Uso de imagens digitais e ferramentas quimiométricas para identificação e determinação de Cr (VI) em solos contaminados

• Ano de defesa: 2012
• Autor(a) principal: Oliveira, Luciana Fontes de
• Orientador(a): Pereira Filho, Edenir Rodrigues
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade Federal de São Carlos
• Programa de Pós-Graduação: Programa de Pós-Graduação em Química - PPGQ
• País: BR
• Palavras-chave em Português: Química analítica; Quimiometria; Crômio; Imagens digitais; Solos
• Área do conhecimento CNPq: CIENCIAS EXATAS E DA TERRA::QUIMICA
• Link de acesso: https://repositorio.ufscar.br/handle/ufscar/6523

The determination of toxic metals in soils is an important factor to access their quality or identify possible contamination. The element Cr has double feature: Cr (III) is essential for human beings and Cr (VI) is toxic. In this context, the aim of this work is to propose a simple test for the determination of Cr (VI) in soil samples using digital images and chemometric tools. The samples used in this project were collected at Maritime Antarctica, around the Comandante Ferraz Antarctic Station (a Brazilian scientific station). Alkaline medium was used to extract Cr (VI) of the samples and the chromogenic reagent was added to color development. To obtain the digital images, the extracts were placed in Petri dishes and a multifunction printer was employed. The results were compared with those calculated by molecular absorption spectrometry. For the images, the MatLab program version R2009a was used to obtain the mean values of colors R, G, B, H, S, V, r, g, b and L, and the Pirouette 4.0 program to the chemometric tools application. The digital images were evaluated in order to be applied for the proposition of models of classification and supervised calibration method PLS (Partial Least Squares). Two types of preprocessing were used to calculate PLS models: autoscaled and mean centered. For the calibration set was used the blanks and standards in a total of 52 solutions and as validation set was used 44 solutions (samples replicas). The model using the mean centered preprocessing was better. Furthermore, when the real values were compared with the predicted the coefficient of correlation was 0,991. It is evident that the proposed method is a simple strategy with low cost for the detection of Cr (VI) in soil samples. Furthermore, the images have great potential to be applied for classification models. The Pb determination using digital images was also tested, but problems with the solutions pH did not permit the application of the methodology for real samples.