In this paper (purchase required) the authors make use of a recently developed ambient ionisation source for mass spectrometry. Self Aspiration Corona Discharge Ionisation (SACDI-MS) enabled the direct measurement of volatile compounds from coffee, and when the authors coupled this with an Air Curtain Sampling Device it meant that, by removing interfering volatiles from neighbouring batches, they could design an in-line sensor suitable for use in a production environment. The use of Deep Learning Algorithms with an Artificial Neural Network enabled them to compensate for other interfering peaks from environmental volatiles, often a problem in direct ionisation mass spectrometry. They constructed a chemometric classification model using a reference set of coffees from 6 different geographic origins and proved that they could differentiate between them in a high-throughput, rapid production environment. They conclude that this makes the approach ideal for in-line screening of coffee authenticity in situations when there is a consistent “expected” origin, used to train a classification model, that needs to be distinguished from substitution by “unexpected” origin coffee.
Photo by Andrew Neel on Unsplash
Comments