Journal of Chromatography A, 1035 (2004) 123–130

Trace level determination of acrylamide in cereal-based foods

by gas chromatography–mass spectrometry
Alain Pittet∗ , Adrienne Périsset, Jean-Marie Oberson
Nestlé Research Center, Nestec Ltd., Vers-chez-les-Blanc, P.O. Box 44, CH-1000 Lausanne 26, Switzerland

Received 20 November 2003; received in revised form 10 February 2004; accepted 11 February 2004

Abstract

A quantitative method has been developed for the determination of trace levels (<50 ␮g/kg) of acrylamide in cereal-based foods. The
method is based on extraction of acrylamide with water, acidification and purification with Carrez I and II solutions, followed by bromination
of the acrylamide double bond. The reaction product (2,3-dibromopropionamide) is extracted with ethyl acetate/hexane (4:1, v/v), dried over
sodium sulfate, and cleaned up through a Florisil column. The derivative is then converted to 2-bromopropenamide by dehydrobromination
with triethylamine and analyzed by gas chromatography coupled to mass spectrometry (GC–MS), employing (13 C3 )acrylamide as internal
standard. In-house validation data for commercial and experimental cereal products showed good precision of the method, with repeatability
and intermediate reproducibility relative standard deviations below 10%. The limit of detection and limit of quantitation are estimated at 2
and 5 ␮g/kg, respectively, and recoveries of acrylamide from samples spiked at levels of 5–500 ␮g/kg ranged between 93 and 104% after
correction of analyte loss by the internal standard. Finally, a comparative test organized with two independent laboratories provided additional
confidence in the good performance of the method, particularly at very low concentration levels.
© 2004 Elsevier B.V. All rights reserved.

Keywords: Cereals; Acrylamide

1. Introduction idation of sensitive and reliable analytical methods for the

low level quantification of acrylamide in different food ma-
In April 2002, researchers from the University of Stock- trices was considered as essential [8].
holm and the Swedish National Food Administration (NFA) Several methods have been developed in the past decade
reported the presence of acrylamide (2-propenamide) in a to determine acrylamide in water, biological fluids and
wide range of fried and oven-cooked foods [1]. These find- non-cooked foods (sugar, mushrooms, and field crops such
ings have attracted considerable attention worldwide be- as corn, potatoes, sugar beets, and beans), and the majority
cause acrylamide has been classified as “probably carcino- are classical methods based on high performance liquid
genic to humans” by the International Agency for Research chromatography (LC) or gas chromatographic (GC) tech-
on Cancer (IARC) [2]. Recent model system studies have niques [9–15]. However, these methods as such are not ap-
shown that acrylamide is formed during the Maillard reac- propriate for the analysis of acrylamide in processed/cooked
tion, and that the major reactants leading to the release of foods at low ␮g/kg levels. In particular, they lack the se-
acrylamide are sugars and asparagine [3,4]. The potential lectivity and additional degree of analyte certainty required
health risks of acrylamide in food have been evaluated by to confirm the presence of a small molecule such as acry-
the Scientific Committee on Food (SCF) [5] and the British lamide in a complex food matrix. This was largely observed
Food Standards Agency (FSA) [6]. In a next step, all avail- in our preliminary attempts to develop a GC method with
able data on acrylamide have been reviewed at international electron-capture detection (ECD) for the determination of
level by expert working groups (e.g. FAO/WHO, JIFSAN acrylamide in various food products. Recently, more se-
Workshop), identifying and listing a number of research gaps lective analytical methods have been published that are
and priorities [7,8]. Among these, the development and val- based mainly on mass spectrometry (MS) as the deter-
minative technique, coupled with a chromatographic step
∗ Corresponding author. Tel.: +41-21-7858245; fax: +41-21-7858553. either by LC [16–23] or GC, the latter in most cases after
E-mail address: [email protected] (A. Pittet). derivatization of the analyte [17,20,23–26]. These methods

0021-9673/$ – see front matter © 2004 Elsevier B.V. All rights reserved.
doi:10.1016/j.chroma.2004.02.037
124 A. Pittet et al. / J. Chromatogr. A 1035 (2004) 123–130

have been reviewed in detail during the Joint European sodium sulfate anhydrous, and zinc sulfate heptahydrate
Commission–Swedish National Administration Workshop were all obtained from Merck (Darmstadt, Germany) and
on “Analytical methods for acrylamide determination in were of analytical grade. Acetone for residue analysis and
food” that was held in Geel, Belgium, during 28–29 April triethylamine were purchased from Fluka (Buchs, Switzer-
2003. In a recent assessment of the performance of 37 lab- land). Florisil 60–100 mesh and hydrobromic acid 48%
oratories in determining acrylamide in crispbread, Clarke were obtained from Aldrich. Ethyl acetate for pesticide
et al. [27] reported that the majority of laboratories use analysis was supplied by SDS (Peypin, France). Potassium
either GC–MS or LC–MS, and that there was no obvious bromide, sodium sulfate anhydrous, and Florisil were puri-
method-dependent difference in results obtained between fied/activated by calcination in a crucible at 600 ◦ C (muffle
the two approaches. furnace) during one night and stored in tightly closed con-
The main advantage of LC–MS-based methods is that tainers at room temperature.
acrylamide can be analyzed without prior derivatization (e.g.
bromination), which considerably simplifies and expedites 2.3. Standards and reagents
the analysis. Due to the low molecular weight of acryl-
amide (71 g/mol) and thus also its low mass fragment ions, Stock solutions of acrylamide (1 mg/ml) and (13 C3 )acryl-
confirmation of the analyte can be achieved with tandem amide (0.1 mg/ml) were prepared by dissolving the com-
mass spectrometry (monitoring of more than one charac- pounds in distilled water. These solutions were then
teristic mass transition) [16–20]. However, acrylamide is a appropriately diluted with water to prepare working stan-
very polar molecule with poor retention on conventional LC dards at 10 and 4 ␮g/ml, respectively. All stock solutions
reversed-phase sorbents [20], and despite the use of tandem and working standards were stored in a refrigerator at 4 ◦ C
mass spectrometry, experience gathered in our laboratory has for maximum 3 months. The saturated bromine–water solu-
shown some limitations of this technique, like the difficulty tion (ca. 1.6%) was prepared by adding bromine (∼8 ml) to
of applying a “universal” clean-up approach valid for many 500 ml of water until precipitation became visible. Carrez I
different food matrices, and the difficulty of achieving a limit solution was prepared by dissolving 144 g of potassium hex-
of quantitation lower than ∼50 ␮g/kg in cereal products [22]. acyanoferrate(II) trihydrate in 500 ml of water, and Carrez
In this report, we present a robust and sensitive GC–MS II solution by dissolving 288 g of zinc sulfate heptahydrate
method for the analysis of trace levels (<50 ␮g/kg) of in 500 ml of water.
acrylamide in cereal-based foods. This method is based on
derivatization of acrylamide through bromination and is 2.4. Safety
used concomitantly with our LC–MS/MS method whenever
there is a need to achieve very low levels of detection. It Acrylamide monomer is toxic and readily absorbed
has been validated in-house and through participation in through the skin. It is classified as probably carcinogenic
a comparative test with two independent laboratories in to humans (group 2A) by the International Agency for Re-
Germany. search on Cancer. Gloves and safety glasses should be worn
at all times, and all standard and sample preparation stages
should be carried out in a fume cupboard.
2. Materials and methods
2.5. Extraction
2.1. Cereal samples
Finely ground cereal samples (15 g) were weighed into
Experiments were conducted with a series of commercial a 250 ml centrifuge bottle and spiked with 250 ␮l of
products of roller dried and extruded cereal-based foods (13 C3 )acrylamide internal standard (4 ␮g/ml). The sample
purchased from retail outlets in Brazil and stored at room was suspended in 150 ml of distilled water and homoge-
temperature, as well as with some cereal bases manufac- nized for 30 s with a Polytron homogenizer (Kinematica
tured for experimental trials in Switzerland. Nine of these AG, Luzern, Switzerland). The suspension was acidified to
samples have been sent to two private laboratories in Ger- pH 4–5 by addition of ∼1 ml of glacial acetic acid, treated
many for comparative analysis. successively with Carrez I and II solutions (each 2 ml), and
centrifuged at 16 000 × g (5 ◦ C) for 15 min. Then, the clear
2.2. Chemicals supernatant was filtered through glass wool into a 250 ml
Erlenmeyer flask.
Acrylamide (99%) and (13 C3 )acrylamide (isotopic purity
99%) were commercially available from Aldrich (Buchs, 2.6. Bromination
Switzerland) and Cambridge Isotope Laboratories (An-
dover, MA, USA), respectively. Acetic acid (glacial) 100%, Calcinated potassium bromide (7.5 g) was dissolved into
bromine, n-hexane, potassium bromide, potassium hexa- the filtrate with stirring, and the pH of the solution was
cyanoferrate(II) trihydrate, sodium thiosulfate pentahydrate, adjusted to a value between 1 and 3 by the addition of a
 A. Pittet et al. / J. Chromatogr. A 1035 (2004) 123–130 125

few drops (∼0.5 ml) of hydrobromic acid. Then, 10 ml of 2.9. GC–MS

saturated bromine–water solution was added to the flask
whilst stirring. The flask was covered with aluminum foil Brominated sample extracts and calibration standards
and transferred into an ice bath where reaction was allowed were analyzed on a Hewlett-Packard (HP) 5890 Series II
to take place for at least 1 h. After the reaction was com- gas chromatograph (GC) coupled to an HP 5972A benchtop
pleted, the excess bromine was decomposed by adding a mass selective detector (MSD) operated in selected ion mon-
few drops (∼1 ml) of 1 M sodium thiosulfate solution until itoring (SIM) mode with positive electron impact (EI) ion-
the yellow color disappeared. The mixture was transferred ization. The GC column was a ZB-WAX capillary column
to a 250 ml separatory funnel and extracted with 50 ml of (30 m × 0.25 mm i.d., 0.25 ␮m film thickness; Phenomenex,
ethyl acetate/hexane (4:1, v/v) by shaking for 1 min. After Torrance, CA, USA) connected to a deactivated fused silica
phase separation, the lower aqueous layer was discarded (in guard column (1 m × 0.53 mm i.d.; Agilent Technologies,
case of emulsion, the mixture was centrifuged at 2600 × g Palo Alto, CA, USA), and the carrier gas was helium at
for 10 min). The organic phase was filtered into a 100 ml 1.6 ml/min. Following injection, the column was held at
round-bottom flask through glass wool covered with ca. 15 g 65 ◦ C for 1 min, then programmed at 15 ◦ C/min to 170 ◦ C,
of calcinated sodium sulfate. The separatory funnel and the 5 ◦ C/min to 200 ◦ C, followed by 40 ◦ C/min to 250 ◦ C, and
filter were rinsed twice with 10 ml aliquots of ethyl ace- held for 15 min at 250 ◦ C (total run time: 30.25 min). In-
tate/hexane (4:1, v/v). Pooled fractions were evaporated to jections by an HP 7683 autosampler (2 ␮l) were made in
∼2 ml on a rotary evaporator (40 ◦ C, 140 mbar), and then to splitless mode (split flow 60 ml/min) with a purge activa-
dryness under a stream of nitrogen. tion time of 1.0 min and an injection temperature of 260 ◦ C.
The GC–MS interface transfer line was held at 280 ◦ C. Un-
2.7. Florisil clean-up der these conditions, the retention time of acrylamide and
(13 C3 )acrylamide derivatives was 11.3 min. Ions monitored
The residue was transferred quantitatively onto a glass were m/z 70, 149, and 151 for 2-bromopropenamide, and
chromatography column (430 mm × 11 mm i.d.) contain- m/z 110 and 154 for 2-bromo(13 C3 )propenamide.
ing 5 g of calcinated sodium sulfate and 5 g of activated
Florisil, using small aliquots taken from 50 ml of hexane. 2.10. Quantification
The column was then eluted with the remainder of this
50 ml of hexane, and the effluent was discarded. The acry- Acrylamide in cereal samples was quantified using the
lamide derivative (2,3-dibromopropionamide) was eluted ion at m/z 149 for 2-bromopropenamide, and the ion at m/z
with 150 ml of acetone, at a slow steady flow rate of 154 for 2-bromo(13 C3 )propenamide. The other ions at m/z
∼6 ml/min. The eluate was evaporated to ∼2 ml on a rotary 70, 110, and 151 were considered only for confirmation
evaporator (40 ◦ C, 200 mbar), and then to dryness under a purposes. A calibration graph was constructed by plotting
stream of nitrogen. The residue was redissolved in 400 ␮l peak area ratios (149/154) against the corresponding ratios
of ethyl acetate, and 40 ␮l of triethylamine was added to of analyte amounts, and the acrylamide contents in sample
convert 2,3-dibromopropionamide to 2-bromopropenamide. extracts were calculated from the calibration slope and in-
The solution was finally filtered through a 0.2 ␮m mi- tercept value. For acrylamide to be considered present in a
crofilter (Spartan 13RC, Schleicher & Schuell GmbH, sample extract, the following three criteria had to meet [28]:
Dassel, Germany) into an autosampler vial for GC–MS (1) the ratio of the retention time of 2-bromopropenamide
analysis. to that of 2-bromo(13 C3 )propenamide had to match the me-
dian ratio calculated for the seven calibration standards with
2.8. Bromination of calibration standards a tolerance of ±0.5%; (2) at least one ratio for the ions
70/149 and 151/149 had to be within ±20% of the median
Aliquots of 0 (blank), 5, 10, 25, 50, 100, 250, and ratio calculated for the seven calibration standards; (3) the
500 ␮l of acrylamide working standard solution (10 ␮g/ml) signal-to-noise ratio of the GC–MS peak had to be greater
were transferred to a set of eight 250 ml volumetric than 3.
flasks containing 100 ml of distilled water, and 250 ␮l of
(13 C3 )acrylamide internal standard (4 ␮g/ml) was added to
each flask. These solutions were brominated according to 3. Results and discussion
the procedure described earlier for sample extracts (with-
out Florisil clean-up). Final residues were redissolved in This work describes a quantitative analytical method for
1000 ␮l of ethyl acetate, and 100 ␮l of triethylamine was the low level determination of acrylamide in cereal-based
added to each flask. The derivatized solutions were then foods. The proposed method is comparable to the US
filtered through 0.2 ␮m microfilters (Spartan 13RC), and EPA method for analysis of acrylamide in water [11],
aliquots (∼150–175 ␮l) of each solution were distributed but improvements have been made particularly regard-
among six to seven separate autosampler vials that were ing sample clean-up (inclusion of a purification step with
stored in a freezer at −18 ◦ C until use. Carrez reagents before bromination, and modification of
126 A. Pittet et al. / J. Chromatogr. A 1035 (2004) 123–130

Florisil clean-up conditions). Moreover, isotopically-labeled anhydrous sodium sulfate salts by calcination in a crucible
(13 C3 )acrylamide is added to the test portion before extrac- at 600 ◦ C (muffle furnace) during one night, which resulted
tion so as to keep control on the recoveries achieved and in much cleaner chromatograms for some difficult cereal
to keep track of possible losses occurring during the whole matrices. The remaining impurities (present essentially in
sample pre-treatment (extraction and clean-up). Despite its potassium bromide) were removed by a Florisil clean-up
low molecular weight (71 g/mol), acrylamide can be ana- procedure, as originally proposed in US EPA Method 8032A
lyzed as such without derivatization [25,27], but when using for analysis of water [11]. However, the EPA clean-up
GC–MS the molecule is normally brominated to form the conditions have been improved/adapted to the analysis of
2,3-dibromopropionamide derivative [10,11]. cereals by using hexane instead of benzene for loading
sample extracts onto the column, and by using a single step
3.1. Bromination elution procedure with 150 ml of acetone instead of diethyl
ether/benzene (1:4) and acetone/benzene (2:1) mixtures.
The advantage of acrylamide bromination is that a more
volatile compound is produced, which leads to improved 3.3. GC–MS
GC characteristics (less polar) and improved MS char-
acteristics (higher mass ions and characteristic 79 Br/81 Br The ions monitored for identification of the ana-
patterns). This results in an increased selectivity, which lyte, 2-bromopropenamide, were [C3 H4 NO]+ = 70,
compensates for a laborious and time-consuming deriva- [C3 H4 79 BrNO]+ = 149, and [C3 H4 81 BrNO]+ = 151, using
tization process. In the present method, conversion of m/z 149 for quantification. The ions monitored for identifi-
acrylamide to 2,3-dibromopropionamide is performed ac- cation of the internal standard, 2-bromo(13 C3 )propenamide,
cording to the protocol originally described by Hashimoto were [13 C2 H3 81 Br]+ = 110 and [13 C3 H4 81 BrNO]+ = 154,
[29], which involves addition of potassium bromide, hydro- using m/z 154 for quantification (Figs. 1 and 2). Quan-
bromic acid, and a saturated bromine–water solution. The tification was performed by comparison with a calibration
excess of bromine is then removed by addition of sodium curve (0.17–17.5 ng of 2-bromopropenamide, correspond-
thiosulfate until the solution becomes colorless. Under ing to 5–500 ␮g/kg of acrylamide in the test portion), and
these conditions, the yield of 2,3-dibromopropionamide samples with acrylamide concentrations >500 ␮g/kg were
is constant and >80% when the reaction time is more diluted in ethyl acetate and re-injected. Calibration curves
than 1 h [11]. This derivative is less polar compared to typically produced correlation coefficients ranging between
the original compound and is therefore easily soluble in 0.997 and 1.000. The limit of detection calculated from the
non-polar organic solvents like ethyl acetate and hexane. measurement of standard solutions is estimated at 2 ␮g/kg
However, Andrawes et al. [30] have shown that under (signal-to-noise ratio of 3), and the limit of quantitation at
certain conditions, 2,3-dibromopropionamide can be con- 5 ␮g/kg, which corresponds to the lowest concentration of
verted to the more stable derivative 2-bromopropenamide acrylamide that can be determined in a cereal matrix with
on the inlet of the GC or directly on the capillary column. an acceptable level of accuracy (repeatability) under the
Because this decomposition (dehydrobromination) may stated conditions. Acrylamide in cereals was confirmed if
yield poor repeatability and accuracy, it is preferable to at least one ratio for the ions m/z 70/149 and 151/149 was
deliberately convert 2,3-dibromopropionamide to the stable matching the median ratio calculated for the seven cali-
2-bromopropenamide prior to GC analysis, which can be bration standards within an acceptable tolerance of ±20%.
readily done by adding 10% of triethylamine to the final However, at concentration levels lower than the limit of
extract before injection. This conversion is almost instan- quantitation (5 ␮g/kg), the two ratios were often found
taneous at room temperature and has been shown to be outside the expected tolerance range. A good estimate of
quantitative and reproducible [30]. the recovery rate of the analytical procedure is obtained
Another bromination recipe based on the use of potassium by comparing the median of 2-bromo(13 C3 )propenamide
bromide–potassium bromate (KBrO3 ) has recently been pro- peak areas in standards with the corresponding peak area in
posed in the literature [26]. This approach offers the advan- sample extracts. Depending on the type of cereal product
tage of eliminating the handling of elemental bromine, but it analyzed, the recovery rate was shown to vary between 40
has not been evaluated in the present study since low bromi- and 60%. No minimum value could be determined for an
nation yields have been reported under certain circumstances acceptable recovery of the internal standard, but, as a rule,
[26]. analysis was repeated if the recovery dropped below 40%.

3.2. Florisil clean-up 3.4. Method performance

Preliminary trials with the proposed method showed the The method was validated by replicate analysis of two
presence of some interfering co-extractives on the GC–MS commercial cereal products under repeatability and inter-
chromatograms. In order to partially eliminate this problem, mediate reproducibility conditions (Table 1). Both prod-
it was found necessary to purify potassium bromide and ucts were first analyzed seven times in parallel so as to
 A. Pittet et al. / J. Chromatogr. A 1035 (2004) 123–130 127

Abundance

6500

6000
m/z = 110 [13C2H381Br]+
5500
m/z = 154 [13C3H481BrNO]+
5000 m/z = 70 [C3H4NO]+
4500 m/z = 149 [C3H479BrNO]+
4000 m/z = 151 [C3H481BrNO]+
3500

3000

2500

2000

1500

1000

500

0
11.00 11.10 11.20 11.30 11.40 11.50 11.60 11.70 11.80 11.90 12.00
(a) Time

Abundance

6500
m/z = 110 [13C2H381Br]+
6000
m/z = 154 [13C3H481BrNO]+
5500
m/z = 70 [C3H4NO]+
5000
m/z = 149 [C3H479BrNO]+
4500
m/z = 151 [C3H481BrNO]+
4000

3500

3000

2500

2000

1500

1000

500

0
11.00 11.10 11.20 11.30 11.40 11.50 11.60 11.70 11.80 11.90 12.00
(b) Time

Fig. 1. GC–MS (EI) chromatograms of five selected ions from the acrylamide derivative (2-bromopropenamide) and internal standard derivative
(2-bromo(13 C3 )propenamide) after bromination of: (a) standard solution containing 100 ng of acrylamide; (b) rice product containing 17 ␮g/kg of acrylamide.

get information about within-day variation. The mean acry- dard deviations (R.S.D.(iR)) calculated at 6.8 and 6.2%, re-
lamide concentrations found in these samples were 17.6 spectively. The cereal product containing ca. 68 ␮g/kg of
and 70.6 ␮g/kg, with repeatability relative standard devia- acrylamide was also analyzed by LC–MS/MS according to
tions (R.S.D.(r)) of 8.6 and 5.7%, respectively. Additional the other method validated in our laboratory [22], and the
precision data were obtained by duplicate analysis of the result was almost identical at 69 ␮g/kg, showing an excel-
same two products on four different days, which gave prac- lent agreement between the two techniques.
tically the same concentration levels (means of 16.3 and The recovery rate of the method was determined by
67.5 ␮g/kg) with intermediate reproducibility relative stan- spiking a wheat semolina sample that had been shown to
128 A. Pittet et al. / J. Chromatogr. A 1035 (2004) 123–130

Abundance

70
280000

260000

240000

220000

200000

180000

160000 106

140000
149
120000

100000

80000

60000
53
40000 133

20000 81
60 87 93 99 113 122 141 155 165 173180
(a) 0
50 60 70 80 90 100 110 120 130 140 150 160 170
m/z-->

Abundance

73
320000

300000

280000

260000

240000

220000

200000
108
180000

160000
152
140000

120000

100000

80000
56
60000

40000 138
79
20000
67 85 94 101 115 123 131 145 159165173179
(b) 0
50 60 70 80 90 100 110 120 130 140 150 160 170
m/z-->

Fig. 2. Mass spectra of acrylamide derivatives: (a) 2-bromopropenamide; (b) 2-bromo(13 C3 )propenamide (internal standard).

contain <5 ␮g/kg of incurred acrylamide (n = 5). Test gave mean recoveries of 93 and 95% with relative standard
portions were fortified with acrylamide at concentrations deviation of 32.6 and 11.9%, respectively. A good accor-
of 5, 10, 50, 100, 200, and 500 ␮g/kg before the extraction dance of calibration slope and intercept values calculated
step, and at the same time isotope-labeled acrylamide was with data obtained from calibration standards and fortified
added to each flask to compensate for variations due to samples confirmed the compensation of matrix effects by
analyte ionization efficiency or injection volume. Taking (13 C3 )acrylamide (data not shown).
into account the correction of analyte loss by the internal The accuracy of the method was further investigated
standard, the overall recoveries of acrylamide in wheat through a comparative test with two independent laborato-
semolina ranged between 93 and 104% (Table 2). Repli- ries. Test portions of commercial and experimental cereal
cate experiments at spiking levels of 5 ␮g/kg (n = 8, four products (n = 9) containing acrylamide concentrations
different days) and 10 ␮g/kg (n = 15, five different days) ranging between 5 and 62 ␮g/kg were analyzed with the
 A. Pittet et al. / J. Chromatogr. A 1035 (2004) 123–130 129

Table 1
Acrylamide levels measured in two commercial cereal products under repeatability and intermediate reproducibility conditions
Rice product Three-cereal producta

Repeatability conditions
Acrylamide results (␮g/kg) Day 1 17, 18, 17, 16, 16, 20, 19 65, 74, 66, 71, 74, 69, 75
Mean (␮g/kg) 17.6 70.6
S.D.(r) (␮g/kg) 1.5 4.0
R.S.D.(r) (%) 8.6 5.7
Intermediate reproducibility conditions
Acrylamide results (␮g/kg) Day 2 16, 17 67, 62
Day 3 16, 14 67, 66
Day 4 17, 16 68, 68
Day 5 17, 17 65, 77
Mean (␮g/kg) 16.3 67.5
S.D.(iR) (␮g/kg) 1.1 4.2
R.S.D.(iR) (%) 6.8 6.2
a The product contained wheat, oats and barley.

Table 2
Recoveries of acrylamide from a wheat semolina sample spiked at levels between 5 and 500 ␮g/kg
Acrylamide added (␮g/kg) n Acrylamide found (␮g/kg) Mean recoverya (%) R.S.D. (%)

Range Mean

0 5b 2–5 4 – –
5 8c 6–11 9 93 32.6
10 15b 12–16 13 95 11.9
50 1 – 54 100 –
100 1 – 108 104 –
200 1 – 191 94 –
500 1 – 505 100 –
a Data corrected for the amount of acrylamide found in unspiked material and for the analyte loss by the internal standard.
b Analyses were performed on five different days.
c Analyses were performed on four different days.

Table 3
Comparison of acrylamide concentrations found in nine roller dried or extruded cereal products by three different laboratories
Product description Type Acrylamide content (␮g/kg)

In-house results Laboratory A Laboratory B

(GC–MS) (GC–MS) (GC–MS/MS)
Rice (roller dried) Commercial 18 16 25
Rice (roller dried) Commercial 10 8 16
Maize (extruded) Commercial 17 19 33
Wheat, oats and barley (roller dried) Commercial 62 44 66
Wheat base (roller dried, dry) Experimental 15 16 22
Wheat base (roller dried, wet) Experimental 14 12 15
Rice base (extruded, agglomerated) Experimental 9 7 10
Rice–soya base (roller dried) Experimental 5 4 5
Wheat base (roller dried) Experimental 5 4 5

present GC–MS method and sent to two laboratories in 4. Conclusion

Germany. Laboratory A used an in-house GC–MS method,
and laboratory B used an in-house GC–MS/MS method. Although the proposed method is similar in its princi-
As shown in Table 3, a good agreement was demonstrated ple to that recently published by Tareke et al. [17], the
between the three laboratories, particularly at very low use of Carrez purification before the bromination step and
concentration levels (<10 ␮g/kg). the inclusion of an additional Florisil clean-up step after
130 A. Pittet et al. / J. Chromatogr. A 1035 (2004) 123–130

derivatization have been shown very effective in removing [8] Report of the Analytical Working Group, 9 November 2002, JIF-
interfering co-extractives identified in some difficult ce- SAN Acrylamide in Food Workshop, 28–30 October 2002, Chicago,
USA.
real matrices. Taken together, these two purification steps [9] J. Tekel, P. Farkas, M. Kovác, Food Addit. Contam. 6 (1989) 377.
combined with bromination allow to achieve a very good [10] L. Castle, J. Agric. Food Chem. 41 (1993) 1261.
sensitivity, with a limit of detection and a limit of quantita- [11] US EPA, SW 846, Method 8032A, US Environmental Protection
tion estimated at 2 and 5 ␮g/kg, respectively. Moreover, the Agency, Washington, DC, 1996.
present work describes for the first time in-house validation [12] L.S. Bologna, F.F. Andrawes, F.W. Barvenik, R.D. Lentz, R.E. Sojka,
J. Chromatogr. Sci. 37 (1999) 240.
characteristics for acrylamide in a wide range of concen- [13] D.S. Barber, J. Hunt, R.M. LoPachin, M. Ehrich, J. Chromatogr. B
trations. The use of (13 C3 )acrylamide as internal standard 758 (2001) 289.
added to the samples at the initial step in the work-up pro- [14] E. Tareke, P. Rydberg, P. Karlsson, S. Eriksson, M. Törnqvist, Chem.
cedure was shown to significantly improve the accuracy and Res. Toxicol. 13 (2000) 517.
precision of measurements, allowing to obtain repeatability [15] K. Kawata, T. Ibaraki, A. Tanabe, H. Yagoh, A. Shinoda, H. Suzuki,
A. Yasuhara, J. Chromatogr. A 911 (2001) 75.
and intermediate reproducibility relative standard deviations [16] J. Rosén, K.-E. Hellenäs, Analyst 127 (2002) 880.
below 10%, even at low concentration levels (<20 ␮g/kg). [17] E. Tareke, P. Rydberg, P. Karlsson, S. Eriksson, M. Törnqvist, J.
Consequently, this method should be regarded as a new Agric. Food Chem. 50 (2002) 4998.
alternative for routine analysis of trace levels of acrylamide [18] A. Becalski, B.P.-Y. Lau, D. Lewis, S.W. Seaman, J. Agric. Food
in complex cereal-based foods. Chem. 51 (2003) 802.
[19] L. Hartig, Ch. Hummert, J. Buhlert, K. von Czapiewski, A. Schreiber,
Poster Presentation at the 17th Symposium on Liquid Chromatogra-
phy/Mass Spectrometry, Montreux, Switzerland, November 2002.
References [20] J.S. Ahn, L. Castle, D.B. Clarke, A.S. Lloyd, M.R. Philo, D.R.
Speck, Food Addit. Contam. 19 (2002) 1116.
[1] Swedish National Food Administration, Information About Acry- [21] F. Höfler, R. Maurer, S. Cavalli, GIT Labor-Fachzeitschrift 48 (2002)
lamide in Food, 24 April 2002, http://www.slv.se. 968.
[2] IARC Monographs on the Evaluation of Carcinogen Risk to Humans, [22] S. Riediker, R.H. Stadler, J. Chromatogr. A 1020 (2003) 121.
vol. 60, International Agency for Research on Cancer, Lyon, 1994, [23] H. Ono, Y. Chuda, M. Ohnishi-Kameyama, H. Yada, M. Ishizaka,
p. 389. H. Kobayashi, M. Yoshida, Food Addit. Contam. 20 (2003) 215.
[3] D.S. Mottram, B.L. Wedzicha, A.T. Dodson, Nature 419 (2002) 448. [24] C. Gertz, S. Klostermann, Eur. J. Lipid Sci. Technol. 104 (2002) 762.
[4] R.H. Stadler, I. Blank, N. Varga, F. Robert, J. Hau, P.A. Guy, M.-C. [25] M. Biedermann, S. Biedermann-Brem, A. Noti, K. Grob, P. Egli, H.
Robert, S. Riediker, Nature 419 (2002) 449. Mändli, Mitt. Lebensm. Hyg. 93 (2002) 638.
[5] Scientific Committee on Food, Opinion of the SCF on New Findings [26] S. Nemoto, S. Takatsuki, K. Sasaki, T. Maitani, J. Food Hyg. Soc.
Regarding the Presence of Acrylamide in Food, expressed on 3 July Jpn. 43 (2002) 371.
2002, http://www.europa.eu.int. [27] D.B. Clarke, J. Kelly, L.A. Wilson, J. AOAC Int. 85 (2002) 1370.
[6] Food Standards Agency, Study on Acrylamide in Food. Background [28] Commission Decision 2002/657/EC of 12 August 2002, Off. J. Eur.
Information and Research Findings, Press briefing of 17 May 2002, Commun. L221/8, 2002.
http://www.food.gov.uk. [29] A. Hashimoto, Analyst 101 (1976) 932.
[7] Health Implications of Acrylamide in Food, Report of the Joint [30] F. Andrawes, S. Greenhouse, D. Draney, J. Chromatogr. 399 (1987)
FAO/WHO Consultation, Geneva, Switzerland, 25–27 June 2002. 269.