Kinetics of acrylamide formation and elimination during heating of an asparagine-sugar model system

Article abstract of DOI:10.1021/jf051197n

The kinetics of acrylamide (AA) was analyzed by heating a simple model system consisting of asparagine and glucose, fructose, or sucrose (0.01 M, pH 6) at temperatures between 140 and 200 °C. The AA concentration appeared to be the net result of simultaneous formation and elimination. A general kinetic model describing the AA yield was identified, and kinetic parameters were obtained by nonlinear regression on the nonisothermally derived data. On the basis of kinetic parameters, the AA formation appeared to proceed faster and to be more temperature sensitive in the asparagine-glucose than in the asparagine-fructose model system. The AA elimination kinetics, on the other hand, was similar. Significantly less AA was formed in the asparagine-sucrose model system as compared to the model systems with glucose or fructose.

Full text of DOI:10.1021/jf051197n

J. Agric. Food Chem. 2005, 53, 9999−10005 9999
Kinetics of Acrylamide Formation and Elimination during
Heating of an Asparagine Sugar Model System
−
WENDIE L. CLAEYS, KRISTEL DE VLEESCHOUWER, AND MARC E. HENDRICKX*
Center for Food and Microbial Technology, Department of Microbial and Molecular Systems,
Faculty of Bioscience Engineering, Katholieke Universiteit Leuven, Kasteelpark Arenberg 22,
B-3001 Heverlee, Belgium
The kinetics of acrylamide (AA) was analyzed by heating a simple model system consisting of
asparagine and glucose, fructose, or sucrose (0.01 M, pH 6) at temperatures between 140 and 200
°C. The AA concentration appeared to be the net result of simultaneous formation and elimination.
A general kinetic model describing the AA yield was identified, and kinetic parameters were obtained
by nonlinear regression on the nonisothermally derived data. On the basis of kinetic parameters, the
AA formation appeared to proceed faster and to be more temperature sensitive in the asparagine-
glucose than in the asparagine-fructose model system. The AA elimination kinetics, on the other
hand, was similar. Significantly less AA was formed in the asparagine-sucrose model system as
compared to the model systems with glucose or fructose.
KEYWORDS: Acrylamide; kinetics; asparagine; glucose; fructose; sucrose
INTRODUCTION
In this paper, a kinetic model allowing the estimation of the
AA yield in terms of processing time and temperature is
identified. Several research groups previously identified the
importance of the amino acid asparagine and the Maillard
reaction in the formation cascade of AA. Other probably minor
pathways have been proposed as well, including acrolein and
acrylic acid (11-14). Because of the inherent complexity of a
food matrix, where many factors are involved that act one upon
another, we chose to analyze kinetics under idealized conditions
in a closed asparagine-glucose model system. Whereas no or
only trace amounts of AA have been reported in a model system
consisting of glucose and an amino acid other than asparagine
(11, 12, 15), a variety of monosaccharides (e.g., glucose,
deoxyglucose, ribose, glyceraldehyde, and glyoxal) can generate
AA from asparagine (14). Therefore, in addition to the aspar-
agine-glucose model system, AA formation/elimination kinetics
were also analyzed in asparagine-fructose and asparagine-
sucrose model systems. Glucose, fructose, and sucrose are the
most common sugars of the potato (constituting >90% of all
sugars in a potato tuber), and potato products yield the highest
AA amounts when fried or baked. Additionally, studying
kinetics in those three model systems allows comparison
between the performance of an aldohexose and a ketohexose
sugar and the performance of a disaccharide and a monosac-
charide.
Acrylamide (AA, CAS 79-06-1) is an industrial chemical used
since the mid-1950s in the production of polyacrylamides for
different technical applications (e.g., in the paper and textile
industry, as soil conditioners, and in wastewater treatment).
Recently, relatively high levels of AA have been measured in
different food and food products (1-3). AA appeared to be
formed during high temperature processes such as frying,
baking, and roasting of foods, especially of carbohydrate-rich
foods. The neurotoxicity of AA in humans is well-known from
occupational and accidental exposures. In addition, experimental
studies with AA in animals have shown reproductive, genotoxic,
and carcinogenic properties (4-6).
Since the finding of AA in food, intensive research in several
areas is ongoing and focuses mainly on the different mechanisms
of AA formation, the toxicology of AA (e.g., bioavailability,
intake evaluation, and margins of exposure), the development
of accurate analytical techniques (e.g., the development of a
uniform method applicable to all food matrices and of cheap
and rapid screening methods), and the effect of the food
composition on the AA yield (e.g., variation due to cultivar and
storage temperature of potato tubers). Several strategies for
reducing AA have been put forward and are under investigation
(7, 8). To predict and to control the amount of AA formed,
additionally, kinetics of AA as a function of process and product
variables need to be known. Reported data are, however, mainly
qualitative in nature, and until now, quantitative data on the
possible interaction between time, temperature, and AA forma-
tion were lacking (9, 10).
MATERIALS AND METHODS
Model System and Heat Treatment. Model systems were prepared
by dissolving 0.01 M L-asparagine (g99.5%, Sigma-Aldrich, Bornem,
Belgium) and 0.01 M D-glucose, D-fructose, or D-sucrose (g99%,
Sigma-Aldrich) in a 0.05 M citrate buffer with a pH of 6.
To eliminate as much side phenomena as possible during heating
(such as fluctuations in water activity due to water evaporation and
* To whom correspondence should be addressed. Tel: +32-16-32.15.85.
Fax: +32-16-32.19.60. E-mail: marc.hendrickx@biw.kuleuven.be.
10.1021/jf051197n CCC: $30.25
© 2005 American Chemical Society
Published on Web 11/24/2005

10000 J. Agric. Food Chem., Vol. 53, No. 26, 2005
Claeys et al.
with Casn, Csugar, and CD the concentration of asparagine, sugar,
and D, respectively, and t the treatment time. It is clear that,
next to isotopically labeled AA, elimination of AA can also be
deduced from the kinetic model fitted on the net formation data
of AA.
Only a fraction of the asparagine and the sugar is converted
into AA (10). Because the asparagine and the sugar are reacted
at an equal concentration, the concentration of asparagine (Casn
)
can be assumed to be equal to the concentration of the sugar
(Csugar) and equal to CR:
dCAA
) kF(CR)2 - kECAA
(2)
dt
Figure 1. Typical temperature−time profiles of samples heated in closed
reactors.
However, whereas both asparagine and sugar undergo the
Maillard reaction, the sugar is consumed by caramelization
reactions as well. Moreover, in the Maillard reaction, the loss
of sugar has been reported to be faster than the loss of amino
acid, which could be explained by the regeneration of asparagine
from the initial condensation products and the possible formation
of diglucosylamine (18). On the basis of these arguments,
asparagine can be considered in excess as compared to the sugar.
As a result, the AA yield can be described by a first-order
formation/first-order elimination kinetic model and eq 2 becomes
(7, 19)
absorption of oil), which affect AA formation, samples were heated in
hermetically closed reactor tubes (inox, 8 mm × 100 mm, custom-
made). After the samples were heated in a thermostated oil bath at
temperatures between 140 and 200 °C for different heating times, the
samples were immediately cooled in ice water to stop any further
reaction. Because the samples went through a heating phase, the
temperature profile of each sample was registered in the reactor tubes
in order to obtain an accurate analysis of formation and elimination
kinetics of AA. An example of typical temperature profiles is given in
Figure 1. Similar profiles were used in each assay, for which
temperature-time data were measured at regular time intervals (2 s)
using thermocouples (type T, Thermo Electric, Balen, Belgium)
connected to a datalogger (TM 9616, Ellab, Norfolk, England).
Analysis of AA. The AA content was determined by a gas
chromatography-mass spectrometry (GC-MS) method (without previ-
ous derivatization of AA), described by Biedermann et al. (16). After
AA extraction and further cleanup of the sample, 1 µL was injected
cool-on-column on an HP-INNOWax column (30 m × 250 µm i.d.,
0.25 µm ft, equipped with a 0.5 m × 530 µm i.d. precolumn of
deactivated fused silica, Agilent Technologies, Diegem, Belgium)
coupled to a quadrupole mass spectrometer operating in positive
chemical ionization with He as the carrier gas and CH4 as the ionization
gas (5973 inert GC-MS system, Agilent Technologies).
dC
AA ) kFCR - kECAA
(3)
dt
Kinetic models are not mechanistic models; they describe the
rates of formation of substances in terms of the rate-limiting
processes.
When kinetics are analyzed under isothermal conditions
(constant temperature), k values can be assumed constant and
the above equations are easily integrated. However, because
heating of the samples involved nonisothermal conditions
(variable temperature), the integrated effect of temperature on
the reaction rate constant has to be taken into account.
Generally, the effect of temperature on the reaction rate
constant k can be expressed by the Arrhenius relation (20), in
which the temperature dependence of k is quantified by the
activation energy Ea (J/mol) according to
The AA concentration at (m/z 72) was quantified by adding an
internal standard, methacrylamide (m/z 86), at the beginning of the
sample preparation step. By comparing this internal standard with a
second internal standard, i.e., butyramide (m/z 88), which was added
to the samples before injection, we could account for losses during
analysis.
Ea
1
1
T
k ) kref exp
-
(4)
(
)
]
DATA ANALYSIS
[
R Tref
Kinetic Parameter Estimation. Simplified, the formation
of AA in food can be given by the following scheme, with kF
and kE the formation and elimination rate constants, respectively,
at the temperature studied (7)
with R the universal gas constant (8.314 J/mol, K), T the
temperature concerned (K), and kref the reaction rate constant
at reference temperature Tref. This equation applies for each rate
constant involved in the kinetic model describing AA formation/
elimination.
kF
kE
R
9
8 AA
9
8
D
After implementation of eq 4 in eqs 2 or 3, the resulting
differential equations were solved by numerical integration
(Euler) of the registered temperature-time profile of each
sample, and kinetic parameters describing AA formation and
elimination (kFref, EaF, and kEref, EaE, respectively) could be
estimated by nonlinear regression (Gauss-Newton algorithm).
Data analysis was performed by the statistical software package
SAS (v8, Cary, NC).
Evaluation of the Kinetic Model. The performance of the
model was evaluated in terms of output statistics that evaluate
the quality of fitting and parameter estimation, i.e., the sum of
squares (SS) and the standard errors (SE) associated with the
parameter estimates. Commonly, the R2 value is used to express
the quality of fit of a linear model. Use of the R2 for evaluating
reactants
AA - protein complex/
AA degradation product/...
Because asparagine and sugar are present at an equimolar
level in the model system, the formation of AA from these
reactants can be modeled by a second-order reaction. Bieder-
mann et al. (17), monitoring AA elimination by D3-AA,
proposed (pseudo-) first-order kinetics for describing the AA
elimination. Consequently, the net AA content (CAA) can be
described by following equation:
dC
AA ) kF(Casn ‚ Csugar) - kECAA
(1)
dt

Acrylamide Formation/Elimination Kinetics
J. Agric. Food Chem., Vol. 53, No. 26, 2005 10001
reactions with asparagine heated at 180 °C. Becalski et al. (23),
on the other hand, observed a lower relative yield of AA from
a dry asparagine hydrate-sugar system when sucrose (0.48)
was used as compared to glucose (1) andsin contrast to our
observationssa higher AA yield (1.8) when fructose was used.
Also, in the model experiments of Biedermann et al. (24),
fructose appeared to be twice as effective in promoting AA
formation as glucose when added to dry potato (5%) and heated
at 150 °C for 30 min. Likewise, Ashoor and Zent (25) measuring
browning in an amino acid-sugar model system (1:1, pH 9)
observed higher browning with fructose as compared to glucose.
Sucrose did not yield detectable browning, probably because
the disaccharide was not hydrolyzed under the conditions studied
(121 °C/10 min). Yaylayan et al. (26), on the other hand,
performing model studies at high temperatures by pyrolysis-
GC-MS, found sucrose to be significantly more efficient as
glucose or fructose in forming AA with asparagine at 350 °C.
Table 1. Relative AA Content (%) in Equimolar Asparagine−Sugar
Model Systems (0.01 M, pH 6) Heated at Different Temperatures
T (°
C)
glucose
fructose
sucrose
140
160
180
200
100
100
100
100
100.0
±
±
±
±
11.1a
28.8
5.0
45.8
±
±
±
±
12.2
78.4
63.5
54.7
40.8
41.6
51.0
16.9
18.5
20.4
2.5
a Averages
± standard deviations of nine samples taken after different treatment
times at the temperature concerned.
nonlinear models, however, can be highly misleading. One of
the problems with the R2 definition is that it requires the presence
of an intercept, which most nonlinear models do not have.
Therefore, a coefficient of determination closely corresponding
to R2 was defined (21)
SSresidual
The difference between observations can be partly explained
by an influence of both temperature and time on the relative
activities of sugars toward the production of AA, indicating the
importance of kinetic data. However, also, other factors such
as the pH and the physical state of the system can influence the
relative reactivity of AA precursors. The latter was demonstrated
by Robert et al. (27) who monitored on-line AA formation in
equimolar asparagine-sugar mixtures by proton transfer reaction
mass spectrometry. Under low-moisture conditions, the molec-
ular mobility of the AA precursors appeared to be the determin-
ing factor, which is linked to the melting behavior and the
release of crystallization water of the reaction system. A melting
point of 126 °C was measured by DSC for D-fructose and of
157 °C for D-glucose (27). Sucrose has a melting point of 190-
192 °C (with decomposition) (28). In liquid systems, the
molecular mobility is no longer a limiting factor and the relative
reactivity of the sugar will be determined by its chemical
reactivity. This largely explains the discrepancy between our
results and those reported in the literature, since most studies
reported are performed in dry systems.
pseudo - R2 ) 1 -
(5)
SScorrected
In addition, residual plots were checked for the absence of
trends or correlations. Normality of residuals was evaluated
based on the Shapiro-Wilk statistic W, which is calculated for
the null hypothesis that the input values are a random sample
from a normal distribution. When Pr < W is smaller than 0.05,
the null hypothesis is rejected at a 5% significance level (22).
The “bias” of the prediction of the model was graphically
evaluated by a scatterplot of the observed vs the predicted
values. If the model is appropriate for the data, the data points
roughly follow the line of equity (i.e., with a slope of 1).
RESULTS AND DISCUSSION
Relative Reactivity of the Sugars. The AA formation in the
asparagine-fructose and -sucrose model systems relative to
the formation in the asparagine-glucose model system at each
temperature studied is given in Table 1. This table, however,
gives only a general notion on the difference in relative reactivity
of the sugars at different temperatures and does not take the
combined effect of temperature and heating time into account.
At 140 °C and to a lesser extent at 160 °C, glucose and fructose
appeared to generate a similar amount of AA. Overall, the
reactivity of the sugars in the AA formation reaction decreases
from glucose over fructose to sucrose and with increasing
temperature. From a chemical point of view, aldohexose sugars
are generally more reactive than ketohexose sugars, explaining
the higher amounts of AA formed with glucose as with fructose.
The lower AA yields from sucrose as compared to glucose and
fructose can be attributed to the fact that sucrose is a nonre-
ducing sugar and, as such, cannot react with asparagine directly
to form AA but first needs to undergo decomposition to reactive
carbonyl compounds. Decomposition of sucrose will be en-
hanced at higher temperatures. Nevertheless, no increase in
relative reactivity of sucrose is observed at higher temperatures.
Despite the fact that hydrolysis of sucrose results in a molar
ratio of sugar to amino acid of 2:1, AA formation from sucrose
remains 40-50% relative to AA formation from glucose at
higher temperatures. In terms of molar ratio, however, a 0.5 to
an equimolar ratio of asparagine to glucose is reported to be
favorable for AA formation (15, 23).
In addition, the relative rates of browning of glucose and
fructose are reported to depend on the extent with which the
reaction mixture is buffered. In unbuffered media, the rate of
browning of fructose with amino acids is reported to be greater
than that of glucose, whereas in buffered media, fructose browns
more slowly than glucose (29).
AA Formation/Elimination Kinetics. The influence of
temperature on the AA concentration has already been dem-
onstrated (11, 23, 30, 31). AA starts to be formed above ∼120
°C, and dependent on the model system studied and the duration
of heating, the AA yield reaches a maximum around 170-180
°C. Knowledge on the combined effect of heating time and
temperature on AA formation, on the other hand, is scarce.
Figure 2A-C shows the AA formation in the asparagine-
glucose/fructose/sucrose model systems heated at different
temperatures for different treatment times. After prolonged
heating, a decrease of the AA content was observed, which can
be attributed to AA elimination becoming predominant over AA
formation. AA is mainly reactive through its double bond and
can react as an electrophile by 1,4-addition to nucleophiles such
as SH- or NH2-groups in biomolecules. AA can thus readily
react with various other components present or formed in the
food or model system (15). The elimination of AA is also
ascribed to AA degradation or polymerization (32).
In literature, different observations are reported concerning
the relative reactivity of the sugars for generating AA. Stadler
et al. (12) investigating the role of different carbohydrates in
the formation of AA observed D-fructose, D-galactose, lactose,
and sucrose to release AA with comparable yields in model
AA formation and elimination was modeled by means of a
second- and a first-order reaction, respectively, but also by two
first-order reactions. Both candidate kinetic models appeared

10002 J. Agric. Food Chem., Vol. 53, No. 26, 2005
Claeys et al.
Figure 2. Net formation of AA as a function of heating time at 140 ([),
160 (2), 180 (b), and 200 °C (9) in an equimolar model system (0.01
Figure 3. Scatter plot of observed AA values (ppb) and AA values (ppb)
predicted by the second-order formation/first-order elimination (
O) and
M, pH 6) consisting of asparagine and (A) glucose, (B) fructose, and (C)
sucrose. The full and broken lines connect the AA values predicted by
the second-order formation/first-order elimination and by the first-order
formation/first-order elimination kinetic model, respectively.
by the first-order formation/first-order elimination ( ) kinetic model in an
equimolar model system (0.01 M, pH 6) consisting of asparagine and (A)
glucose, (B) fructose, and (C) sucrose. Full lines have a slope of 1.
b
the activation energies EaF and EaE could be calculated by
nonlinear regression. The kinetic parameters together with their
SEs are summarized in Table 2. Kinetic parameters were similar
for both candidate kinetic models. Although they have different
units, the kFref values differed approximately by a factor of 102,
which can be explained by the fact that it concerns a second-
order reaction at one hand and a first-order reaction at the other
hand. On the basis of results and evaluation of the kinetic
models, one model cannot be considered superior to the other.
Nevertheless, considering the argumentation given under Data
Analysis and because of its simplicity, the first-order formation/
first-order elimination kinetic model seems to be favorable.
The approach used in this paper is a rather pragmatic one, in
which the overall formation kinetics are considered rather than
to describe the net formation of AA in the model systems
accurately, as indicated by the good fit of the model on the
data (Figure 2A-C) and confirmed by the high correlation
between observed and predicted AA concentrations (Figure
3A-C). It is of course clear that the actual reaction mechanism
of AA formation is much more complicated. However, the
remarkable good fit suggests that the rate-determining step in
the cascade of formation and elimination reactions is a bimo-
lecular (second-order) reaction or a bimolecular reaction with
one reactant in large excess (pseudo-first-order) and a mono-
molecular (first-order) reaction, respectively.
On the basis of the temperature-time profile and AA value
of each sample, the reaction rate constants kFref and kEref and

Acrylamide Formation/Elimination Kinetics
J. Agric. Food Chem., Vol. 53, No. 26, 2005 10003
Table 2. Kinetic Parametersa Describing AA Formation/Elimination in
tion were similar. Correspondingly, Biedermann et al. (24)
observed a similar elimination of D3-AA at 150 °C for 30 min
when glucose or fructose was added to a dry potato model
system. AA thus seems to be formed more rapidly and to be
more temperature sensitive when asparagine reacts with glucose
as compared to fructose. Because AA elimination proceeds
similarly, a higher net content of AA is generated in the model
system with glucose.
Equimolar Asparagine
between 140 and 200
−Sugar Model Systems (0.01 M, pH 6) Heated
°
C
Tref
)
160
°
C
glucose
fructose
sucrose
second-order formation/first-order elimination
-
-
-1
kFref
kEref
(
(
×
×
10 3 M 1 min
)
42.90
105.0
161.06
157.33
0.969
±
±
±
±
2.33 a
9.2 a
30.20
111.0
139.10
148.23
0.917
±
±
±
±
2.27 b
12.4 a
5.37 b
6.57 a
63.40
±
±
±
±
22.50 a,b
-
-1
10 3 min
)
508.4
48.88
26.47
0.734
0.475
130.6 b
3.23 c
1.40 b
EaF (kJ/mol)
EaE (kJ/mol)
pseudo-R2
Pr < W
3.84 a
4.68 a
When sucrose is the reactive sugar in the system, the rate
constant of the AA elimination is significantly higher as
compared to the other two model systems studied, which
corresponds to a lower AA yield. Activation energies for AA
formation and elimination are significantly lower implying that
the rates of AA formation/elimination are not highly affected
by a change in temperature. The model, however, does not take
into account the hydrolysis of sucrose to glucose and fructose.
The difference in kinetic parameter values with the ones
estimated for the two other model systems can be due to the
fact that other reaction steps in the process are rate determining.
Nevertheless, the ratio kFref/kEref increases from the model system
with sucrose over the one with fructose to the one with glucose,
which correlates with an increased reactivity of the sugars in
the AA formation reaction.
0.634
0.644
first-order formation/first-order elimination
-
-1
kFref
kEref
(
(
×
×
10 3 min
)
)
0.451
111.1
168.25
167.21
0.975
±
±
±
±
0.023 a
8.9 a
0.303
111.4
140.81
151.91
0.912
±
±
±
±
0.024 b 0.601
±
±
±
±
0.214 a,b
138.9 b
3.15 c
10 3 min
13.1 a
5.84 b
7.04 a
508.7
48.50
26.95
0.729
0.611
-
-1
EaF (kJ/mol)
EaE (kJ/mol)
pseudo-R2
Pr < W
3.80 a
4.30 a
1.53 b
0.934
0.619
a Values of the same parameter with a different letter are significantly different
based on 95% asymptotic confidence intervals. Number of data points 35;
pseudo-R2
SS(residual)/SS(corrected total).
)
)
1 −
details of the complex chemistry. Nevertheless, use of a simple
reaction order for complex formation pathways can be useful
for modeling chemical changes during processing, when knowl-
edge of pure chemistry or mechanism of the reaction is of less
importance. By means of multiresponse modeling, more than
one reactant can be followed and modeled. In this approach,
various reaction steps (reactants, intermediates, and products)
are monitored to gain insight in the reaction mechanism.
Recently, Wedzicha et al. (10) presented a kinetic model for
the formation of AA in potato and rye products by multiresponse
modeling of reducing sugar, amino acid, asparagine, and AA
concentrations with time. The kinetic mechanism features a rate-
limiting intermediate and additional reaction steps of this
intermediate, which are competitive with respect to AA forma-
tion. Similarly to our model, a pathway representing physical
and/or chemical loss of AA accounts for the reduction of AA
measured at high temperatures or long reaction times. Knol et
al. (9) proposed a kinetic model for AA formation/elimination
in an aqueous asparagine-glucose model system by multire-
sponse modeling as well, in which AA was considered as an
intermediate of the Maillard reaction.
Irrespectively of the kinetic model considered, correlation
coefficients between kFref and kEref and between EaF and EaE were
high (0.94 and 0.74-0.83, respectively) for the model systems
with glucose and fructose. In the model system with sucrose, a
high correlation was observed between kFref and kEref as well
(∼0.94) but also between EaF and kFref and kEref (∼0.78). The
correlation between parameters is a consequence of the process
of measuring dependent variables at a finite number of data
points over a limited range of the independent variable and the
subsequent fitting process. The correlation coefficient does not
give information about possible relationships in the chemistry
or about the physical relationships between parameters (33). If
the correlation is high, it may be that a parameter is redundant,
that the data points do not span a sufficiently wide range of x
values, or that too few data points have been collected (34).
The activation energies quantifying the temperature sensitivity
of the AA formation and elimination in the asparagine-glucose
and -fructose model systems are around 140-160 kJ/mol. On
the basis of asymptotic 95% confidence intervals, the kinetic
parameters describing AA formation in the asparagine-glucose
and in the asparagine-fructose model system differed signifi-
cantly, whereas kinetic parameters describing the AA elimina-
For all three systems studied, kEref was much higher than kFref
.
This does not necessarily indicate that the elimination of AA
proceeds much faster than the formation. Next to the relative
AA formation and elimination reaction rate constants, the extent
of AA formation and of AA elimination depends on the amount
of “precursors” available for the reaction. First, a sufficient level
of AA and possibly also of other reaction products must be
formed before elimination can predominate over formation. In
the asparagine-sugar model systems, numerous reactions of,
for example, Maillard and caramelization occur, which depend
on the nature and the molar concentration of the precursor sugar
and which could determine the extent of the AA elimination
reaction.
Biedermann and Grob (35) studying AA elimination by D3-
AA at temperatures between 120 and 200 °C observed an
increased elimination above 160 °C when 40% sucrose was
added to wheat flour. It was suggested that the enhanced
elimination at higher temperatures might be due to decomposi-
tion products of sucrose. The addition of fructose increased the
AA elimination as well but to a higher extent than sucrose and
already at 120 °C.
Figure 4 depicts the temperature dependency of the reaction
rate constants kF and kE in the different model systems and is
based on eq 4. Arrhenius curves related to the AA elimination
are similar in the glucose and fructose model systems, whereas
Arrhenius curves of AA formation in both systems intersect
around 140 °C. Below this temperature, the relative reactivity
of the sugars will change, assuming extrapolation of results.
The low Ea values associated with AA formation/elimination
in the sucrose model system are reflected by the nearly
horizontal Arrhenius curves.
Note that AA kinetics were studied in a buffer to eliminate
as much as possible variation between the model systems studied
and a change of pH due to reactions taking place. The rate of
the Maillard reaction (and consequently of the AA formation
reaction) depends on the rate at which the sugar ring opens to
the reducible, open-chained form, which increases with increas-
ing pH (29). In the Maillard reaction, H+ ions are formed that
decrease the pH of the system. As the pH falls, the rate of the
Maillard reaction thus decreases (the optimum pH for the
Maillard reaction is >7). A drawback of using a buffer is that

10004 J. Agric. Food Chem., Vol. 53, No. 26, 2005
Claeys et al.
only becomes feasible by combining AA formation/elimination
kinetics with kinetic data of safety and quality aspects.
LITERATURE CITED
(1) FSA. Food Standard Agency Study of Acrylamide in Foods
Background Information & Research Findings; Press briefing
17.05.02; Food Standard Agency: United Kingdom, 2002; http://
www.food.gov.uk/multimedia/pdfs/acrylamideback.pdf.
(2) SNFA. Analytical Methodology and SurVey Results for Acryla-
mide in Foods; Swedish National Food Administration, 2002;
http://www.slv.se/templates/SLV_Pages4547.aspx.
(3) FDA. Acrylamide in Food; Food and Drug Administra-
tion: United States, 2004; http://www.cfsan.fda.gov/∼lrd/
pestadd.html#acrylamide.
(4) Rude´n, C. Acrylamide and cancer risksExpert assessments and
the public debate. Food Chem. Toxicol. 2004, 42, 335-349.
(5) Friedman, M. Chemistry, biochemistry, and safety of acrylamide.
A review. J. Agric. Food Chem. 2003, 51, 4504-4526.
(6) Tareke, E. Identification and origin of potential background
carcinogens: Endogenous isoprene and oxiranes, dietary acryl-
amide. Ph.D. Thesis, Department of Environmental Chemistry,
Stockholm University, Sweden, 2003.
(7) Claeys, W.; De Vleeschouwer, K.; Hendrickx, M. Quantifying
the formation of carcinogens during food processing: Acryla-
mide. Trends Food Sci. Technol. 2005, 16, 181-193.
(8) Taeymans, D.; Wood, J.; Ashby, P.; Blank, I.; Studer, A.; Stadler,
R.; Gonde´, P.; Van Eijck, P.; Lalljie, S.; Lingnert, H.; Lindblom,
M.; Matissek, R.; Mu¨ller, D.; Tallmadge, D.; O’Brien, J.;
Thompson, S.; Silvani, D.; Whitmore, T. A review of acryla-
mide: An industry perspective on research, analysis, formation,
and control. Crit. ReV. Food Sci. Nutr. 2004, 44, 323-347.
(9) Knol, J.; Van Loon, W.; Linssen, J.; Ruck, A.-L.; Van Boekel,
M.; Voragen, A. Toward a kinetic model for acrylamide
formation in a glucose-asparagine reaction system. J. Agric. Food
Chem. 2005, 53, 6133-6139.
(10) Wedzicha, B.; Mottram, S.; Elmore, J.; Koutsidis, G.; Dodson,
A. Kinetic models as a route to control acrylamide formation in
food. In Chemistry and Safety of Acrylamide in Food; Friedman,
M., Mottram, D., Eds.; Springer Science+Business Media: 2005;
pp 235-253.
(11) Mottram, D.; Wedzicha, B.; Dodson, A. Acrylamide is formed
in the Maillard reaction. Nature 2002, 419, 448-449.
(12) Stadler, R.; Blank, I.; Varga, N.; Robert, F.; Riediker, S.
Acrylamide from Maillard reaction products. Nature 2002, 419,
449-450.
(13) Yasuhara, A.; Tanala, Y.; Hangel, M.; Shibamoto, T. Gas
chromatographic investigation of acrylamide formation in brown-
ing model systems. J. Agric. Food Chem. 2003, 51, 3999-4003.
(14) Zyzak, D.; Sanders, R.; Stokanovic, M.; Tallmadge, D.; Eberhart,
B.; Ewald, D.; Gruber, D.; Morsch, T.; Strothers, M.; Rizzi, G.;
Villagran, M. Acrylamide formation mechanism in heated foods.
J. Agric. Food Chem. 2003, 51, 4782-4787.
(15) Leufve´n, A.; Lingnert, H. Factors Influencing Acrylamide
Formation in Food ProcessingsIntroductory Model Experiments
Performed at SIK; Public report of the Swedish Institute for
Food and Biotechnology; Swedish Institute for Food and
Biotechnology: 2003; http://www.bi.se/dokument/news/nyheter/
030331_Akrylamid_Projektrapport.pdf.
(16) Biedermann, M.; Biedermann-Brem, S.; Noti, A.; Grob, K. Two
GC-MS methods for the analysis of acrylamide in foodstuffs.
Mitt. Lebensmittelunters. Hyg. 2002, 93, 638-652.
(17) Biedermann, M.; Biedermann-Brem, S.; Noti, A.; Grob, K.
Methods for determining the potential of acrylamide formation
and its elimination in raw materials for food preparation, such
as potatoes. Mitt. Lebensmittelunters. Hyg. 2002, 93, 653-667.
(18) Martins, S.; Van Boekel, M. A kinetic model for the glucose/
glycine Maillard reaction pathways. Food Chem. 2005, 90, 257-
269.
Figure 4. Arrhenius plot of AA formation (full line) and elimination (broken
line) in equimolar model systems (0.01 M, pH 6) consisting of asparagine
and glucose (b), fructose (2), or sucrose (9) based on (A) second-
order formation/first-order elimination kinetics and (B) first-order formation/
first-order elimination kinetics.
buffers can affect the rate of browning. Particularly phosphate
buffer, a buffer mostly used to control the pH during the
Maillard reaction, has been shown to increase browning and
glycine or glucose loss with increasing concentration at pH 5.6
and 100 °C or at pH 7 and 25 °C (36, 37). No glycine loss or
browning was observed at pH 7 and 25 °C when citrate buffer
was used (36). At pH 3.5 and 121 °C, on the other hand, a
higher browning was observed when citrate buffer was used as
compared to phosphate buffer (38). The acceleration of the
Maillard reaction in the presence of buffers can be explained
by the buffer reducing the fall in pH value under alkaline
conditions and by an interaction of the buffer with the reactants
of the Maillard reaction, but more research on this topic is
needed. Remark that the extent of the effect of the buffer on
the Maillard reaction rate depends on the temperature as well
as on the pH studied. The effect of the buffer ions on the
Maillard reaction rate, however, can be of practical importance
since food usually contain sugars, amino acids, and both
phosphates and organic acid salts.
In this paper, AA kinetics was identified by means of
simplified model systems. However, not only the nature of the
reactants but also the molar ratio, the aw, the food matrix, the
heating equipment, etc. will influence the AA content in
products. Factors influencing the AA content and AA kinetics
are interrelated. To obtain a full characterization of AA, the
formation/elimination rate needs to be quantified not only in
terms of temperature but also in terms of reaction variables such
as pH, aw, concentration of reactants, etc. A next step involves
the extrapolation of kinetics from model systems to in situ
conditions or real food products.
Before altering processes or foodstuffs to decrease the AA
content, the impact of the modification on other food safety
issues (e.g., microorganisms and mutagens) and on food quality
(e.g., color, odor, flavor, and texture) should be considered. This
(19) Smith, J. Chemical Engineering Kinetics, 3rd ed.; McGraw-Hill
Int. Ed.: London, United Kingdom, 1981.

Acrylamide Formation/Elimination Kinetics
J. Agric. Food Chem., Vol. 53, No. 26, 2005 10005
(20) Arrhenius, S. About the reaction rate of the inversion of
nonrefined sugar at souring. Z. Phys. Chem. 1889, 4, 226-248.
(21) Schabenberger, O. Nonlinear Regression with the SAS
System; SAS Institute, Inc.: Cary, NC, 1998; http://home.nc.
rr.com/schabenb/SASNlin.htm#Calculating%20a%20R2-
Type%20Measure.
(31) Rydberg, P.; Eriksson, S.; Tareke, E.; Karlsson, P.; Ehrenberg,
L.; To¨rnqvist, M. Investigations of factors that influence the
acrylamide content of heated foodstuffs. J. Agric. Food Chem.
2003, 51, 7012-7018.
(32) Stadler, R.; Verzegnassi, L.; Varga, N.; Grigorov, M.; Studer,
A.; Riediker, S.; Schilter, B. Formation of vinylogous compounds
in model Maillard reaction systems. Chem. Res. Toxicol. 2003,
16, 1242-1250.
(22) SAS Institute Inc. SAS Procedures Guide, Version 6, 3rd ed.;SAS
Institute Inc.: Cary, NC, 1990.
(23) Becalski, A.; Lau, B.; Lewis, D.; Seaman, S. Acrylamide in
foods: Occurrence, sources and modeling. J. Agric. Food Chem.
2003, 51, 802-808.
(24) Biedermann, M.; Noti, A.; Biedermann-Brem, S.; Mozzetti, V.;
Grob, K. Experiments on acrylamide formation and possibilities
to decrease the potential of acrylamide formation in potatoes.
Mitt. Lebensmittelunters. Hyg. 2002, 93, 668-687.
(33) Johnson, M.; Faunt, L. Parameter estimation by least-squares
methods. Methods Enzymol. 1992, 210, 1-37.
(34) Motulsky, H.; Ransnas, L. Fitting curves to data using nonlinear
regression: A practical and nonmathematical review. FASEB J.
1987, 1, 365-374.
(35) Biedermann, M.; Grob, K. Model studies on acrylamide forma-
tion in potato, wheat flour and corn starch; ways to reduce
acrylamide contents in bakery ware. Mitt. Lebensmittelunters.
Hyg. 2003, 94, 406-422.
(25) Ashoor, A.; Zent, J. Maillard browning of common amino acids
and sugars. J. Food Sci. 1984, 49, 1206-1207.
(26) Yaylayan, V.; Wnorowski, A.; Perez Locas, C. Why asparagine
needs carbohydrates to generate acrylamide. J. Agric. Food
Chem. 2003, 51, 1753-1757.
(27) Robert, F.; Vuataz, G.; Pollien, P.; Saucy, F.; Alonso, M.-I.;
Bauwens, I.; Blank, I. Acrylamide formation from asparagine
under low-moisture Maillard reaction conditions. 1. Physical and
chemical aspects in crystalline model systems. J. Agric. Food
Chem. 2004, 52, 6837-6842.
(28) Becalski, A.; Lau, B.; Lewis, D.; Seaman, S.; Hayward, S.;
Sahagian, M.; Ramesh, M.; Leclerc, Y. Acrylamide in French
fries: Influence of free amino acids and sugars J. Agric. Food
Chem. 2004, 52, 3801-3806.
(29) Davies, K.; Labuza, T. The Maillard reaction: Application to
confectionery products. In Confectionery Science; Zeigler, G.,
Ed.; Penn State University Press: 1997; pp 35-66.
(30) Tareke, E.; Rydberg, P.; Karlsson, P.; Eriksson, S.; To¨rnqvist,
M. Analysis of acrylamide, a carcinogen formed in heated
foodstuffs. J. Agric. Food Chem. 2002, 50, 4998-5006.
(36) Bell, L. Maillard reaction as influenced by buffer type and
concentration. Food Chem. 1997, 59, 143-147.
(37) Potman, R.; van Wijk, T. Mechanistic studies of the Maillard
reaction with emphasis on phosphate-mediated catalysis. In
Thermal Generation of Aromas; Parliament, T., McGorrin, R.,
Ho, C., Eds.; American Chemical Society: Washington, DC,
1989; pp 182-195.
(38) Saunders, J.; Jervis, F. The role of buffer salts in nonenzymic
browning. J. Sci. Food Agric. 1966, 17, 245-249.
Received for review May 24, 2005. Revised manuscript received October
2, 2005. Accepted October 8, 2005. This research was supported by
the Belgian National Fund for Scientific Research Flanders (NFWO)
and the Flemish Institute for the Promotion of Scientific-Technological
Research in Industry (IWT).
JF051197N