Qualidade do azeite de oliva extravirgem por Espectroscopia Raman e análise multivariada de dados

• Ano de defesa: 2021
• Autor(a) principal: Barros, Iago Henrique Assis Souza
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade Federal do Espírito Santo; BR; Mestrado em Química; Centro de Ciências Exatas; UFES; Programa de Pós-Graduação em Química
• Palavras-chave em Português: Portabilidade; azeite de oliva extravirgem; óleo de soja; fraude; quimiometria; subject.br-rjbn; Química
• Link de acesso: http://repositorio.ufes.br/handle/10/15649

In recent years, the countless cases involving fraud in the food industry have made consumers suspicious of the quality of the products they are consuming. The main targets of fraud in the food industry are those products that have a high added value, such as extra virgin olive oil. A study was carried out to evaluate the quality of extra virgin olive oils and commercial compound oils using Raman spectroscopy and chemometric devices, in order to identify possible adulterations in products by soybean oil. For this, an exploratory analysis was initially carried out, with the objective of evaluating the similarity of commercial products with different types of vegetable oils. Analytical calibration curves were also prepared, obtaining the best figures of merit using partial least squares regression by intervals with normalization preprocessing (detection limit and quantification of 0.9731% m / m and 3.2436% m / m, prediction and calibration determination coefficient greater than 0.99 and calibration and prediction errors less than 2% (m/m)). Of the 12 samples analyzed, 7 had a higher soybean oil concentration than that determined by the manufacturer, thus configuring the practice of adulteration of the products under study. In addition to the commercial samples, a study was carried out on the quantification of soy oil in 59 samples of 5 different brands of extra-virgin olive oil seized by the civil police of Espírito Santo in 2019 with another portable Raman spectrometer. For this purpose, three different multivariate calibration models have been proposed: PLS, PLS with pre-treated data by SNV and PLS with pre-treated data by MSC. The model chosen was PLS SNV due to the lower prediction error for determining adulteration of the seized oils (RMSEP PLS SNV = 3.0527% w / w; RMSEP PLS MSC = 3.0640% w / w; RMSEP PLS = 3.3546 % m / m). All the specimens evaluated had a soybean oil concentration above 85% by mass, thus characterizing adulteration.