Metodologie chemiometriche e loro applicazioni allo studio della corrosione - Chemometric techniques and their application to the study of corrosion

The aim of this paper is to introduce the reader to the applications of chemometric techniques (i.e. Design of Experiment (DoE), classification and modelling techniques) in the field of corrosion science, which currently are not commonly spread. By means of key examples, it will be shown how they can be fruitfully applied and enhance the quality of a laboratory workflow. Chemometrics is defined by IUPAC as ''the science of relating measurements made on a chemical system or process to the state of the system via application of mathematical or statistical methods". It is currently applied in diverse field of chemistry by using the following methodologies: 1) signal processing, 2) Design of Experiment (DoE), 3) pattern recognition and 4) calibration. The present paper will mainly focus on the last three, since the available literature in signal processing is wide and well established [1,2,3]. After dealing with the mathematical aspects of DoE and the response surface modelling [7], we will show a key application with references to other valuable examples [8]. As highlighted in another paper [7], chemometrics plays a major role in the screening phase of a test as it leads to reduce the number of the needed experiments, when compared with the univariate approach. Further, it allows deriving a mathematical relation among the selected parameters that can be also plotted in the form of a contour surface for a better representation of the results. We will analogously proceed to describe the pattern recognition throughout an example of its application to the study of a bronze statue. Once pattern recognition is applied, corrosion products are grouped as a function of microclimates, thus giving useful information about the conservation state and the relationships among microclimatic factors and degradation process. Finally, we will discuss theory and applications of calibration methodologies, principally the partial least square technique [24] that leads to define a mathematical model based on 5 parameters out of 82 as mainly responsible for the state of conservation of ancient bronze findings.