E-liquids are liquids used in e-cigarettes. They are responsible for creating the vapor that is inhaled by the user. E-cigarettes were created with the intention of assisting smokers in quitting because they are believed to be less dangerous than conventional combustible cigarettes. With a range of flavors and nicotine levels to choose from, as well as the presentation as “the healthier option” and cost-effectiveness, it's no wonder that e-liquids have become increasingly popular in the last 20 years, particularly with younger people. Though, vaping is not without health concerns. The long-term effects are still not fully understood, and there have been reports of health problems associated with e-cigarette use, including lung damage. Therefore, it is necessary to monitor the compositions of the products on the market and conduct surveillance studies. In Belgium this is performed under the coordination of the Federal Public Service of public health.
E-liquids typically contain four main ingredients: propylene glycol, glycerin, flavorings and nicotine. The latter is of course not present in zero liquids. The primary issue is the content of nicotine. Previous reports showed that nicotine concentrations are not compliant to the label claim, with mean deviations between 5% and 20%. Also zero liquids containing nicotine were encountered. Another issue is the presence of additives, illegal according to the European Directive. Some of the most important are caffeine, taurine and vitamin E. Cannabidiol (CBD), which is in the gray zone of the legislation, was also taken into account of this study, due to its popularity.
In this study spectroscopic approaches were explored to check the label conformity for nicotine and the presence of caffeine, taurine, vitamin E and CBD in e-liquids. Therefore a set of 236 e-liquids was selected. Since samples with caffeine, taurine, CBD and vitamin E as additives were hard to find, some of the samples were split in two and one of both aliquots was spiked with the targeted additives in realistic concentrations. After sample set preparation, mid-infrared, near-infrared and Raman spectra were recorded for all samples. After obtaining the data matrices, chemometric analysis was performed. For each type of data, the sample matrix was split into a training and test set. Spectroscopic data were preprocessed before modelling. Next to baseline correction, classical preprocessing techniques like standard normal variate (SNV), derivatives and scaling were explored. In a first step, supervised modelling, using partial least squares-discriminant analysis (PLS-DA) and soft independent modelling by class analogy (SIMCA), was used to classify samples according to the presence of nicotine and/or the targeted additives. The results allowed to classify samples according to nicotine and present additives. So, their presence in new suspected samples could be detected. In a second step, PLS regression was applied to estimate the nicotine concentration and check the label conformity.
