Determinação simultânea de metais em cachaças utilizando a voltametria de redissolução com eletrodos de nanotubo de carbono e calibração multivariada

• Autor(a) principal: Ferreira, Romário Junior
• Data de Publicação: 2015
• Tipo de documento: Dissertação
• Idioma: por
• Título da fonte: Repositório Institucional da Universidade Federal do Espírito Santo (riUfes)
• Texto Completo: http://repositorio.ufes.br/handle/10/1608
• topic: Voltammetry; Cachaça; Eletroquímica; Química analítica; Voltametria; Espectroscopia; Química; 54

Cachaças are drinks produced from sugarcane with an alcohol content of approximately 40% in volume. Due to its large consumption in the country and their complexity in chemical constitution, the determination of the metals: copper, zinc, cadmium and lead, it was promoted in this study, which in high amounts, are harmful to health and the human metabolism. It was determined the concentrations of these metals by voltammetric anodic dissolution with square wave, analyzed in situ by voltammetry. With the obtained currents by voltammetry, along with the concentrations found by a reference technique (Inductively Coupled Plasma - Optical Emission Spectrometry - ICP OES) , along with the currents obtained by voltammetry, two chemometric methodologies were developed to predict these metals in samples of Cachaça. Simultaneous determinations of these metals were constructed by means of Artificial Neural Networks (ANN) and the Partial Least Squares (PLS), using the currents obtained by voltammetry using mercury and carbon commercial electrodes, and a carbon nanotube electrode built in the laboratory. Before the development of these techniques, particular statistical pretreatments were developed in voltammetry currents data in order to obtain a better response of forecast of the metals. Both for the voltammetry and ICP OES, the determination of lead was ineffective, and can justify that the detection limit of the equipment was greater than the concentration of this metal in the samples. After the development of quimiometricos methods, it was calculated root mean squared error of prediction (RMSEP) for all metals, both in PLS as in neural networks, in addition to a construction of correlation charts of values obtained by ICP OES for predicted values by chemometrics. However, the F-test with the values found by chemometrics techniques determines that there is no variation of expected concentrations in statistical analysis. The proposed method showed itself simple and objective, making it a good alternative method of analysis.