Automated chemical fingerprinting of Mexican spirits derived from Agave (tequila and mezcal) using direct-injection electrospray ionisation (DIESI) and low-temperature plasma (LTP) mass spectrometry. Issue 34 (8th June 2017)

• Record Type: Journal Article
• Authors: Martínez-Jarquín, Sandra; Moreno-Pedraza, Abigail; Cázarez-García, Daniel; Winkler, Robert
• Is Part Of: Analytical methods. Volume 9:Issue 34(2017)
• Page Start: 5023
• Page End: 5028
• Publication Date: 2017-06-08
• Subjects: Chemistry, Analytic -- Periodicals; Analytical biochemistry -- Periodicals; Chemical laboratories -- Standards -- Periodicals; 543.1905
• DOI: 10.1039/c7ay00793k
• Languages: English
• ISSNs: 1759-9660

Abstract

Abstract : Tequila and mezcal are Mexican spirits prepared from Agave plants. Abstract : Tequila and mezcal are Mexican spirits prepared from Agave plants. The regulations for the designation 'tequila' are very strict with respect to used Agave species, the geographical origin, and the manufacturing process. Therefore, admixtures with cheaper and less regulated mezcal spirits are frequent. These counterfeit tequila products represent a serious problem for consumer health and the liquor market economy. With this motivation, we developed two high-throughput workflows for the classification of tequila and mezcal spirits. Both strategies are based on chemical fingerprinting by mass spectrometry (MS). The first method uses direct-injection electrospray ionisation (DIESI) with automated sampling, the second method employs low-temperature plasma (LTP) ionisation. For the ambient sampling with LTP we used a 3D-printed probe, as well as a home-build robot based on RepRap (3D printer) technology. Both methods are fully automated. With DIESI-MS, a sample analysis takes 6 min, with LTP-MS only 20 seconds. The data of both methods permit the discrimination between tequila and mezcal products, and thus are suitable for the quick screening of spirits for authenticity. We also found that pre-processing the raw spectra data with Random Forest models greatly supports the separation of informative signals from noise. This data preparation strategy is therefore especially useful for analyzing noisy data from ambient ionisation analyses.