GC-MS Determination of BADGE and BFDGE in Vegetable Oil

This article was downloaded by: [Gazi University]
On: 06 February 2015, At: 02:47
Publisher: Taylor & Francis
Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer
House, 37-41 Mortimer Street, London W1T 3JH, UK
Food Additives & Contaminants
Publication details, including instructions for authors and subscription
information:
http://www.tandfonline.com/loi/tfac19
Improved sample extraction and clean-up for the
GC-MS determination of BADGE and BFDGE in
vegetable oil
a
a
a
b
b
C. Brede , I. Skjevrak , H. Herikstad , E. Ånensen , R. Austvoll & T.
Hemmingsen
b
a
Naeringsmiddeltilsynet for Midt-Rogaland, Forusbeen 3, N-4033 Stavanger,
Norway
b
Stavanger University College, PO Box 2557 Ullandhaug, N-4091 Stavanger,
Norway
Published online: 10 Nov 2010.
To cite this article: C. Brede , I. Skjevrak , H. Herikstad , E. Ånensen , R. Austvoll & T. Hemmingsen (2002)
Improved sample extraction and clean-up for the GC-MS determination of BADGE and BFDGE in vegetable oil, Food
Additives & Contaminants, 19:5, 483-491, DOI: 10.1080/02652030110088293
To link to this article: http://dx.doi.org/10.1080/02652030110088293
PLEASE SCROLL DOWN FOR ARTICLE
Taylor & Francis makes every effort to ensure the accuracy of all the information (the “Content”)
contained in the publications on our platform. However, Taylor & Francis, our agents, and our licensors
make no representations or warranties whatsoever as to the accuracy, completeness, or suitability
for any purpose of the Content. Any opinions and views expressed in this publication are the opinions
and views of the authors, and are not the views of or endorsed by Taylor & Francis. The accuracy of
the Content should not be relied upon and should be independently verified with primary sources
of information. Taylor and Francis shall not be liable for any losses, actions, claims, proceedings,
demands, costs, expenses, damages, and other liabilities whatsoever or howsoever caused arising
directly or indirectly in connection with, in relation to or arising out of the use of the Content.
This article may be used for research, teaching, and private study purposes. Any substantial or
systematic reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution
in any form to anyone is expressly forbidden. Terms & Conditions of access and use can be found at
http://www.tandfonline.com/page/terms-and-conditions
Food Additives and Contaminants , 2002, Vol. 19, No. 5, 483 ± 491
Improved sample extraction and clean-up for the GC-MS
determination of BADGE and BFDGE in vegetable oil
C. Bredey*, I. Skjevraky, H. Herikstady,
E. AÊnensenz, R. Austvollz and T. Hemmingsenz
Downloaded by [Gazi University] at 02:47 06 February 2015
yNñringsmiddeltilsynet for Midt-Rogaland, Forusbeen 3, N-4033
Stavanger, Norway
zStavanger University College, PO Box 2557 Ullandhaug, N-4091
Stavanger, Norway
(Received 9 April 2001; revised 26 July 2001; accepted 7
August 2001)
A straightforwar d method was established for the
determination of migration contaminants in olive oil
with a special focus on the two can-coating migration
compounds bisphenol A diglycidyl ether (BADGE) and
bisphenol F diglycidyl ether (BFDGE). The preferred
sample preparation was a single liquid±liquid extraction
of compounds from the oil into 20% (v/v) methanol in
acetonitrile, followed by clean-up with solid-phase
extraction on aminopropy l bonded to silica. This puri®cation procedure selectively removed all free fatty
acids from the extracts without removing phenolic
compounds of interest. The solid-phas e extraction
columns were used many times by implementing a
procedure of washing out the strongly retained fatty
acids with 2% acetic acid in methanol. Gas chromatography coupled with full scan (m/z 33±700) electron
ionization mass spectrometry was used for the determination of several model compounds in olive oil samples.
BADGE and BFDGE could be determined in the
0.05±2 mg kg 1 range in oil samples with a relative
SD of <6% (six replicates). The method was used in
an enforcement campaign for the Norwegian Food
Control Authority to analyse vegetable oil samples
from canned ®sh-in-oil.
Keywords : migration, olive oil, gas chromatography,
mass spectrometry, bisphenol A diglycidyl ether, bisphenol F diglycidyl ether, screening method
* To whom correspondence should be addressed. e-mail: [email protected]
Introduction
Migration of compounds from food-packaging
materials to food is a well-known problem.
According to EU legislation, such migration is not
allowed if it leads to unhealthy concentrations in the
food or to unacceptable changes in the composition
or organoleptic properties of the food (EEC 1988).
Further, EU legislation gives a list of authorized
monomers and starting compounds for use in the
production of food-contact plastics (EEC 1990).
Some of the compounds listed are associated with
restrictions, which can be speci®c migration limits
(SML). An SML is the maximum allowed concentration of a migrated compound in the food. Another list
contains authorized plastic additives. This is still an
incomplete list, which means that non-listed substances can also be used today. Nevertheless, the use
of other plastic additives must not lead to unhealthy
or unacceptable migration levels in the food.
Several reports exist on the migration of compounds
from can coatings to canned food (Simal-Gandara
et al. 1998, Grob et al. 1999, Simoneau et al. 1999a, b,
Berger and Oehme 2000, Theobald et al. 2000, Berger
et al. 2001). Epoxyphenolic lacquers and organosols
are two frequently used can coatings. Bisphenol A
diglycidyl ether (BADGE) and bisphenol F diglycidyl
ether (BFDGE) are typical candidates of migration
from these coatings since they are monomers in the
former and might be used as process aids in the latter.
Today, BADGE is listed with an SML(T) of
1 mg kg 1 of food, which is a total limit that includes
some hydrolysis and chlorination products of
BADGE. A similar SML(T) is planned for BFDGE,
with the sum of BADGE components added (EEC
2001).
High-performance liquid chromatography (HPLC) is
the preferred technique for the determination of
BADGE and BFDGE in food and food simulants
(Losada et al. 1991, Biedermann and Grob 1998,
Summer®eld et al. 1998, Biles et al. 1999, Simoneau
Food Additives and Contaminant s ISSN 0265±203X print/ISSN 1464±5122 online # 2002 Taylor & Francis Ltd
http://www.tandf.co.uk/journals
DOI: 10.1080/0265203011008829 3
484
C. Brede et al.
Downloaded by [Gazi University] at 02:47 06 February 2015
et al. 1999a, Rauter et al. 1999). It provides robustness and low detection limits partly due to a relatively
large sample injection volume. In addition, gas chromatography coupled with mass spectrometry (GCMS) can be useful, often to verify a positive result
from the HPLC analysis. One paper describes the
combination of solid-phase micro-extraction with
GC-MS for the determination of BADGE in aqueous
food simulants (Salafranca et al. 1999). GC-MS oŒers
compound identi®cation based on both retention time
and mass spectrum recognition. From a food control
perspective, it is important to obtain such conclusive
evidence when checking for compliance with the
regulations.
Many compounds can be expected to migrate from
food-contact materials, and testing for compliance
provides some challenge to an o cial food control
authority. To save time and costs, it is wise to develop
generic methods of analysis. One such generic method
is the Dutch screening method for potential migrants
in food contact plastics (van Lierop 1994, 1997).
Similar screening methods should exist also for the
analysis of food simulants used in the tests for speci®c
migration. GC provides a highly e cient separation
of numerous compounds in a single chromatographic
run, thus allowing for development of such a generic
screening method. In the present work, GC-MS was
used for the determination of migration contaminants.
Olive oil is one of the EU food simulants and is used
to simulate fatty foods in migration experiments. It is
de®nitely more complex than aqueous food simulants
because it contains free fatty acids and many other
compounds in addition to the triglycerides . Free fatty
acids usually produce broad peaks in the gas chromatogram, while triglycerides have a low volatility
and might be retained in the injector liner or at the
column entrance. This puts a high demand on the
sample preparation for avoiding interference in the
chromatography. Both solid-phase extraction (SPE)
and liquid±liquid extraction (LLE) were evaluated for
the sample preparation in the present work, aiming to
achieve maximum sensitivity, precision and linearity
for the determination of BADGE and BFDGE, in
addition to the determination of a selection of other
model compounds in olive oil. This resulted in an
optimized LLE method with SPE clean-up, which was
used to determine BADGE and BFDGE in vegetable
oil samples from canned ®sh-in-oil products on the
Norwegian market.
Materials and methods
Chemicals
Olive oil with speci®cations according to the CEN
prestandard ENV 13130-1 was purchased from Norsk
Medisinaldepo t (Norway). The following solvents
were of analytical grade: acetone, methanol, acetonitrile, dichloromethane and n-hexane (Merck,
Darmstadt, Germany). The following compounds
were of analytical or high-purity grade and were
used as model compounds: bis(2-ethylhexyl) adipate (Merck), methyl-tert-butylether (Rathburn,
Walkerburn, UK), bisphenol A diglycidyl ether, bisphenol F diglycidyl ether, p-cresol, bisphenol A, 3tert-butyl-4-hydroxyanisole , hydroquinone (Fluka,
Buchs, Switzerland), butylated hydroxytoluene, dibutylphthalate (Sigma, St Louis, MO, USA), Irgafos
168, Irganox 1076, Tinuvin 327 (Ciba, Basel,
Switzerland), phenol, 2,2 0 -methylenbis(6-tert-butyl4-methylphenol), 2,2 0 -bisphenol F, 4,4 0 -bisphenol F,
bis(2-ethylhexyl) phthalate, diethylphthalate, benzyl
buylphthalate (Aldrich, Milwaukee, MI, USA), 2,4di-tert-butylphenol, 2,6-di-tert-butyl-p-benzoquinone,
1-chloro-dodecane (Acros, New Jersey, USA), di-noctylphthalate (Chem Service, West Chester, USA),
acetic acid (Riedel-de Haen Laborchemikalien ,
Seelze, Germany) and 4,4 0 -di¯uorbenzophenone
(Fluorchem, Derbyshire, UK). The bisphenol F diglycidyl ether (BFDGE) was a mixture of equal
amounts of the three isomers o,o 0 -BFDGE, o,p 0 BFDGE and p,p 0 -BFDGE, which were assumed to
elute on the gas chromatography column in this
order. Irgafos 168 phosphate was kindly donated by
Borealis, Norway.
Sample preparation methods
A total of 100 ml standard solutions in acetone or
100 ml neat acetone (sample blanks) were added to
10 g olive oil samples to establish standard curves. A
total of 100 ml neat acetone was also added to 10 g
samples of vegetable oil from canned ®sh-in-oil.
Solid-phase extraction (SPE) columns (1 g sorbent,
6 ml volume) containing silica, octadecyl bonded to
silica (C18) and aminopropyl bonded to silica (NH2 )
were purchased from IST (Mid-Glamorgan, UK).
A Supelco-5-7030 vacuum manifold (Supelco,
Bellefonte, USA) was used for the SPE. The ¯ow
Sample extraction and clean-up for the determination of BADGE and BFDGE
rates through the SPE columns were kept at 1±
2 ml min 1 .
Downloaded by [Gazi University] at 02:47 06 February 2015
The optimized sample preparation method based on
SPE consisted of the following steps.
. Conditioning of the silica SPE column with 5 ml
acetone and 5 ml hexane.
. Solution of a 0.5 g oil sample in 5 ml hexane.
. Application of the sample onto the SPE column.
. Washing the SPE column with 5 ml hexane.
. Elution of the sample with 5 ml acetone.
. Addition of internal standard (1-chloro-dodecane)
to the eluate.
. Evaporation to dryness at 55°C under a stream of
99.999% nitrogen followed by solving the sample in
1 ml hexane before GC-MS analysis.
485
was a Zebron ZB-5 (Phenomenex, USA) with a 5%
phenyl- and 95% methylsiloxane stationary phase of
0.25 mm thickness. The carrier gas was 99.9999%
helium (Hydrogass, Norway). Temperature programme: 40°C in 5 min, then programmed
10°C min 1 to 350°C and hold for 15 min. Pressure
programme: 7 psi in 2 min during the injection followed by 4 psi in 3 min, then programmed 0.5
psi min 1 to 20 psi and hold for 15 min. Injector
temperature: 250°C. Detector temperature: 300°C.
The mass spectrometer was a Hewlett-Packard model
5972 mass selective detector operated by the
ChemStation software G1701BA v.B.01.00 . Mass
spectra were recorded in the full scan mode
(m/z 33±700) by using 70 eV electron impact ionization.
The optimized sample preparation method based on
LLE followed by clean-up with SPE consisted of the
following steps.
. Addition of 20 mg internal standard (4,4 0 -di¯uorbenzophenone) in 10 ml acetone to a 10 g oil sample
contained in a 20 ml glass vial.
. Addition of 5 ml 20% (v/v) methanol in acetonitrile
to the 10 g oil sample.
. Manual shaking of the capped sample vial for 1
min.
. Sonication of the sample glass for 10 min followed
by resting for 30 min.
. Conditioning of the SPE column (C18 or NH2 )
with 4 ml 2% (v/v) acetic acid in methanol (NH2
only), 4 ml methanol, 4 ml acetone and 4 ml acetonitrile, followed by brief drying with vacuum.
. Transfer of exactly 3.5 ml of the top solvent phase
from the extracted sample to the SPE column.
. Elution of the sample followed by washing with
1 ml methanol and collection into a 4 ml glass vial.
. Evaporation to dryness at 55°C under a stream of
99.999% nitrogen followed by solving the sample
in 1 ml methyl-tert-butylether prior to GC-MS
analysis
Gas chromatography coupled with mass spectrometry
The gas chromatograph was a Hewlett-Packard
model 5890 series II (Agilent, Palo Alto, CA, USA).
1 ml of each sample was injected in the split-less mode
with a Hewlett-Packard model 7673 auto-injector.
The 30 m and 0.25 mm i.d. GC capillary column
Results and discussion
Solid-phase extraction
Both LLE and SPE have previously been reported as
sample preparation techniques for the determination
of BADGE in the olive oil fatty food simulant (EEC
1997). In the present work, SPE was initially evaluated for the extraction of polar and semipolar model
compounds from the non-polar triglyceride matrix.
When using silica as stationary phase, recoveries of
70±108% were found for BADGE and the three
BFDGE isomers (table 1).
There is a substantial quantity of extractable compounds in olive oil, including free oleic acid. When
using SPE, it is vital that the amount of extractable
material does not exceed the capacity of the SPE
column. Using SPE on 1 g silica, an olive oil sample
of 1 g produced a dried extract of 0.22 g, which is
equivalent to 22% of the weight of stationary phase.
The extract contained a high amount of oleic acid. Oil
samples of diŒerent weights were applied, and the
weight of dry extract was plotted against the weight
of oil sample (®gure 1). This showed that oil samples
>1 g produced extracts of similar weight, indicating
a limit of the SPE sample capacity. Hence, a larger
amount of stationary phase must be utilized if the
sample amount is to be increased. However, this will
require larger solvent volumes and was therefore not
an option in the present work.
486
C. Brede et al.
Table 1. Recoveries for the model compounds extracted from olive oil by using the SPE method (silica column) or by
using the LLE-SPE method (NH2 column).
Downloaded by [Gazi University] at 02:47 06 February 2015
Compounds
2,2 0 -Bisphenol F
2,4-di-tert-butylphenol
2,6-di-tert-butyl-p-benzoquinon
4,4 0 -Bisphenol F
BADGE; bisphenol A diglycidylether
BBP; butyl benzyl phthalate
BFDGE; o,o 0 -bisphenol F diglycidylether
BFDGE; o,p 0 -bisphenol F diglycidylether
BFDGE; p,p 0 -bisphenol F diglycidylether
BHA; 3-tert-butyl-4-hydroxyanisole
BHT; Butylated hydroxytoluene
Bisphenol A
BKF; 2,2 0 -methylenbis(6-tert-butyl-4-methylphenol)
DBP; dibutyl phthalate
DEHA; bis(2-ethylhexyl) adipate
DEHP; bis(2-ethylhexyl) phthalate
DEP; diethyl phthalate
DnOP; di-n-octyl phthalate
Hydroquinone
Irgafos 168; tris(2,4-di-tert-butylphenyl) phosphite
Irganox 1076; octadecyl 3-(3,5-di-tert-butyl-4-hydroxyphenyl) propionate
p-Cresol
Phenol
Tinuvin 327; 2-(5-chloro-2H-benzotriazol-2-yl)-4,6-bis(1,1-dimethylethyl)-phenol
Recoveries (%)
with SPE method
Recoveries (%)
with LLE-SPE method
n.a.
n.a.
n.a.
n.a.
70
n.a.
108
87
78
n.a.
4
57
n.a.
n.a.
55
51
n.a.
n.a.
n.a.
0
0
54
n.a.
4
81
21
9
62
42
35
53
55
68
35
11
60
17
33
n.a.
14
44
8
48
0*
5
24
14
n.a.
n.a., not applied.
*Con®rmed oxidation to Irgafos 168 phosphate.
Liquid±liquid extraction
Figure 1. Weight of dry SPE extract plotted against
weight of olive oil sample. Stationary phase: 1 g 50 ·m
silica particles with 60 AÊ pore size.
LLE was evaluated for the sample preparation of
olive oil samples, initially with a previously reported
method for the determination of BADGE in olive oil
(EEC 1997). With SPE, it was impractical to use
larger sample amounts than 0.5±1 g, which limited
the possibility of producing more concentrated
samples before GC-MS analysis. With LLE, larger
amounts of oil could be applied, leading to higher
concentrations of the prepared samples and thus
lower detection limits. LLE was performed with
mixtures of methanol and acetonitrile. With only
20% (v/v) of methanol added to the acetonitrile, a
signi®cant increase in the recovery was observed for
polar compounds like bisphenol A and p-cresol compared with using neat acetonitrile. However, methanol concentrations >50% (v/v) resulted in lower
recoveries for BADGE and BFDGE. This was either
due to a decreased amount of compound transferred
Downloaded by [Gazi University] at 02:47 06 February 2015
Sample extraction and clean-up for the determination of BADGE and BFDGE
487
in the LLE or possibly due to a methanol hydrolysis
of BADGE and BFDGE. Methanol hydrolysis of
BADGE and BFDGE did not occur when adding
20% (v/v) of methanol to the acetonitrile, as the
recoveries for BADGE and BFDGE were unchanged.
In addition, the recoveries did not change when using
neat acetonitrile and varying the methanol concentration of the washing solvent mixture. Hence, a mixture
of 20% (v/v) methanol in acetonitrile was chosen for
the extraction of several model compounds from olive
oil. The recoveries found for BADGE and the three
BFDGE isomers (table 1) were acceptable for quantitative work. Nevertheless, repeated LLE of the same
sample is expected to increase the recoveries, but this
was not investigated in the present work due to the
demand for a quick sample preparation.
Clean-up of extracts
The dry extract from the LLE weighed only 0.0175 g.
This indicated relatively clean samples compared with
the samples previously obtained with SPE. It is a
common procedure to clean up LLE extracts of fatty
samples by using SPE, and often with octadecyl
bonded to silica (C18) as stationary phase. C18 was
evaluated in the present work but was not found
e cient in terms of removing the free fatty acid from
the LLE extracts. For this purpose, aminopropyl
bonded to silica (NH2 ) was evaluated as the stationary phase, resulting in cleaned-up extracts completely
free from fatty acids. Methanol was used for washing
the SPE column to elute all of the polar model
compounds. Phenolic compounds were completely
eluted and could not be found in any of the following
column conditioning solvents, as con®rmed by GCMS analysis.
Figure 2 shows the total-ion chromatograms of LLE
extracts of oil samples from canned tuna ®sh, with
clean-up on C18 and with clean-up on NH2 . It was
observed that BADGE and the BFDGE isomers elute
between 28 and 32 min in these chromatograms, in an
area with heavy fatty acid interference when using
C18 for clean-up. The immense broadening of the
fatty acid peak cannot be avoided when using GC
columns with similar non-polar stationary phases.
The conditioning procedure implemented into the
LLE-SPE method allowed the columns to be used
multiple times. All free fatty acid content was washed
out of the SPE columns when using 2% acetic acid in
Figure 2. Total-ion chromatogram s (TIC) of LLE extracts of oil samples from canned tuna ®sh, with clean-up
on C18 (upper) and with clean-up on NH2 (lower).
DiŒerent scales of abundances .
methanol to promote ion suppressio n of the longchain fatty acids and thereby increasing their solubility in the organic solvent. The e ciency of this
column reconditioning was con®rmed by a GC-MS
analysis and by weighing of the washing extract.
Mass spectrometric detection
The MS was operated in the full-scan mode allowing
for reconstructed ion chromatograms (RIC) to be
made afterwards. For each compound, a RIC was
drawn using an ion of high abundance, which resulted
in increased selectivity and a lower detection limit.
Downloaded by [Gazi University] at 02:47 06 February 2015
488
C. Brede et al.
Figure 3. Overlay of reconstructed ion chromatogram s
(RIC) at m/z 181, 197, 312 and 325 showing BADGE
and the BFDGE isomers in a cleaned-up LLE extract of
olive oil containing 0.5 mg kg 1 of each isomer.
This enabled a low level determination of the model
compounds in olive oil. o,o 0 -BFDGE, o,p 0 -BFDGE,
p,p 0 -BFDGE and BADGE were determined using
RICs at m/z 181, 197, 312 and 325 respectively (®gure
3). RICs recorded from an olive oil sample blank did
not contain any peaks at the retention times of
BADGE and the BFDGE compounds. Mass spectrum recognition by the computer software was successful for these compounds, even at the LOQ
(0.05 mg kg 1 ), which provided an additional identi®cation parameter.
When olive oil is used as a fatty food simulant, it can
contain up to 0.5% free oleic acid. Oleic acid can also
be found in vegetable oil samples of canned foods.
This acid is a serious interference with the determination of bisphenol A due to close elution on the GC
column and because it produces a peak at m/z 213
RIC, which was utilized for the determination of
bisphenol A at low levels. By removing free oleic acid
as described above, it was possible to obtain higher
peak purity for bisphenol A (®gure 4). Bisphenol A
and bisphenol F are used in the production of
BADGE and BFDGE respectively, and are thus
potential migration contaminants in canned foods.
It was found that Irgafos 168 is readily oxidized to a
phosphate. This was probably the main reason for the
absence of Irgafos 168 at m/z 441 RIC. Instead,
Irgafos 168 phosphate appeared at m/z 316 and 647
RICs and could be detected at a level corresponding
Figure 4. Reconstructed ion chromatogram s (RIC) at
m/z 213 of LLE extracts of olive oil samples containing
1 mg kg 1 bisphenol A, with clean-up on C18 (upper) and
with clean-up on NH2 (lower).
to 0.5 mg kg 1 Irgafos 168 in olive oil. The mass
spectrum and retention time of Irgafos 168 phosphate
was con®rmed by injections of a standard solution.
Analytical characteristics
In this work, the limit of detection (LOD) was de®ned
as the amount of compound to give a signal-to-noise
ratio of 2. With the SPE method, and with full-scan
GC-MS analysis, the LOD for BADGE and the
BFDGE isomers were close to 1 mg kg 1 olive oil.
By using selected-ion monitoring (SIM) at abundant
ions, it was possible to reduce the LODs by a factor
Sample extraction and clean-up for the determination of BADGE and BFDGE
489
Table 2. Limits of detection (S/N = 2) for the model compounds in olive oil samples
using the LLE-SPE method.
Limits of detection
(mg kg 1 olive oil)
Downloaded by [Gazi University] at 02:47 06 February 2015
Compounds
2,2 0 -bisphenol F
2,4-Di-tert-butylphenol
2,6-Di-tert-butyl-p-benzoquinon
4,4 0 -Bisphenol F
BADGE; bisphenol A diglycidylether
BBP; butyl benzyl phthalate
BFDGE; o,o 0 -bisphenol F diglycidylether
BFDGE; o,p 0 -bisphenol F diglycidylether
BFDGE; p,p 0 -bisphenol F diglycidylether
BHA; 3-tert-butyl-4-hydroxyanisole
BHT; butylated hydroxytoluene
Bisphenol A
BKF; 2,2 0 -methylenbis(6-tert-butyl-4-methylphenol)
DBP; dibutyl phthalate
DEHP; bis(2-ethylhexyl) phthalate
DEP; diethyl phthalate
DnOP; di-n-octyl phthalate
Hydroquinone
Irgafos 168; tris(2,4-di-tert-butylphenyl) phosphite
Irganox 1076; octadecyl 3-(3,5-di-tert-butyl-4-hydroxyphenyl) propionate
p-Cresol
Phenol
0.1
0.01
0.09
0.01
0.01
0.04
0.01
0.02
0.01
0.01
0.01
0.01
0.01
0.02
0.05
0.01
0.03
0.06
0.5*
0.5
0.03
0.01
*Determined as the Irgafos 168 phosphate.
of 10. However, with GC-MS in the SIM mode, it
was impossible to identify the compounds by mass
spectrum recognition, but only by retention times.
Interference might exist in vegetable oil extracts from
canned ®sh, making the SIM approach less favourable.
by adjusting the extraction solvent polarity. Future
work will explore the usefulness of the method for
migration testing of materials and articles in contact
with the olive oil food simulant.
With the combined LLE±SPE method, and with fullscan GC-MS analysis, substantially lower LODs were
observed (table 2). Especially the LODs for BADGE
and BFDGE were low enough for compliance testing, giving limits of quantitation (LOQ) of 0.05 mg
kg 1 olive oil. Six consecutive injections of prepared samples from olive oil containing 1 mg kg 1
BADGE and the BFDGE isomers produced relative
SDs <6%. The same level of precision was found by
injecting six extracted samples containing phthalates
(DEP, DBP, BBP, DEHP, DnOP). Linear standard
curves (R2 = 0.995±0.999) were obtained for
BADGE and the BFDGE isomers in the 0.05±
2 mg kg 1 range.
Real samples
For some compounds, higher recoveries will be required for the method to provide reliable quantitative
data for diŒerent vegetable oil samples. This might be
achieved by performing repeated LLE per sample or
Real samples of vegetable oil from canned ®sh in oil
were analysed as part of an enforcement campaign for
the Norwegian Food Control Authority. Cans of 17
diŒerent products were collected, which were mostly
tuna ®sh and sardines. The cans were weighed before
and after drainage of the oil, and also after removing
the ®sh meat. The relative amount of oil in a product
was estimated and used for calculating the speci®c
migration level of BADGE and BFDGE in the food.
The amount of BADGE and BFDGE in ®sh meat
was set to zero. This simpli®cation was possible
because ®sh meat is likely to contain 20 times lower
amounts of BADGE and BFDGE than the oil
(Summer®eld et al. 1998). In addition, the reaction
products for BADGE and BFDGE were not determined to simplify the analysis. Thus, with the current
C. Brede et al.
490
Table 3. Levels of BADGE and BFDGE found in
samples of canned ®sh-in-oil. All samples originated
from diŒerent cans, but equal numbers represent the
same product.
Downloaded by [Gazi University] at 02:47 06 February 2015
Total concentration of
Concentration of
BFDGE in the food BADGE in the food
Sample number
(mg kg 1 )
(mg kg 1 )
1A
1B
1C
1D
2A
2B
2C
2D
3A
3B
3C
0.416
0.457
0.564
0.587
0.233
0.813
0.622
0.537
0.569
0.863
0.982
0.017
0.018
0.023
0.024
0.009
0.03
0.025
0.021
0.023
0.03
0.04
n.d.
n.d.
0.006 0.001
0.006 0.001
n.d.
0.032 0.002
n.d.
n.d.
0.013 0.001
0.008 0.001
0.007 0.001
selective MS detection. The GC-MS method developed here was suitable for the determination of the
can-coating migration contaminants BADGE and
BFDGE in vegetable oil. On the Norwegian market,
three of 17 products of canned ®sh-in-oil contained
BFDGE and traces of BADGE, but at total levels
<1 mg kg 1 in the food. Hence, no evidence of noncompliance was found.
The established method can be applied for screening
of several migration contaminants in vegetable oil,
and might prove to be useful for migration testing of
materials and articles in contact with the olive oil
food simulant. However, for some compounds, higher
recoveries must be achieved to provide reliable quantitative data.
Acknowledgements
method, any evidence of non-compliance could only
be obtained be revealing levels of BADGE and
BFDGE above the limit. Three products contained
BFDGE in all cans and traces of BADGE in some
cans (table 3). The variation within cans of the same
product was large, as indicated by 16±44% relative
SD (BFDGE). The levels of BADGE in some of the
cans were close to the detection limit of the method
and signi®cantly lower than the levels of BFDGE.
The total levels of BADGE and BFDGE were always
<1 mg kg 1 food, and thus no evidence of noncompliance with the current legislation was found in
the present work.
No analyses of the can lacquers were performed.
However, information was received from the manufacturer of one canned product which was found to
contain BFDGE and traces of BADGE. This information revealed that the can had an organosol lacquer inside the body and an epoxyphenolic lacquer
inside the lid.
Conclusions
Compared with SPE, the LLE method followed by
clean-up with SPE gave the lowest detection limits for
the model compounds in prepared olive oil samples.
The optimized sample preparation produced an extract free from fatty acid content which otherwise can
be a major interference in GC, even when using a
The authors thank their Project Coordinator Turid
Hellstrùm, Norwegian Food Control Authority, for
the funding of the work.
References
Berger, U., and Oehme, M., 2000, Identi®cation of derivatives of
bisphenol A diglycidyl ether and novolac glycidyl ether in can
coatings by liquid chromatography/ion trap mass spectrometry.
Journal of the Association of O cial Analytical Chemists, 83,
1367±1376.
Berger, U., Oehme, M., and Girardin, L., 2001, Quanti®cation of
derivatives of bisphenol A diglycidyl ether (BADGE) and novolac glycidyl ether (NOGE) migrated from can coatings into
tuna by HPLC/¯uorescence and MS detection. Fresenius
Journal of Analytical Chemistry, 369, 115±123.
Biedermann, M., and Grob, K., 1998, Food contamination from
epoxy resins and organosols used as can coatings: analysis by
gradient NPLC. Food Additives and Contaminant s, 15, 609±618.
Biles, J. E., White, K. D., and McNeal, T. P., 1999, Determination
of the diglycidyl ether of bisphenol A and its derivatives in
canned foods. Journa l of Agricultural and Food Chemistry, 47,
1965±1969.
EEC, 1988, Council Directive of 21 December 1988 on the approximation of the laws of the Member States relating to materials and
articles intended to come into contact with foodstuŒs (89/109/
EEC). O cial Journa l of the European Communities, L40, 38.
EEC, 1990, Commission Directive of 23 February 1990 relating to
plastic materials and articles intended to come into contact with
foodstuŒs (90/128/EEC). O cial Journal of the European
Communities, L75, 19.
EEC, 1997, Development of methods of analysis for monomers and
other starting substances with SML and/or QM limits in directives 90/128/EEC and 92/39/EEC.
Downloaded by [Gazi University] at 02:47 06 February 2015
Sample extraction and clean-up for the determination of BADGE and BFDGE
EEC, 2001, Draft Commission Directive on the use of some epoxy
derivatives in materials and articles intended to come into contact with foodstuŒs (SANCO/2001/0336).
Grob, K., Spinner, C., Brunner, M., and Etter, R., 1999, The
migration from the internal coatings of food cans summary of
the ®ndings and call for more eŒective regulation of polymers in
contact with foods: a review. Food Additives and Contaminant s,
16, 579±590.
Losada, P. P., Mahia, P. L., Oderiz, L. V., Lozano, J. S., and
Gandara, J. S., 1991, Sensitive and rapid reversed-phase liquid
chromatography-¯uorescence method for determining bisphenol A diglycidyl ether in aqueous-based food simulants. Journal
of the Association of O cial Analytical Chemists, 74, 925±928.
Rauter, W., Dickinger, G., Zihlarz, R., and Lintschinger , J.,
1999, Determination of Bisphenol A diglycidyl ether
(BADGE) and its hydrolysis products it canned oily foods from
the Austrian market. Zeitschrift fur Lebensmittel-Untersuchung
und -Forschung A, 208, 208±211.
Salafranca, J., Batlle, R., and Nerin, C., 1999, Use of solid-phase
microextraction for the analysis of bisphenol A and bisphenol
A diglycidyl ether in food simulants, Journa l of Chromatography A, 864, 137±144.
Simal-Gandara, J., Paz-Abuin, S., and Ahrne, L., 1998, A critical
review of the quality and safety of BADGE-based epoxy coat-
491
ings for cans: implications for legislation on epoxy coatings for
food contact. Critical Reviews in Food Science and Nutrition, 38,
675±688.
Simoneau, C., Theobald, A., Hannaert, P., Roncari, P.,
Roncari, A., Rudolph, T., and Anklam, E., 1999a,
Monitoring of bisphenol-A-diglycidyl-ether (BADGE) in
canned ®sh in oil. Food Additives and Contaminant s, 16, 189±
195.
Simoneau, C., Theobald, A., Wiltschko, D., and Anklam, E,
1999b, Estimation of intake of bisphenol-A-diglycidyl-ether
(BADGE) from canned ®sh consumption in Europe and migration survey. Food Additives and Contaminant s, 16, 457±463.
Summerfield, W., Goodson, A., and Cooper, I., 1998, Survey of
Bisphenol A diglycidyl ether (BADGE) in canned foods. Food
Additives and Contaminant s, 15, 818±830.
Theobald, A., Simoneau, C., Hannaert, P., Roncari, P.,
Roncari, A., Rudolph, T., and Anklam, E., 2000,
Occurrence of bisphenol-F-diglycidyl ether (BFDGE) in ®sh
canned in oil. Food Additives and Contaminant s, 17, 881±887.
Van Lierop, J. B., 1994, Enforcement of European Community legislation at the national level. Food Additives and Contaminant s,
11, 131±139.
Van Lierop, J. B., 1997, Enforcement of food packaging legislation.
Food Additives and Contaminant s, 14, 555±560.