Parreira T. F., Ferreira M. M. C., Sales H. J. S.,“Quantitative Determination of Epoxidized Soybean Oil Using Near Infrared Spectroscopy and Multivariate Calibration”. Villeneuve d'Ascq, France, 01-06/09/2002: XXVIth  European Congress on Molecular Spectroscopy (EUCMOS XXVI), Abstracts, 81 (2002). Oral OC12. Section: ORAL CONTRIBUTIONS C.

                                                     OC12

             QUANTITATIVE DETERMINATION OF EPOXIDIZED SOYBEAN OIL
              USING NEAR INFRARED SPECTROSCOPY AND MULTIVARIATE
                                                        CALIBRATION

                      Thais F. Parreira,^a M. M. C. Ferreira,^a Henrique J. S. Sales^a,b

^aInstituto de Química, Universidade Estadual de Campinas, 13083-970, Campinas S.P., Brazil
                            ^bHenkel S/A Indústrias Químicas, 12300-000, Jacareí, S.P., Brazil
                                                marcia@iqm.unicamp.br, rudolf@iqm.unicamp.br

The  epoxidized  soybean  oil  (ESO)   is  industrially  used  as   plasticizer  and  heat
stabilizer  toPVC  films.   During  the  thermal  decomposition,  hydrogen  chloride    is
eliminated  from  PVC,  conjugated  double  bondsare  formed,  and   the  processis
followed by changes in polymer's coloration. The epoxide ring from  ESO  reacts  with
the hydrogen chloride generated  by  polymer degradation  and restores  the chlorine
atoms back  into the polymer's chains.  This prevents  the  PVC from further
dehydrochlorination.

In order to follow  the soybean  oil  epoxidation  process,  it  is  necessary  to  quantify
three analytes related to the product's quality and process efficiency:

1-The Epoxide Index (E.I.) directly related to the stabilizer feature  of  the product, i.e.
the  higher  the  epoxide  index,  the  more  efficient  will  be  the  additive  as  thermal
stabilizer.

2-The Iodine Index (I.I.),an  additive  of  theamount  of  unsaturations present  in  the
epoxidized  soybean  oil,  which  is  unknown  a  priori.

3-The water percentage resulting  from  washing  of  the final product  in the industrial
process.  Its  concentration must be minimal, since  the water  causes  degradation  of
the  epoxide  group.

In this work¹,Near  Infrared  (NIR)spectra recorded  off  line at the chemical  industry
during  the epoxidation process,  combined  with multivariate regression method  PLS,
was used  for the quantification of  I.E.,  I.I.,  and  water  in  the epoxidized soybean oil.
The  water  levels  in  the  ESO  were  determined  by  the Karl  Fischer  titration.  The
content  of iodine was determined  by iodometry  and  the epoxide index was obtained
by the hydrogen bromide reaction with the epoxide ring.  In average, 50 samples were
used for modeling and 10 for external validation.  The PLS models were optimized  by
applying   a   few   different   variable   selection  procedures,   such   as   the  spectra
subtraction and the correlogram, besides the simultaneous inspection  of  the loadings
and regression vector.  The models were validated  by  leave-one-out  crossvalidation
and  using  an  external  data  set.  The  results,  expressed  in  terms  of  the  percent
relative  error  of  prediction,  were  in  the  range  1-2%.
It is shown in this work, that the use of NIRS combined with multivariate regression  is
a good alternative  to the widely established techniques,  especially  for the  industrial
purposes where time,  effort and money play  a crucial  role.
 
 

[1] T. F. Parreira, M. M. C. Ferreira, H. S. S. Sales, W. B. Almeida, submitted for publication.
 
 
 

                                    EUCMOS XXVI - VILLENEUVE D'ASCQ  2002                  81