A black box or a red strawberry? The use of SIFT MS in aroma analysis.

Strawberry fruit is worldwide one of the most popular fruits, typically praised for its flavour. Flavour can be subdivided into taste and aroma, the latter being the main driver for consumer perception and appreciation. Strawberry has one of the most complex aromas, consisting of many different components comprised of different chemical classes. To measure aroma gas chromatography – mass spectrometry (GC-MS) is currently considered the golden standard. However, this technique is quite time consuming and, given the large sample sizes needed to account for biological variation, not applicable for high-throughput analyses. This thesis, therefore, aimed to apply a new technique called selected ion flow tube – mass spectrometry (SIFT-MS) for the measurements of strawberry aroma.

SIFT-MS was originally developed to study ion-molecule reactions in interstellar clouds in the 1980s. Since then it has expanded its scope to more earthly topics, including food science. Unlike most analytical MS techniques SIFT-MS does not require calibration and can be applied in real-time. Furthermore, it can measure the sample directly as no sample pre-treatment or concentration is necessary. Since SIFT-MS lacks a chromatographic step, the impact of this reduced information content was studied in an exploratory study by reconstructing datasets of reduced information content from GC-MS data, mimicking the output of SIFT-MS. The effect of this reduced information content was studied to see how this would affect the discriminating power of the aroma analyses compared to the standard deconvolution method in which both the chromatographic and mass spectral information are used. Both the ‘total ion chromatogram’ and the ‘cumulative mass’ spectrum successfully discriminated large differences such as differences between cultivars, but smaller differences such as intra-cultivar variation proved difficult to detect and in both approaches the reproducibility was less compared to the deconvolution method.

After the exploratory data study, the feasibility of SIFT-MS for strawberry aroma measurements and its strengths and weaknesses were investigated and compared to GC-MS. SIFT-MS and GC-MS were used in parallel to measure the aroma of strawberries at different stages of ripening. Compared to GC-MS, the SIFT-MS analyses were 11 times as fast. Both techniques showed similar results; however, the limit of detection of SIFT-MS seemed lower since the data points were more evenly distributed over the principal component analysis (PCA) plot, whereas the GC-MS samples showed more overlap. The most important volatiles per ripening stage (red or green) were tentatively identified based on the GC-MS data. Since GC-MS requires sample pre-treatment, which in general is a blending step with salt, the pre-treated samples were measured with SIFT-MS as well to also study the effect of sample treatment on analysis. This showed that artefacts may possibly be introduced by the treatment, but this needs further research.

One of the weaknesses of SIFT-MS is its kinetic library, which is far smaller than the widespread GC-MS spectral library of the National Institute of Standards (NIST). Three strawberry-specific compounds – ethyl isovalerate, mesifuran and trans-2-hexenyl acetate – were studied and their reaction rate constants and product ion distribution were determined and added to the kinetic library for further use in this dissertation. The knowledge of these parameters will allow for the identification and quantification of these compounds, and eventually to the calculation of odour activity values. 

Currently, quantification of SIFT-MS mass spectra is done using the m/z values that do not suffer from overlapping fragments. This often results in the use of only one m/z value and this makes quantification of complex samples almost impossible. Therefore, a new data analysis methodology that uses the full spectra of all three reagent-ions was developed. The method is based on fitting the measured SIFT-MS spectrum to a calculated spectrum that is based on previously selected compounds with known SIFT-MS spectrum. The method was tested and its applicability discussed based on the analysis of different samples. The best results were obtained in samples of simple and known composition; the methodology is not suited for analysis of complex or unknown samples. The selection of compounds at the start and fragments that share m/z values complicate the analysis and may lead to incorrect identification and quantification of volatiles. The applicability of the method to compound identification was tested but proved merely statistical and had no realistic value (the compounds identified by the software were not present in the actual sample, of which the composition was known). Another issue is the yet limited SIFT-MS kinetic library, which it is not as complete as for GC-MS so product ions of unknown compounds can be incorrectly assigned to known compounds. Furthermore, the library does not contain much information on secondary products (H2O adducts) since their formation is humidity dependent. These secondary products may have an effect on the product ion distribution and intensities, though. This can be overcome by studying the compounds of interest under the conditions of the sample, since then not only the secondary products will be known but also the ratio of primary to secondary products. Further modifications of the method are required to decrease false positive identifications of volatiles that are not present in the headspace. Finally, tentative odour activity values were calculated for the sensory most relevant strawberry compounds; however, these values have to be interpreted with caution, since the quantification is dependent on the compound selection at the start and overlapping product ions might be assigned to the wrong compound.

This research has demonstrated that SIFT-MS can be successfully applied in the field of postharvest biology and technology. It can serve as a fast technique to classify fruits based on their spectral fingerprint according to cultivar, ripening stage, moment of harvest and in the future probably also storage condition. Disadvantages are the lack of chromatography and the incompleteness of the library. Another means of introducing a separation step to SIFT-MS would be the use of a thermal desorption unit, which would separate volatile compounds based on their boiling points and introduce them to the SIFT-MS consecutively. With regards to the completeness of the SIFT-MS kinetic library, new compounds can be added by studying their reactions with the three reagent ions H3O+, NO+ and O2+. With regards to humidity it would be advised to study these reactions under the conditions of the final sample, since as such the possibility and extent of H2O cluster formation will be accounted for.