Innovative strategies of multivariate analysis and vibrational spectroscopy for an advanced characterization of tyre materials

This doctoral thesis presents research carried out in collaboration between academia, research institutions, and industry, with the ambitious goal of improving technology transfer from scientific research to industrial practice. Specifically, the work was conducted in partnership with the University of Genoa, the National Research Council – Giulio Natta Institute of Chemical Sciences and Technologies of Genoa, and Pirelli Tyre S.p.A. The study is framed within the tyre industry, a sector in which manufacturing processes remain highly complex from an engineering standpoint. Tyres themselves are chemically and physically intricate materials composed of multiple components, each fulfilling a specific function in the final product. These components are formulated from numerous raw materials, precisely combined to meet stringent performance requirements. In this context, extensive chemical and physical analyzes are routinely performed on raw materials as well as on intermediate and final products to investigate their chemical structure and mechanical properties. However, such analyzes rely on sophisticated instrumentation, are costly, require highly trained personnel, and are typically carried out off-line on limited sample sets. As a result, they do not fully align with the industry’s objectives for sustainable and efficient production. Non-destructive spectroscopic techniques offer a promising alternative, as they are inherently more suitable for automated and in-line monitoring within industrial environments. Nonetheless, these techniques produce complex datasets that require advanced statistical and chemometrics methods for proper interpretation and implementation. Integrating spectroscopy with multivariate analysis is therefore essential to maximize information extraction and enhance practical usability. This doctoral project specifically explores the potential of two spectroscopic techniques — near-infrared (NIR) spectroscopy and time-domain nuclear magnetic resonance (TD-NMR) spectroscopy — combined with chemometrics approaches. In more detail, NIR spectroscopy was investigated in conjunction with hierarchical classification strategies to develop a model capable of simultaneously performing raw material identification (RMID) and compliance verification for the most commonly used raw materials in the tyre industry. The hierarchical framework consisted of a nested partial least squares–discriminant analysis (PLS-DA) model for RMID, implemented through the automated hierarchical model builder (AHIMBU) method, followed by soft independent modeling of class analogy (SIMCA) models for compliance verification. TD-NMR spectroscopy was studied using regression approaches to develop models capable of determining key cross-linking parameters in tyre compounds. In particular, PLS regression models were developed, optimized, and validated. Moreover, detailed investigation of both raw and pre-processed TD-NMR signals was conducted to understand how variations in cross-linking properties are reflected in the corresponding profiles. Furthermore, given the increasing adoption of TD-NMR within the tyre and rubber sectors, a multivariate design of experiments (MDoE) approach was combined with a feature-extraction strategy to evaluate how TD-NMR signals change as the main factors influencing the measurement vary. This study provided deeper insight into the technique and enabled more informed selection of experimental conditions based on the analytical objectives, thereby improving its applicability in industrial settings. Overall, this research demonstrates that, through the application of appropriate chemometrics methods of multivariate analysis, spectroscopic techniques can be more effectively exploited within the complex environment of tyre manufacturing, ultimately contributing to the improvement of the production chain.