Parreira T. F., Ferreira, M. M. C., Sales H. J. S., de Almeida W. B., “Quantitative Determination of Epoxidized Soybean Oil Using Near Infrared Spectroscopy and Multivariate Calibration”. Appl. Spectrosc., 56(12), 1607-1614 (Dec 2002).
[Article.]
 

Abstract.
The quality of epoxidized soybean oil (ESO), industrially used as a plasticizer and heat stabilizer for PVC films, is given by the degree of
epoxidation (EI), the number of double bonds expressed as the iodine index (II), and the water percentage in the final product. In this work,
near-infrared (NIR) spectra recorded off line at the chemical industry during the epoxidation process, combined with the multivariate regression
method partial least squares (PLS), was used for the quantification of EI, II, and water percentage in the ESO. The water levels were determined
by the Karl Fischer titration. The content of iodine was determined by iodometry. The epoxide index was obtained by the hydrogen bromide
reaction with the epoxide ring. On average, 50 samples were used to model each analyte and 10 were used for external validation. The results,
expressed in terms of the statistical parameters RPD (ratio of standard deviation of experimental concentrations to the standard deviation of the
residuals) and RER (ratio error range), were excellent for the El (26.02 and 80.70, respectively). For the water % and II, the obtained results were
fairly good. Good correlations between predicted and real concentrations were attained (0.984, 0.966 and 0.974 for water %, II, and EI,
respectively). It is shown in this work that the use of NIRS combined with chemometric methods are of great importance, especially for industrial
purposes.
 

Keywords.
Soybean oil; Near-infrared spectroscopy; NIRS; Poly(vinyl chloride), PVC; Chemometrics; Partial Least Squares, PLS

Keywords Plus.
Regression; Selection; Tutorial; Spectra.