Discrimination and Quantification of Moroccan Gasoline Adulteration with Diesel Using Fourier Transform Infrared Spectroscopy and Chemometric Tools

In this work, transform-infrared spectroscopy (FTIR) was associated with chemometric tools, especially principal component analysis (PCA) and partial least squares regression (PLSR), to discriminate and quantify gasoline adulteration with diesel. The method is composed of a total of 100 mixtures wer...

Full description

Saved in:
Bibliographic Details
Published in:Journal of AOAC International 2019-05, Vol.102 (3), p.966-970
Main Authors: Barra, Issam, Mansouri, Mohammed Alaoui, Bousrabat, Mohammed, Cherrah, Yahia, Bouklouze, Abdelaziz, Kharbach, Mourad
Format: Article
Language:English
Subjects:
Backup software
Chemometrics
Composition
Diesel fuels
Fourier transform infrared spectroscopy
Gasoline
Gasoline - analysis
Infrared spectroscopy
Least-Squares Analysis
Measurement
Methods
Morocco
Principal Component Analysis
Spectroscopy
Spectroscopy, Fourier Transform Infrared - methods
Testing
Citations: Items that cite this one
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:In this work, transform-infrared spectroscopy (FTIR) was associated with chemometric tools, especially principal component analysis (PCA) and partial least squares regression (PLSR), to discriminate and quantify gasoline adulteration with diesel. The method is composed of a total of 100 mixtures were prepared, and then FTIR fingerprints were recorded for all samples. PCA was used to verify that mixtures can be distinguished from pure products and to check that there are no outliers. As a result of using just PC1 and PC2, more than 98% of the general variability was explained. The PLSR model based on infrared spectra has shown its capabilities to be suitable for predicting gasoline adulteration in the concentration range of 0 to 98% (w/w), with a high significant coefficient of determination ( ² = 99.25%) and an acceptable calibration and prediction errors (root mean squared error of calibration = 0.63 and root mean square of external validation and/or prediction = 0.69).
ISSN:1060-3271
1944-7922
DOI:10.5740/jaoacint.18-0179