Packaging materials for foodstuffs or consumer usage must not alter the flavor or odor of the substance being packaged. Examples of packaging materials include the paper used to wrap cigarettes, the paper cartons used in orange juice or milk and the Polyethylene (PE) often used for caps to seal soft drink bottles. Sometimes, due to differences in the raw materials or changes in the production processes, chemical inconsistencies can be found in these packaging materials. For example, a common issue with packaging materials is encountered when they emit off-odors.
Traditionally acceptance of batches is often based on sensory panels. However this is very subjective and time consuming. Ideally an objective and fast screening system should be located directly at the production floor. This is a typical application for chemical sensors.
In this study, paper and PE samples classified as acceptable (no-odor) and unacceptable (off-odor) were used to train a mass spectral based chemical sensor. This sensor incorporates well-known mass spectrometry technology with multivariate data analysis. The mass spectral fingerprint obtained with the sensor was used to create multivariate classifi cation models that were later used to classify unknown samples as acceptable or unacceptable.
Unknown samples for these two applications were successfully classified using KNN classifi cation models. In addition the results of the PE samples were compared to those obtained by the sensory panel. These results are encouraging, due to the level of accuracy and shorter analysis times compared to traditional (e.g. GC/MS) techniques.
In a third application we analyzed ppb amounts of TPGDA in milk and orange juice cartons with the Headspace ChemSensor. This was possible using the MSD in selected ion monitoring mode.
