Identification of Residual Layers of Intumescent Fire-Retardant Coating Through Laser-Induced Breakdown Spectroscopy and Partial Least Squares Discriminant Analysis

This study explores the utilization of laser-induced breakdown spectroscopy (LIBS) combined with chemometric techniques in fire investigation. We employed two distinct heating devices to generate fire residues from intumescent fire-resistant coatings (IFRC). The residues' top and bottom layers were analyzed using LIBS. We processed the spectral data to minimize variables and applied partial least squares discriminant analysis (PLS-DA) for categorizing the residual layers. Furthermore, we used principal component analysis (PCA) for visualizing the class separability and identified key wavelengths and their associated elements through the regression coefficients of PLS models. The training and test phase classification accuracies range from 0.944 to 0.992. Notable wavelengths with significant intensity disparities include 251.63 nm (Si I), 251.91 nm (Al I), 387.03 nm (Ca I), and 383.85 nm (Mg I). These findings affirm that integrating LIBS with chemometric methods is an effective strategy for characterizing IFRC in fire investigations.