Desenvolvimento de procedimentos de preparo de amostras de difícil mineralização para análise direta por laser-induced breakdown spectroscopy (LIBS)

Abstract

This doctoral thesis proposes sample preparation combined with the use of laser-induced breakdown spectroscopy (LIBS) for analysis of difficult-to-mineralize samples focusing on the analytical development of LIBS as well as for the development of sample preparation for the technique. The LIBS technique consists of focusing a laser pulse with high fluence (> GW s-1) under the surface of a sample. Following the incidence of this laser pulse, atoms are vaporized, atomized, ionized and brought into an excited energy state. When atoms and ions return to their lowest energy levels, they release photons of specific wavelengths that are translated into the electromagnetic spectrum. The technique has inherent challenges in the analysis of solid samples, which are mostly its applications, such as microheterogeneity and high limits of detection (LOD). Although the technique has great ability in solid sample analysis, it can also be used for liquid sample analysis, however, some specific problems arise, such as sample splashing with laser pulse impact and low reproducibility, and sample preparation can help minimize them. As goal, the present thesis focuses on the development of sample preparation of nickel ores, solder masks and sunscreens for analysis of elements of interest in each of the presented samples. The strategy for the analysis of nickel ore and sunscreen samples consisted of converting the matrix of liquid to solid in order to minimize the problems inherent to liquid analysis by LIBS. For this, polyvinyl alcohol (PVA), a water-soluble polymer, was used. For solder masks, the samples were completely dried on glass support for further analysis. Chemometric tools were used during the work to aid in data processing (normalization), exploratory analysis (PCA (principal component analysis)) and univariate and multivariate calibration (PCR (principal component regression), PLS (partial least squares), MLR (multiple linear regression)).