Towards "greener" strategies in quality control: rapid volatilomics of cocoa based on HS-GC-IMS and machine learning.

Gas chromatography-ion mobility spectrometry (GC-IMS) has emerged as a powerful analytical platform in quality control of food, beverages, and flavor products. The technology allows for point-of-care application without the need for sample preparation, which makes it advantageous in resource-limited and equipment-hostile environments. One of these fields is the quality assessment of raw cocoa, which is fundamental for ensuring authenticity, product quality, as well as food safety and compliance. At the same time, analytical departments are facing an increasing urge to turn to a more sustainable use of resources, as well as substantial cost pressure. In the present study, a fast GC-IMS strategy was used to evaluate the provenance of cocoa in combination with machine learning. While most of the commercially available GC-IMS systems are based on nitrogen as a carrier gas, this approach was optimized and translated to a fast, hydrogen-based GC method. This was applied to a set of commercial cocoa liquor and data were evaluated by machine learning approaches, such as multivariate curve resolution-alternating least squares (MCR-ALS) and partial least squares-discriminant analysis (PLS-DA). By cutting down the analysis time by a factor of 2.5, this study demonstrates that in contrast to most conventional gas chromatography-mass spectrometry (GC-MS) systems, GC-IMS can be easily optimized towards higher throughput using the faster flow rates possible with hydrogen. Furthermore, this leads to enhanced signal quality and thus, a better basis for machine learning and finally, to an optimal tool for the classification of raw cocoa origins. Therefore, H