15) 81 (17) 77 (4) 91 (7) 79 (9) 75 (8) 85 (10) 77 (12) 75 (6) 92 (15) 79 (5) 77 (4) 93 (13) 77 (6) 79 (5)
Pyr 84 (14) 90 (16) 77 (4) 67 (18) 76 (10) 82 (15) 58 (18) 69 (5) 73 (4) 73 (19) 86 (19) 79 (4) 68 (8) 75 (11) 75 (6)
Chry–B[a]Ab 84 (9) 85 (15) 81 (5) 95 (6) 82 (6) 78 (8) 77 (10) 68 (4) 70 (6) 95 (7) 78 (9) 80 (5) 92 (14) 75 (5) 79 (7)
B[b]F 103 (18) 73 (19) 71 (4) 61 (15) 70 (12) 68 (5) 56 (8) 61 (7) 67 (4) 66 (13) 70 (9) 69 (5) 59 (14) 61 (15) 70 (7)
B[k]F 91 (18) 74 (16) 72 (5) 57 (9) 71 (7) 74 (3) 57 (9) 73 (5) 70 (4) 59 (5) 70 (6) 71 (7) 53 (18) 74 (8) 74 (5)
B[a]P 116 (11) 94 (18) 86 (3) 102 (15) 83 (9) 85 (9) 76 (13) 86 (7) 88 (3) 94 (13) 86 (7) 88 (4) 92 (13) 87 (9) 92 (5)
Db[a,h]A 85 (10) 83 (17) 81 (4) 93 (5) 88 (7) 84 (7) 89 (8) 83 (5) 83 (5) 96 (14) 85 (5) 86 (5) 100 (14) 84 (8) 90 (4)
B[g,h,i]P 104 (14) 99 (12) 82 (11) 112 (11) 80 (13) 73 (5) 112 (12) 72 (19) 76 (4) 118 (14) 92 (4) 78 (12) 119 (13) 72 (7) 82 (13)
I[1,2,3-cd]P 79 (16) 81 (19) 82 (6) 69 (11) 72 (5) 74 (6) 58 (9) 75 (13) 75 (5) 54 (16) 76 (16) 74 (6) 69 (16) 71 (18) 84 (5)
a
Mean % recovery from five independently sample extractions (n = 5).
b
Calculated based on “chrysene + benzo[a]anthracene” co-elution peak.

3.4. Analysis of honey samples 4. Conclusions

The applicability of the proposed method has been evaluated
by analyzing commercial available honey samples from different
floral origins originated by Greek region. All samples were treated
according to the procedure described in Section 2.3. No detectable
levels of the studied compounds were found with the proposed
approach and this could be attributed to the low environmental pol-
lution in the geographical region or the accurate smoking process.
However, considerably high B[a]P level has been found in other
beehive products (e.g. propolis) as reported by Moret et al. [38].
The applicability and accuracy of the proposed method in real
samples was also demonstrated by spiking the samples at different
concentration levels in the range of 0.1–1.0 ␮g L−1 (0.4–4 ng g−1 ).
Table 4 shows representative results from the analysis of five honey
samples. The average recoveries for all analytes ranged from 51 to
118% while the relative standard deviation was lower than 19%. In
the case of B[a]P, B[a]A, B[b]F and Chry the recoveries were between
56 and 116% in all studied fortification levels which is in compliance
with the criteria established by Commission Regulation (EC) No
836/2011 [32]. A representative chromatogram of a spiked honey
sample at concentration level of 0.1 ␮g L−1 (0.4 ng g−1 ) of all the
compounds is illustrated in Fig. 5. The proposed SALLE method is
feasible for quantitative analysis of the selected PAHs in real sam-
ples, and could be used in routine analysis.
In this work, we proposed a simple analytical scheme for the
determination of twelve polycyclic aromatic hydrocarbons (PAHs)
residues in honey after SALLE and UHPLC coupled to fluorescence
detection. Various experimental parameters were investigated
and optimized. Compared to other recently reported methods,
the proposed method offers several advantages such as simplic-
ity, low operation cost employing typical laboratory equipment,
high extraction efficiency at a relative short extraction time. Since
acetonitrile, the extraction solvent, is compatible with various tech-
niques, the proposed method has the potential to be compatible
with these systems for sample analysis. The analytical figures of
merit are satisfactory and met the EU legislation criteria. Applica-
tion of the method in routine analyses would imply a significant
reduction of both effort and time.
References
[1] G. Purcaro, S. Moret, L.S. Conte, Overview on polycyclic aromatic hydrocar-
bons: occurrence legislation and innovative determination in foods, Talanta
105 (2013) 292–305.
[2] US Environmental Protection Agency, Guidelines Establishing Test Procedures
for the Analysis of Pollutants. Proposed Regulations. Federal Register, Vol. 49,
No 209, USEPA, Washington, DC.
[3] L. Corredera, S. Bayarri, C. Pérez-Arquillué, R. Lázaro, F. Molino, A. Herrera, Mul-
tiresidue determination of carcinogenic polycyclic aromatic hydrocarbons in
54 A. Koltsakidou et al. / J. Chromatogr. A 1377 (2015) 46–54
honey by solid-phase extraction and high-performance liquid chromatogra-
phy, J. Food Prot. 74 (2011) 1692–1699.
[4] Commission Regulation, (EU) No 835/2011 of 19 August 2011 Amending Regu-
lation (EC) No 1881/2006 as Regards Maximum Levels for Polycyclic Aromatic
Hydrocarbons in Foodstuffs, Official Journal of the European Union L215 (2011).
[5] S. Dobrinas, S. Birghila, V. Coatu, Assessment of polycyclic aromatic hydrocar-
bons in honey and propolis produced from various flowering trees and plants
in Romania, J. Food Compos. Anal. 21 (2008) 71–77.
[6] A. Ciemniak, A. Witczak, K. Mocek, Assessment of honey contamination with
polycyclic aromatic hydrocarbons, J. Environ. Sci. Health. B 48 (2013) 993–998.
[7] B. Albero, C. Sánchez-Brunete, J.L. Tadeo, Determination of polycyclic aromatic
hydrocarbons in honey by matrix solid-phase dispersion and gas chromatog-
raphy/mass spectrometry, J. AOAC Int. 86 (2003) 576–582.
[8] C.E. Matkovich, Salting-out of acetone from water-basis of a new solvent extrac-
tion system, Anal. Chem. 45 (1973) 1915–1921.
[9] C.E. Matkovich, G.D. Christian, Solvent extraction of metal chelates into water-
immiscible acetone, Anal. Chem. 46 (1974) 102–106.
[10] A.N. Anthemidis, K.-I.G. Ioannou, Recent developments in homogeneous and
dispersive liquid-liquid extraction for inorganic elements determination. A
review, Talanta 80 (2009) 413–421.
[11] S.-Y. Wei, M.-I. Leong, Y. Li, S.-D. Huang, Development of liquid phase microex-
traction based on manual shaking and ultrasound-assisted emulsification
method for analysis of organochlorine pesticides in aqueous samples, J. Chro-
matogr. A 1218 (2011) 9142–9148.
[12] I.M. Valente, L.M. Goncalves, J.A. Rodrigues, Another glimpse over the
salting-out assisted liquid–liquid extraction in acetonitrile/water mixtures, J.
Chromatogr. A 1308 (2013) 58–62.
[13] M.A. Farajzadeh, S. Sheykhizadeh, P. Khorram, Salting-out homogeneous
liquid-liquid extraction in narrow-bore tube: extraction and preconcentration
of phthalate esters from water, J. Sep. Sci. 36 (2013) 939–946.
[14] M. Rezaee, Y. Assadi, M.-R. Milani Hosseini, E. Aghaee, F. Ahmadi, S. Berijani,
Determination of organic compounds in water using dispersive liquid-liquid
microextraction, J. Chromatogr. A 1116 (2006) 1–9.
[15] F. Myasein, E. Kim, J. Zhang, H. Wu, T.A. El-Shourbagy, Rapid, simultaneous
determination of lopinavir and ritonavir in human plasma by stacking protein
precipitations and salting-out assisted liquid/liquid extraction, and ultrafast
LC-MS/MS, Anal. Chim. Acta 651 (2009) 112–116.
[16] R.M. Ramos, I.M. Valente, J.A. Rodrigues, Analysis of biogenic amines in wines
by salting-out assisted liquid–liquid extraction and high-performance liquid
chromatography with fl uorimetric detection, Talanta 124 (2014) 146–151.
[17] Y. Wen, J. Li, F. Yang, W. Zhang, W. Li, C. Liao, L. Chen, Salting-out assisted
liquid-liquid extraction with the aid of experimental design for determina-
tion of benzimidazole fungicides in high salinity samples by high-performance
liquid chromatography, Talanta 106 (2013) 119–126.
[18] J. Liu, M. Jiang, G. Li, L. Xu, M. Xie, Miniaturized salting-out liquid-liquid extrac-
tion of sulfonamides from different matrices, Anal. Chim. Acta 679 (2010)
74–80.
[19] D. Du, G. Dong, Y. Wu, J. Wang, M. Gao, X. Wang, Y. Li, Salting-out induced
liquid-liquid microextraction based on the system of acetonitrile/magnesium
sulfate for trace-level quantitative analysis of fluoroquinolones in water, food
and biological matrices by high-performance liquid chromatography with a
fluorescence detector, Anal. Methods 6 (2014) 6973–6980.
[20] H. Sereshti, M. Khosraviani, M. Sadegh Amini-Fazl, Miniaturized salting-
out liquid–liquid extraction in a coupled-syringe system combined with
HPLC–UV for extraction and determination of sulfanilamide, Talanta 121 (2014)
199–204.
[21] Commission Decision EC 2002/657 of 12 August 2002 Implementing Council
Directive 96/23/EC Concerning the Performance of Analytical Methods and,
2002.
[22] M.W. Kujawski, J. Namieśnik, Challenges in preparing honey samples for chro-
matographic determination of contaminants and trace residues, TrAC Trends
Anal. Chem. 27 (2008) 785–793.
[23] P.S. Albright, Experimental tests of recent theories descriptive of the salting-out
effect, J. Am. Chem. Soc. 59 (1937) 2098–2104.
[24] B. Wang, T. Ezejias, H. Feng, H. Blaschek, Sugaring-out A novel phase separation
and extraction system, Chem. Eng. Sci. 63 (2008) 2595–2600.
[25] L. Campone, A.L. Piccinelli, I. Pagano, S. Carabetta, R. Di Sanzo, M. Russo, L.
Rastrelli, Determination of phenolic compounds in honey using dispersive
liquid-liquid microextraction, J. Chromatogr. A 1334 (2014) 9–15.
[26] C.K. Zacharis, I. Rotsias, P.G. Zachariadis, A. Zotos, Dispersive liquid-liquid
microextraction for the determination of organochlorine pesticides residues
in honey by gas chromatography-electron capture and ion trap mass spectro-
metric detection, Food Chem. 134 (2012) 1665–1672.
[27] A. Sarafraz-Yazdi, A. Amiri, Liquid-phase microextraction, TrAC Trends Anal.
Chem. 29 (2010) 1–14.
[29] S. Song, E.N. Ediage, A. Wu, S. De Saeger, Development and application of
salting-out assisted liquid/liquid extraction for multi-mycotoxin biomarkers
analysis in pig urine with high performance liquid chromatography/tandem
mass spectrometry, J. Chromatogr. A 1292 (2013) 111–120.
[30] M.T.O. Jonker, B. Muijs, Using solid phase micro extraction to determine salting-
out (Setschenow) constants for hydrophobic organic chemicals, Chemosphere
80 (2010) 223–227.
[31] Y. Zhang, P.S. Cremer, Interactions between macromolecules and ions: The
Hofmeister series, Curr. Opin. Chem. Biol. 10 (2006) 658–663.
[32] Commission Regulation, (EU) No 836/2011 of 19 August 2011 Amending Regu-
lation (EC) No 1881/2006 as Regards Maximum Levels for Polycyclic Aromatic
Hydrocarbons in Foodstuffs, Official Journal of the European Union L215 (2011).
[33] A.M. Almeida, M.M. Castel-Branco, A.C. Falcão, Linear regression for calibration
lines revisited: weighting schemes for bioanalytical methods, J. Chromatogr. B
774 (2002) 215–222.
[34] Method Validation and Quality Control Procedures for Pesticides Residues
Analysis in Food and Feed, SANCO/10684/2009.
[35] M. Thompson, Recent trends in inter-laboratory precision at ppb and sub-ppb
concentrations in relation to fitness for purpose criteria in proficiency testing,
Analyst 125 (2000) 385–386.
[36] A.M. Gustavo González, Á. Herrador, A practical guide to analytical method vali-
dation, including measurement uncertainty and accuracy profiles, TrAC Trends
Anal. Chem. 26 (2007) 227–238.
[37] L.R.E. V.J. Barwick, AM Project 3.2.1. Development and Harmonisation of Mea-
surement Uncertainty Principles. Part D: Protocol for Uncertainty Evaluation
from Validation Data., 2000.
[38] S. Moret, G. Purcaro, L.S. Conte, Polycyclic aromatic hydrocarbons (PAHs) levels
in propolis and propolis-based dietary supplements from the Italian market,
Food Chem. 122 (2010) 333–338.