Asbestos identification in bulk samples using FTIR and multivariate data analysis

This study aimed to investigate a laboratory procedure for identifying asbestos in asbestos-containing materials (ACMs) using Fourier Transform Infrared (FTIR) spectroscopy combined with Partial Least Squares-Discriminant Analysis (PLS-DA). A PLS-DA model was trained with the six regulated asbestos reference materials using FTIR with the diffuse reflectance infrared Fourier transform (DRIFT) technique to identify asbestos type(s) in unknown samples. The prediction of the PLS-DA model was evaluated using laboratory-generated asbestos-containing samples and suspected ACMs collected from the remediation and building management industries. Predicted asbestos type(s) from the PLS-DA model in the suspected ACMs were compared to the results of standard polarized light microscope (PLM) analysis. The PLS-DA model predicted 100 % correct classification in laboratory-generated single asbestos-containing samples (n = 30) and 80 % correct classification in mixed-asbestos samples (n = 45). The model predicted 96 % (24/25) correct classification in chrysotile containing bulk samples with extra sample treatment steps (ashing and acid washing). The PLS-DA model prediction with the samples containing multiple asbestos types was less accurate and will require further optimization through expanded datasets and model refinement. The proposed method is cost-effective, rapid, and may reduce reliance on the analyst's experience by directly measuring the distinctive chemical properties of the asbestos.