Emprego de técnicas espectroanalíticas para a análise de amostras de solos e de coprólitos : análise de textura e exploratória
Resumen
In this
study, samples of coprolites (fossilized feces) were analyzed in LIBS (Laser-Induced
Breakdown Spectroscopy) to evaluate the chemical profile in these samples using
hyperspectral images, which enable us to assess the different concentration of several
elements in the same image. One of these samples was submitted to analysis by ICP
OES (Inductively Coupled Plasma Optical Emission Spectroscopy) separately (rock and
coprolites) to assess the actual concentration of each element in both of the arrays.
Through the results of both techniques, it was possible to infer the diet of animals that
generated the coprolites. The samples correspond to the Oligo-Miocene, Cretaceous, and
Permian periods (28, 85 and 260 million years, respectively). Coprolites samples showed
positive correlation between the Ca and P elements, indicating the presence of Ca3(PO4)2,
compound generally present in carnivorous animal waste. In rock samples, the
predominant elements were Al, K, Na and Si. Through a PCA analysis (Principal
Component Analysis), it was found that the elements that characterized the samples
changed as the age. Through scores graph analysis, it is possible to identify groups of
samples in chronological order. The soil samples were analyzed by NIRS (Near-Infrared
Spectroscopy) in an attempt to create classification models for soil texture, which is
characterized by sand fractions, clay and silt soil. These are differentiated by particle size.
Various strategies have been adopted in an attempt to improve the created models and
several chemometric tools were utilized, such as SIMCA, KNN and PLS-DA for
classification, giving satisfactory results as accuracy 83-100% for the calibration set. PLS
to quantify the levels was also tested, but without success. The samples were also
subjected to analysis by LIBS and soil encapsulation strategy with a polymer was tested,
also without success, since the LIBS generates the spectral profile elements and does not
distinguish particle size.