Ensuring analytical quality in regulated markets through multivariate, multiway and doe strategies

Resumen

When facing an analytical determination problem, both physicochemical and instrumental features should be considered so as to develop a valid analytical method. All those variables are not actually independent from each other, so a multivariate framework where correlation and covariance play a key role strengthens the way to that success, as every study in this doctoral thesis consistently shows. In this sense, multivariate chemometric tools such as Partial Least Squares regression, detection capability estimation based on multivariate curves, D-optimal designs, both full and fractional factorial designs and desirability functions, together with multiway techniques like PARAFAC and PARAFAC2 decomposition have proven their suitability in enhancing accuracy, sensitivity, specificity and robustness of both spectroscopy- and chromatography-mass-spectrometry-based multiresidue procedures on different chemical families (tranquillizers, bisphenols and plasticizers), foodstuffs (Queso Zamorano, meat) and migrating food materials (tableware, glasses and infant dummies). In particular, the specificity required for every analytical method and specifically for those dealing with regulated substances has been fully ensured by the use of PARAFAC decompositions on three-way tensors; that specificity, which is founded on its so-called second-order advantage, can fail if traditional methodologies are followed whenever a poor resolution is achieved or whether interferents are present in the sample. The importance of a wise selection of the response variable for an analytical study has also been revealed, which should be kept in mind to succeed in the statement of any analytical Design Space, as the recent approaches to extend the Quality by Design paradigm into the analytical laboratory claim.