Please use this identifier to cite or link to this item:
Type: Artigo de periódico
Title: Identification Of Gasoline Adulteration Using Comprehensive Two-dimensional Gas Chromatography Combined To Multivariate Data Processing.
Author: Pedroso, Marcio Pozzobon
de Godoy, Luiz Antonio Fonseca
Ferreira, Ernesto Correa
Poppi, Ronei Jesus
Augusto, Fabio
Abstract: A method to detect potential adulteration of commercial gasoline (Type C gasoline, available in Brazil and containing 25% (v/v) ethanol) is presented here. Comprehensive two-dimensional gas chromatography with flame ionization detection (GCxGC-FID) data and multivariate calibration (multi-way partial least squares regression, N-PLS) were combined to obtain regression models correlating the concentration of gasoline on samples from chromatographic data. Blends of gasoline and white spirit, kerosene and paint thinner (adopted as model adulterants) were used for calibration; the regression models were evaluated using samples of Type C gasoline spiked with these solvents, as well as with ethanol. The method was also checked with real samples collected from gas stations and analyzed using the official method. The root mean square error of prediction (RMSEP) for gasoline concentrations on test samples calculated using the regression model ranged from 3.3% (v/v) to 8.2% (v/v), depending on the composition of the blends; in addition, the results for the real samples agree with the official method. These observations suggest that GCxGC-FID and N-PLS can be an alternative for routine monitoring of fuel adulteration, as well as to solve several other similar analytical problems where mixtures should be detected and quantified as single species in complex samples.
Subject: Chromatography, Gas
Citation: Journal Of Chromatography. A. v. 1201, n. 2, p. 176-82, 2008-Aug.
Rights: fechado
Identifier DOI: 10.1016/j.chroma.2008.05.092
Date Issue: 2008
Appears in Collections:Unicamp - Artigos e Outros Documentos

Files in This Item:
File SizeFormat 
pmed_18571187.pdf472.54 kBAdobe PDFView/Open

Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.