Near infrared hyperspectral imaging as a tool for quantifying atmospheric carbonaceous aerosol

Abstract

The quantitative determination of atmospheric organic carbon (OC), elemental carbon (EC) and total carbon (TC) is of utmost relevance due to its influence on global warming, visibility degradation, and human health. Thermal-optical transmission (TOT) is the most common analytical method used to analyze them. Nevertheless, it is destructive, time-consuming, and only a small portion of the sample is analyzed. Here, near infrared spectroscopy coupled with a hyperspectral detector (HSI-NIR), together with multivariate data analysis, comes as a reliable alternative.This work aims at establishing a methodology for the quantitation of the carbonaceous content in atmospheric particulate matter (PM) fusing HSI-NIR and multivariate data analysis, and being validated with TOT analysis. Principal Component Analysis and Multivariate Curve Resolution analysis, used as qualitative tools, confirmed concentration trends in the PM samples, and contributions from the chemical composition of both the filter material and aerosol samples. The Partial Least Squares (PLS) regression models developed were optimized using interval Partial Least Squares (iPLS) selection and showed a high correlation between the HSI-NIR spectra and TOT measurements: R2Cal = 0.99, 0.94 and 0.99 for the OC, EC and TC contents, respectively. Finally, the spatial distribution of the carbonaceous content in the filter samples was assessed using a homogeneity percentage (%H). This showed values higher than 50% for all cases, which indicated that no distributional inhomogeneities were observed in the filters and corroborated the efficient operation of the sampling system. This study demonstrated the feasibility of using HSI-NIR for the quantitative assessment of carbonaceous particulate matter in air pollution monitoring.