Batch Cross-Contamination Monitoring With An Electronic Nose

Food and beverage processing systems are under constant scrutiny for cleanliness, accuracy, and repeatability. In order to meet the quality requirements, production batches must be reproducible and conform to specifications, and batch cross-contamination must be avoided

Contamination detection is currently performed using traditional analytical instruments such as GC, GC-MS, and sensory assessment (olfactory, visual, etc.). However, these methods are very time-consuming, and human testing can be unpleasant or raise safety concerns for panelists.

Electronic nose analyzers based on ultrafast gas chromatography can also be applied to measure low levels of contaminants in food and beverage products. This application note presents analyses conducted with the HERACLES E-Nose (Alpha M.O.S., Toulouse, France) on various teas, some of which were contaminated with low concentrations of fruit flavors (from 0.5 to 10 ppm).

Figure 1 - HERACLES system.

The instrument measures odors and chemical compounds. The specificity of electronic nose analyzers is that their working principle mimics human olfaction. Instead of measuring and identifying the various compounds responsible for the flavor, as many analytical techniques do, an electronic nose captures the global profile of a flavor/odor. Based on ultrafast gas chromatography technology, the HERACLES system (Figure 1) integrates two short columns of different polarities placed in parallel and associated with two flame ionization detectors (FIDs) to generate two chromatograms simultaneously. It achieves volatile organic carbon (VOC) analysis in less than 60 sec, with a sensitivity reaching ppb concentrations (trap using an absorbent support). It can be used to directly analyze liquid, gas, or solid sample analysis via liquid or headspace injection mode, without any prior preparation. The E-Nose is used in the pharmaceutical industry, chemicals and petrochemicals, and in other areas to achieve organoleptic or chemical quality control, quantify low amounts of compounds in products, monitor the manufacturing process, and screen large numbers of products during the development stages.

Aroma analysis in tablets

Tea samples containing various amounts of four different flavors were collected on different batches and analyzed by the HERACLES E-Nose. The purpose of the analysis was to differentiate the various levels of the same flavor in tea and set up a calibration curve to establish the detection threshold. For each of the four contaminating flavors (peach, apple, grape, and strawberry), six concentrations were analyzed (0.1, 0.5, 1, 2.5, 5, and 10 ppm).

To run the analysis, samples were heated in a vial. A fraction of the headspace generated was then injected into the system. The analysis parameters are shown in Table 1.

Table 1    -    Analysis parameters of the HERACLES E-Nose

The aroma compounds were separated by ultrafast gas chromatography in less than 45 sec (Figure 2) in contrast to dozens of minutes using classical gas chromatography. A data library developed using the chromatograms shows the most significant retention times and peak areas.

Figure 2 - Superimposed chromatograms of tea contaminated with apple flavor (column 1).

Differentiating batches contaminated with varying amounts of the same flavor

Figure 3 - PCA chart of tea batches contaminated with apple flavor.

In order to visualize the differences or similarities between various batches that are free of contaminant or are contaminated with a flavor, data were processed using principal components analysis (PCA) (Figure 3). Based on multivariate treatment, PCA provides a 2-D representation of various sample odors and chemical composition. The mathematical processing consists of determining two axes accounting for the maximum amount of information. The percentage shown below each axis represents the part of the information brought by this axis. In this example, the horizontal axis accounts for the main part of the information (94.87%). The distances between two clusters on the map provide information on the similarity or difference between products. The discrimination index is calculated by dividing the surface of groups by the surface intergroups. It gives the discrimination quality based on an indication of the surface between groups. When groups overlap each other, the discrimination index is negative. When groups are distinct, the discrimination index is positive; the higher this value, the better the discrimination. The arrow in Figure 3 shows the ranking of samples by increasing concentration of apple flavor.

The PCA map (Figure 3) of tea batches contaminated with apple flavor shows a clear differentiation of all contamination concentrations, with a high discrimination index (>90). Samples were classified based on their apple flavor contamination levels. The detection threshold of apple flavor in green tea was below 0.1 ppm since this concentration was clearly identified.

Table 2    -    SNR and estimated detection threshold by contaminating flavor

To precisely determine the detection threshold of flavor contaminant in tea batches, the signal-to-noise ratio (SNR) was calculated at a contamination level of 0.1 ppm in tea, based on selected peaks. This gave the expected detection threshold limit of the HERACLES E-Nose (Table 2).

Determination of contaminant concentration in a tea batch

Figure 4 - PLS model for the quantification of apple contaminant in tea batches.

In order to quantify the amount of contaminant contained in a batch, a calibration model consisting of partial least squares (PLS) was constructed. This graph presents the known value on the X-axis and the concentration measured by the E-Nose on the Y-axis. Each sample was analyzed in three repetitions in order to check the repeatability of the measurement. The higher the correlation coefficient, the more reliable the model.

The PLS model (Figure 4) constructed from tea samples contaminated with apple flavor shows a very high level of correlation (99.57%). This model is thus valid for quantitative analysis and prediction.

Conclusion

From an analytical point of view, the HERACLES E-Nose is a reliable (RSD <5%), sensitive (low detection limit, i.e., in the ppb range), and fast (analysis in seconds) instrument. In the food and beverage industry, electronic nose analyzers are used in:

  • Quality control departments to perform batch-to-batch consistency testing of raw materials and finished products and to ensure process conformity
  • R&D departments working on product development to speed up the screening of formulations, select the most consistent raw material, improve the acceptability of the product, and guarantee the authenticity of the product by measuring its characteristic aroma.

Mrs. Bonnefille is Communication Manager, Alpha M.O.S., 20 ave. Didier Daurat, 31400 Toulouse, France; tel.: +33 5 62 47 64 55; fax: +33 5 61 54 56 15; e-mail: communication@alpha-mos.com. This article is based on a poster presented at Pittcon® 2007, Chicago, IL, U.S.A.

Comments