Double pulse laser-induced breakdown spectroscopy (DP LIBS) has attracted much attention for analyzing trace elements due to its higher sensitivity when compared to single pulse laser-induced breakdown spectroscopy (SP LIBS). However, the development of quantitative methods in LIBS for the analysis of complex samples, such as sediments, is a great challenge due to the matrix effects that are very accentuated in this technique. In this study, different spectral treatments and calibration strategies were investigated to obtain calibration models that allow determinations with satisfactory accuracy and precision of Cr, Cu, and Ni in river sediments from different hydrographic basins. The best model developed for Cr was using MMC without spectral normalization and for Cu and Ni it was using MMC with spectral normalization, and using inverse regression, an increase in the accuracy of the determinations of all analytes was obtained. These models showed limit of quantification (LOQ) of 7.87 mg kg-1, 1.62 mg kg-1, and 2.21 mg kg-1 and root mean square error of prediction (RMSEP) of 7.54 mg kg-1, 14.53 mg kg-1, and 8.29 mg kg-1 for Cr, Cu, and Ni, respectively. Therefore, the models have adequate sensitivity and precision for the quantification of the potentially toxic elements (PTEs) evaluated, since, according to Brazilian legislation, the lower concentration of threshold effect level (TEL) for Cr, Cu, and Ni is <37.3 mg kg-1, <35.7 mg kg-1, and <18 mg kg-1, respectively. The concentrations of Cr, Cu, and Ni determined by DP LIBS allowed to obtain a partial ecological risk assessment of the studied sediments. Also, the chemometric tool Kohonen self-organizing map (KSOM) were used for data interpretation.
Read full abstract