Characterization of chrysotile, lizardite, and antigorite Raman spectra by multivariate analysis on serpentinite samples

The distinction between asbestiform (chrysotile) and non-asbestiform (lizardite and antigorite) serpentine polymorphs is critical due to the significant health risks associated with asbestos exposure. Chrysotile, the most common form of asbestos, is known for its fibrous morphology, which can lead to severe respiratory diseases when inhaled, including asbestosis, lung cancer, and mesothelioma. Differentiating asbestiform chrysotile from its non-asbestiform counterparts, lizardite and antigorite, is therefore essential for regulatory, industrial, and environmental purposes. Raman spectroscopy provides a powerful approach for differentiating these polymorphs based on their unique spectral features. However, the characterization of individual spectra remains heavily reliant on operator expertise, which can introduce variability and limit performance. To address this limitation, this study integrates Raman spectroscopy with multivariate analysis techniques, such as Principal Component Analysis (PCA) and Partial Least Squares Discriminant Analysis (PLS-DA), to develop a classification model that automates the identification process. The proposed approach aims to recognize diagnostic spectral patterns and to classify chrysotile, lizardite, and antigorite reducing the need for manual interpretation. The model is trained and validated on spectral data obtained from pure mineral samples and from serpentinite samples collected at the tunneling excavation site in Cravasco (Genoa, Italy), in asbestos containing Green Rocks with related potential asbestos exposure for numerous workers. The proposed approach enhances the reliability and efficiency of serpentine polymorphs differentiation, offering a robust tool for standardized and automated characterization of these minerals in real geological settings, allowing better risk assessment for exposed workers and timely adoption of more precautionary measures.