Anal. Chem. 2003, 75, 5488-5494
Proton Transfer Reaction Mass Spectrometry, a Tool for On-Line Monitoring of Acrylamide Formation in the Headspace of Maillard Reaction Systems and Processed Food Philippe Pollien, Christian Lindinger, Chahan Yeretzian, and Imre Blank*
Nestle´ Research Center, Vers-chez-les-Blanc, 1000 Lausanne 26, Switzerland
The formation of acrylamide was measured in real time during thermal treatment (120-170 °C) of potato as well as in Maillard model systems composed of asparagine and reducing sugars, such as fructose and glucose. This was achieved by on-line monitoring of acrylamide released into the headspace of the samples using proton transfer reaction mass spectrometry (PTR-MS). Unambiguous identification of acrylamide by PTR-MS was accomplished by gas chromatography coupled simultaneously to electronimpact MS and PTR-MS. The PTR-MS ion signal at m/z 72 was shown to be exclusively due to protonated acrylamide obtained without fragmentation. In model Maillard systems, the formation of acrylamide from asparagine was favored with increasing temperature and preferably in the presence of fructose. Maximum signal intensities in the headspace were obtained after ∼2 min at 170 °C, whereas 6-7 min was required at 150 °C. Similarly, the level of acrylamide released into the headspace during thermal treatment of potato was positively correlated to temperature. Acrylamide (2-propenamide), widely used in polymer synthesis, water treatment, and electrophoretic separations, has been classified as “probably carcinogenic to humans”1,2 and is also neurotoxic via its epoxide metabolite glycidamide.3-5 High acrylamide amounts were reported in carbohydrate-rich foods that are subjected to high heat treatment, such as French fries, potato chips, and crispbread.6,7 Recently, we have identified the Maillard reaction as a major source of acrylamide promoting its formation in foodstuffs at high temperatures and under low moisture * To whom correspondence should be addressed. Telephone: +21/785-8607. Fax: +21/785-8554. E-mail:
[email protected]. (1) International Agency for Research on Cancer (IARC). Some industrial chemicals, IARC monographs on the evaluation for carcinogenic risk of chemicals to humans; IARC: Lyon, 1994; Vol. 60, pp 435-453, 389-433. (2) Tareke, E.; Rydberg, P.; Karlsson, P.; Eriksson, S.; To ¨rnqvist, M. Chem. Res. Toxicol. 2000, 13, 517-522. (3) Costa, L. G.; Deng, H.; Greggotti, C.; Manzo, L.; Faustman, E. M.; Bergmark, E.; Calleman, C. J. Neurotoxicology 1992, 13, 219-224. (4) Dearfield, K. L.; Douglas, G. R.; Ehling, U. H.; Moore, M. M.; Sega, G. A.; Brusick, D. J. Mutat. Res. 1995, 330, 71-99. (5) Tilson, H. A. Neurobehav. Toxicol. Teratol. 1981, 3, 445-461. (6) Rose´n, J.; Hellena¨s, K.-E. Analyst 2002, 127, 880-882. (7) Tareke, E.; Rydberg, P.; Karlsson, P.; Eriksson, S.; To ¨rnqvist, M. J. Agric. Food Chem. 2002, 50, 4998-5006.
5488 Analytical Chemistry, Vol. 75, No. 20, October 15, 2003
conditions, in particular, in the presence of free asparagine, which is directly converted into acrylamide.8 Various mechanistic principles underlying the formation of acrylamide in food are being discussed in the scientific community.8-10 HPLC/MS and GC/MS methods using labeled acrylamide as internal standard and special cleanup procedures have been used to obtain quantitative data.6-8 These results, however, show only the situation at a given reaction time. Because thermal processes in foods are generally complex and the formation of acrylamide via the Maillard reaction is directly linked to flavor generation, real-time on-line measurement tools are needed to better monitor, control, and optimize thermal processes. Currently, there is no information available in the scientific literature describing on-line, real-time techniques that are able to detect acrylamide during processing. Proton-transfer reaction mass spectrometry (PTR-MS) has been shown to be a suitable method for rapid and on-line measurements of volatile compounds of headspace samples.11-13 It combines a soft, sensitive, and efficient mode of chemical ionization with a quadrupole mass filter. The headspace gas is continuously introduced into the drift tube, which contains a buffer gas and a controlled ion density of H3O+. Volatile organic compounds (VOCs) that have proton affinities larger than water are ionized in the drift tube by proton transfer from H3O+, that is, VOC + H3O+ f [VOC + H]+ + H2O. The protonated VOCs are extracted from the drift tube and mass-analyzed in the quadrupole mass spectrometer.12 The four key features of PTRMS can be summarized as follows: (i) it is fast, and timedependent variations of headspace profiles can be monitored with a time-resolution of ∼0.1 s; (ii) the volatiles are not subjected to workup or thermal stress, and little fragmentation is induced by the ionization step; hence, mass spectral profiles closely reflect (8) Stadler, R. H.; Blank, I.; Varga, N.; Robert, F.; Hau, J.; Guy, Ph. A.; Robert, M.-C.; Riediker, S. Nature 2002, 419, 449-450. (9) Mottram D. S.; Wedzicha, B. L.; Dodson, A. T. Nature 2002, 419, 448449. (10) Gertz, C.; Klostermann, S. Eur. J. Lipid Sci. Technol. 2002, 104, 762-771. (11) Lindinger, W.; Hirber, J.; Paretzke, H. Int. J. Mass Spectrom. Ion Processes 1993, 129, 79-88. (12) Lindinger, W.; Hansel, A.; Jordan, A. Int. J. Mass Spectrom. 1998, 173, 191241. (13) Yeretzian, C.; Jordan, A.; Brevard, H.; Lindinger, W. In Flavor Release; Roberts, D. D., Taylor, A. J., Eds.; ACS Symposium Series 763; American Chemical Society: Washington, DC, 2002, pp 58-72. 10.1021/ac0344586 CCC: $25.00
© 2003 American Chemical Society Published on Web 09/05/2003
Figure 1. Experimental setup for on-line analysis of acrylamide from Maillard systems and from potatoes. The bottom frame shows the temperature-controlled oven with the reaction vessel. The top frame shows the experimental setup for dynamic sampling and PTR-MS analysis of VOCs released into the headspace of the reaction vessel. See text for detailed explication.
genuine headspace distributions; (iii) mass spectral intensities can be transformed into absolute headspace concentrations; and (iv) it is not invasive. All of these features make PTR-MS particularly suited to investigate fast dynamic processes, such as formation of aroma and volatile contaminants in Maillard reactions. We report here on the use of PTR-MS as a sensitive on-line method for real-time detection of acrylamide in the headspace of thermally treated samples using Maillard model systems and potato as examples. EXPERIMENTAL SECTION Materials. L-Asparagine, D-fructose, D-glucose, D-glucose monohydrate, and acrylamide (99.5%) were from Aldrich/Fluka (Buchs, Switzerland). Potatoes were from the local market. They were cut into 2-mm-thick slices. One part was oven-dried (Binder FD 53 APT-line, Tuttlingen, Germany) at 50 °C for 5 h (with active N2 circulation), and the other part was used as such.
Sample Preparation. Asparagine (0.33 g, 2.5 mmol) was drymixed with fructose (0.45 g, 2.5 mmol), glucose (0.45 g, 2.5 mmol), or glucose monohydrate (0.495 g, 2.5 mmol) and placed in a reaction vessel that was held at a defined temperature in an oven (Figure 1). After adding the reactants, the formation of volatile compounds was monitored by PTR-MS. Alternatively, freshly sliced potato (2 g) or dried potato (0.5 g) were placed in the reaction vessel and heated as described above at 150 °C and 170 °C, respectively. Potatoes were homogenized under inert gas (nitrogen), followed by oven-drying at 50 °C for 5 h in a nitrogen atmosphere, finally obtaining a white powder. On-Line Analysis. The complete setup for on-line analysis of released volatiles is shown in Figure 1. The reaction vessel consisted of a glass vessel with a 240-mL volume. It was placed inside a temperature-controlled oven and had on the top cover a gas inlet and gas outlet for purging the headspace and a feedAnalytical Chemistry, Vol. 75, No. 20, October 15, 2003
5489
Figure 2. Schematic diagram of the GC/EI-MS/PTR-MS setup for chemical characterization of PTR-MS ion peaks. VOCs trapped on Tenax are thermally desorbed in the automatic thermal desorber (ATD) and separated by GC. The effluent from the GC is split and simultaneously injected into EI-MS and PTR-MS for detection and identification. See text for detailed explanation.
through for a thermocouple to measure the temperature inside the vessel. Gas flows were controlled with two flow controllers, FC1 and FC2 (mass flow controller, Brooks Instruments B. V., Holland), and the lines were heated with heating wires (active control of all temperatures with thermocouples). FC2 maintained the purging gas flow (zero-air) at 670 standard cubic centimeters per minute (sccm). The air was preheated to 130 °C and introduced through 0.25-in. Teflon tubing into the hot reaction vessel. The headspace inside the vessel was swept with the purge gas and left the reaction vessel through 0.25-in. stainless steel tubing. The exit-side tubing was heated to 120 °C to avoid condensation. The temperature and humidity of the sample gas was reduced by admixture of dry air at 55 °C (position T2) to maintain simple reaction kinetics between the primary ion and the reactant gas in the reaction chamber (drift tube) of the PTRMS, thus ensuring a proper functioning of the drift tube. FC1 controlled the flow of dilution gas and maintained it at 5250 sccm, except for the trials with dried potato, during which FC1 was at 1150 sccm. The 0.25-in. Teflon line for the dilution gas was kept at 55 °C. For all experiments discussed here, the full mass spectrum from m/z 21 to 220 was monitored by PTR-MS on-line, with a 0.2-s dwell time per mass. Only a small amount of sample gas needed to be introduced into the drift tube. Therefore, most of the 5950 sccm gas at T2 (or 1820 sccm in the experiments with dried potato) was released through the exhaust line, while only 14 sccm was introduced through a needle valve into the drift tube. This needle valve controlled the split of the gas flow at T2 between the exhaust and the drift tube. The pressure in the line, between the needle valve and the drift tube, was in the range of a few millibars. The line was made of deactivated fused-silica tubing of 0.53-mm i.d. Part of the sample gas released through the exhaust line was drawn through a Tenax trap to collect the VOCs generated during the reaction. The constant trapping flow through the Tenax was adjusted to 50 sccm by a membrane pump combined with a flow controller. Proton-Transfer Reaction Mass Spectrometry. The PTRMS was from Ionicon Analytik (Innsbruck, Austria) and consisted of four main components: an ion source, a drift tube, a mass 5490
Analytical Chemistry, Vol. 75, No. 20, October 15, 2003
analyzer (quadrupole), and an ion detector/amplifier. The hollow cathode ion source and the drift tube were pumped differentially by a 50 L/s turbo molecular pump (Pfeiffer Vacuum, TMH 071P, Assler, Germany), maintaining a pressure of 2 mbar in the drift tube (under operating conditions). The quadrupole mass spectrometer (QMS 422, Balzers, Switzerland) and detector (secondary electron multiplier) were pumped by a 200 L/s turbo pump (Pfeiffer Vacuum, TMH 261, Assler, Germany), maintaining a vacuum of 3 × 10-5 mbar. The headspace gas was sampled continuously and introduced into the drift tube, where it reacted under defined conditions with a controlled density of H3O+ primary ions following proton-transfer reactions. Nitrogen or air can be used as buffer gas in the drift tube. The protonated VOCs were extracted with an electric field out of the drift tube and massanalyzed by the quadrupole mass spectrometer. The drift tube was specifically designed to reach high sensitivity, induce little fragmentation, and permit absolute quantification of VOCs. To accomplish these targeted specifications, the generation of the primary H3O+ ions on one hand and the chemical ionization process on the other hand were spatially and temporally separated and individually optimized. For these experiments, we used a newly developed fast drift tube, which increased the speed and reduced the background signal of the experiments, therefore increasing the sensitivity. Gas Ghromatography/Mass Spectrometry (GC/MS). For unambiguous identification of VOCs, we have developed an experimental setup in which a gas chromatograph (Trace 2000 Series) from Finnigan (Les Ulis, France) is coupled via a split to two MS detectors, an electron-impact (EI)-MS (Automass Multi MS, Finnigan) and a PTR-MS equipped with the novel fast-drift tube. The analytical approach, which is briefly outlined here, can be divided into two steps (a more complete discussion of the GC/ EI-MS/PTR-MS setup will be given elsewhere). In a first step and simultaneously with the on-line analysis, a portion of the sample gas passed through a Tenax cartridge for trapping the VOCs (Figure 1). In a second step, analysis of the trapped volatiles was performed off-line in a setup as shown in Figure 2. Volatiles were thermally desorbed onto an automatic thermal desorber (ATD,
Figure 3. PTR-MS time-intensity ion traces of six VOCs, including m/z 72 for acrylamide, generated from the asparagine/fructose Maillard reaction system at 150 °C.
Perkin-Elmer; ATD 400) from the Tenax (250 °C, 10 min) with a helium flow of 20 sccm. They were cryofocused at -30 °C and injected from the cold trap (250 °C, 3 min) into the GC. The GC was equipped with a DB-Wax capillary column (J&W Scientific, Folsom, CA) of 60-m length, 0.53-mm i.d., and 1-µm phase thickness. Helium was used as carrier gas at 5 sccm. The column was kept at 20 °C for 5 min, increased at 4 °C/min to 220 °C, and maintained for 10 min at 220 °C. The column outlet was split into the two mass spectrometers, the EI-MS and the PTR-MS, for simultaneous analysis of the GC-separated compounds. The split introduces a problem related to differences in the carrier gas and pressure between GC and PTR-MS. GC/EI-MS works with He carrier gas at 1 × 10-4 to 1 × 10-5 mbar, while the drift tube of the PTR-MS runs with air or N2 as buffer gas at 2.0 mbar. To allow a proper coupling of GC with PTR-MS, the effluent gas from the GC column was mixed with a 10-fold volume of N2 gas (or alternatively with air). A flow controller (FC2) controlled admixture of N2 to the He carrier gas. In addition, to avoid back-diffusions of N2 to the EI-MS and to stabilize the pressure at the entrance of the PTR-MS, a 1-m capillary line was introduced between the split and the admixture of N2 (deactivated fused-silica tubing with 0.25-mm i.d.). On the EI-MS side, a 1-m capillary (deactivated fused-silica tubing with 0.1 mm i.d.) reduced the pressure to 1 × 10-4 to 1 × 10-5 mbar and, in addition, ensured that the signals on both MS are synchronized. Identification of the VOCs was performed on the basis of GC/ EI-MS using commercial databases on EI fragmentation14 and, in the case of acrylamide, by injection of a reference compound. The retention index and fragmentation pattern unequivocally indicates the compositional purity of a given ion. Since the GC outlet was injected in parallel into the PTR-MS, the corresponding PTR-MS ion distribution could be unequivocally assigned to the compound identified by EI-MS. Analyzing the whole GC chromatogram, one can assign each single ion peak of the PTR-MS profile to the corresponding compound. In case several compounds contribute (14) Wiley Registry of Mass Spectral Data, 7th ed.; McLafferty, F., Ed.; Database, with NIST Spectral Data; Chichester: 2003.
to the same PTR-MS ion peak, one can determine the relative contributions of each compound. RESULTS AND DISCUSSION Experimental Setup. On-line measurement of headspace samples was performed by PTR-MS. The configuration used in this study is shown in Figure 1. It consisted of a reaction unit with precise temperature and flow controls which was coupled to the PTR-MS via a heated interface. Because PTR-MS is a onedimensional method that characterizes compounds by their mass, an identification of VOCs by on-line PTR-MS analysis alone is questionable. For unambiguous identification of the released VOCs, the system was completed with a headspace sampling device for trapping volatile compounds onto a Tenax cartridge. This allowed the composition of the sample gas to be determined by off-line GC/EI-MS/PTR-MS (Figure 2). The retention index in combination with the EI-fragmentation pattern unequivocally indicates the compositional purity of a given ion, and allows assigning the ion peaks in the PTR-MS spectrum. For all samples, the 240-mL headspace volume was sampled at a high gas flow rate of 670 sccm to monitor dynamically the release of VOCs from the reaction systems. The sample gas was then diluted at T1, not only to reduce the humidity and temperature of the sample gas, but also to decrease the concentration of VOCs, thus avoiding saturation of the PTR-MS. Considering that the model reaction systems generated very high concentrations of VOCs during the first few minutes of the reaction, a high gas flow rate of 5250 sccm was needed. The dried potato produced less VOCs, and therefore, an admixture of 1150 sccm dilution gas was sufficient to obtain an adequate signal intensity. For PTR-MS analysis, only 14 sccm was introduced into the fast drift tube, while the remaining 5906 sccm sample gas volume from the model Maillard systems (or the 1806 sccm from the dried potato) was discarded into the exhaust at T2. For a given time window of the reaction, a portion of the discarded gas flow was trapped on a Tenax cartridge for subsequent off-line GC/MS analysis. Maillard Model Reactions. The formation of acrylamide was studied in Maillard systems composed of asparagine and reducing Analytical Chemistry, Vol. 75, No. 20, October 15, 2003
5491
Figure 4. Identification of acrylamide by GC/EI-MS/PTR-MS from the Maillard reaction system asparagine/fructose at 150 °C. The bottom trace shows the total ion current (TIC) in GC/EI-MS. It corresponds to the Tenax trapped headspace of the reaction system, shown in Figure 3, for the time window from 5 to 10 min. The inset shows the EI-MS of acrylamide eluting at 49.6 min. The top trace shows the GC/PTR-MS trace at m/z 72 obtained simultaneously with the TIC. The PTR-MS ion signal at m/z 72 is exclusively related to the GC peak eluting at 50 min and corresponds to acrylamide.
sugars. The total headspace obtained from the Maillard system asparagine and fructose at 150 °C (Asn/Fru/150) was monitored on-line over the mass range from m/z 21 to 220. Figure 3 shows a selection of six ion masses from more than 200 that were monitored to illustrate some typical time-intensity patterns. For instance, the ion trace at m/z 47, which corresponds exclusively to formic acid, had the highest headspace concentration over the first 25 min. While all ion traces shown in Figure 3 have been fully characterized via off-line GC/EI-MS/PTR-MS, here we focus only on the ion trace at m/z 72. It corresponds to protonated acrylamide, ([M + H]+), which was found to be homogeneous, as ascertained by off-line GC/MS analysis (see below). For unambiguous identification of the VOCs contributing to the PTR-MS ion traces in Figure 3, 50 sccm of the headspace gas was drawn through a Tenax during the reaction time-window between 5 and 10 min (Figure 1). The trapped VOCs were thermally desorbed on the ATD, separated by GC, and analyzed in parallel by EI-MS and PTR-MS (Figure 2). The bottom trace in Figure 4 shows the total ion counts (TIC) of the GC/EI-MS. The compound eluting at 49.6 min corresponds to a peak that showed identical retention index and EI mass spectrum as the acrylamide reference compound. Furthermore, the EI spectrum, shown as an inset to the GC/EI-MS, matches the spectrum of acrylamide of the EI database. Finally, the GC/PTR-MS ion signal at m/z 72 (top trace) is related solely to the compound eluting at 49.6 min. It should be noted that acrylamide does not fragment in PTR-MS and appears only at the protonated parent mass m/z 72. Combining all these, we can conclude that (i) the PTR-MS trace at m/z 72 in Figure 3 is, indeed, acrylamide and (ii) there is no other compound in the headspace interfering with the mass at m/z 72. 5492 Analytical Chemistry, Vol. 75, No. 20, October 15, 2003
Figure 5. Formation of acrylamide from asparagine/glucose, asparagine/glucose‚H2O, and asparagine/fructose reacted at 150 °C for 50 min.
The formation of acrylamide in Maillard model systems could, therefore, be studied by on-line monitoring of the PTR-MS ion trace at m/z 72. The formation of acrylamide from asparagine was favored in the presence of fructose, as compared to glucose or glucose monohydrate (Figure 5). At 150 °C, the acrylamide yields were, on the basis of the peak height, higher in the Asn/Fru sample by a factor of 6-7. These results clearly suggest fructose to be more efficient than glucose in generating acrylamide from asparagine. However, the release of acrylamide into the headspace may also depend on the structural properties of the reaction system. On the other hand, no differences were found between glucose and glucose monohydrate in generating acrylamide at 150 °C. In all cases, the formation of acrylamide at 150 °C was very rapid and essentially instantaneous: the Maillard precursors
Figure 6. Formation of acrylamide from asparagine/fructose as affected by the reaction temperature: 120; 150; and 170 °C. The traces at 150 and 120 °C were magnified by a factor of 10 and 60, respectively.
generated acrylamide already after a few minutes of reaction time. The yield and reaction rate of acrylamide formation was very much temperature-dependent (Figure 6). Significantly higher amounts of acrylamide were found at 170 °C, as compared to 150 and 120 °C. The maximum headspace concentration at 170 °C was 10 times as high as at 150 °C. This is in good agreement with the data recently reported, which shows increasing amounts of acrylamide generated at elevated temperatures with 170 °C as maximum and a rapid decrease at higher temperatures.8,9 Our measurements also indicate that the maximum yield of acrylamide was reached already after 2 min at 170 °C, whereas 6-7 min was required at 150 °C. From this we can conclude that the yields and reaction rates were markedly lower at 150 °C, as compared to 170 °C, and even lower at 120 °C. However, it should be noted that elevated temperatures also favor the release of acrylamide from the matrix into the headspace. Therefore, the signal intensities observed result from both formation yield and release, both of which are influenced by temperature and structural properties of the reaction system. Thermally Processed Potato. The applicability of the PTRMS approach for monitoring on-line the formation of acrylamide was evaluated in real food systems using thermally treated potato as an example. The mass trace at m/z 72 shown in Figure 7 indicated the presence of acrylamide in the headspace obtained by heating potato slices at 170 °C for 70 min. The mass at m/z 72 was found to be homogeneous without contamination from other volatile compounds (Figure 8) using the off-line coupling method described above. Retention index and EI mass spectrum were identical to those of the acrylamide reference compound, and only one peak with the mass at m/z 72 was detected by PTR-MS. The EI spectrum of the compound eluting at 49.6 min (see inset) was conclusively identified by the Wiley EI database as acrylamide. The slight differences in the EI spectra among the insets shown in Figures 4 and 8 are due to the large differences in absolute concentrations of the respective compounds in the ion source, which may distort the spectral pattern. The formation of acrylamide at 170 °C showed a curve similar to that of Asn/Fru/150 °C (Figure 5). After a rapid increase, a
Figure 7. PTR-MS time-intensity profile for the formation of acrylamide from dried potato slices at 170 °C.
broad maximum was observed at 6-10 min of reaction time, followed by a slow decline of the curve. However, the amounts of acrylamide in the headspace were very low compared to the Maillard model systems, which is most likely due to the lower concentration of the precursors (reducing sugars and asparagine). According to literature data,15 fresh potato contains ∼1000 mg/ kg of free asparagine. Taking into account the high water content of ∼80%,16 an estimated 2.5 mg (19 µmol) of asparagine was available in 0.5 g dried potato for generating acrylamide. Thus, the concentration of asparagine as major precursor of acrylamide was about 130-times lower than in the Maillard model experiments. Despite the low precursor amounts in the experiment with potato, these data show that PTR-MS is sufficiently sensitive to monitor the formation of acrylamide under food processing conditions. Analogous data were obtained with fresh raw potato: 2 g potato contains ∼2 mg free asparagine, which corresponds to 15 µmol direct precursor for acrylamide formation. The measurements with fresh potato were complicated by the release of large amounts of (15) Martin, F. L.; Ames, J. M. J. Agric. Food Chem. 2001, 49, 3885-3892. (16) Souci, S. W.; Fachmann, W.; Kraut, H. Food Composition and Nutrition Tables, 6th ed.; CRC Press: Boca Raton, 2000; pp 639-641.
Analytical Chemistry, Vol. 75, No. 20, October 15, 2003
5493
Figure 8. Identification of acrylamide by GG/EI-MS/PTR-MS, from dried potato at 170 °C. The bottom trace shows the total ion current (TIC) in GC/EI-MS. It corresponds to the Tenax trapped headspace of the reaction system, shown in Figure 7, for the time window from 0 to 30 min. The inset shows the EI-MS of the compound eluting at 49.6 min. The top trace shows the corresponding GC/PTR-MS trace at m/z 72 obtained simultaneously with the TIC. The PTR-MS ion signal at m/z 72 is exclusively related to the GC peak eluting at 50 min and corresponds to acrylamide.
water vapor into the headspace during the first few minutes of the reaction (drying of the potato). This saturated the ion source with H2O, led to a strong reduction of the primary H3O+ ion concentration in the drift tube, and hindered proper operation of the PTR-MS. Nevertheless, we were able to conclusively observe acrylamide formation also from fresh potato (data not shown). CONCLUSION We demonstrate for the first time the feasibility of monitoring on-line and in real time the processing contaminant acrylamide in Maillard reaction samples and food systems using PTR-MS. The experimental setup is based on direct headspace sampling of volatiles released from the reaction system. Unequivocal identification of acrylamide was accomplished by a novel experimental approach, which couples GC analysis to two parallel MS detectors, an EI-MS and a PTR-MS. These features make PTRMS a valuable tool for studying the formation of volatile compounds in thermal food processing. Since PTR-MS allows moni-
5494 Analytical Chemistry, Vol. 75, No. 20, October 15, 2003
toring released acrylamide in real time and at high time resolution (1-s time resolution is feasible if one monitors exclusively acrylamide), it holds promise to become a valuable approach for process control and process optimization during food production. This may provide the basis for limiting the formation of acrylamide while concurrently maintaining the desired color and flavor of food products. For the time being, signal intensities measured in the headspace by PTR-MS cannot be linked to the actual concentration of acrylamide in the product. Experiments are in progress describing this correlation and will be reported elsewhere. ACKNOWLEDGMENT We thank Elizabeth Prior for linguistic proofreading and Marcel A. Juillerat and Richard H. Stadler for critical discussions. Received for review May 1, 2003. Accepted July 30, 2003. AC0344586
