| Article ID | Journal | Published Year | Pages | File Type |
|---|---|---|---|---|
| 1235070 | Spectrochimica Acta Part A: Molecular and Biomolecular Spectroscopy | 2013 | 7 Pages |
This study revealed that it was possible to determine the enantiomeric composition of with multivariate regression models of spectral data obtained by ordinary UV–vis spectrophotometry of enantiomeric guest–host complexes. The total 60 samples involving three concentration levels of metalaxyl as low, medium and high were prepared for spectral collecting. Four methods of modeling were subsequently proposed and compared including two common ways and two compensating ways for variations in total analyte concentration. Firstly, without normalization robust modeling was failed to achieve while employing the medium concentration levels as calibration and the other two levels as a validation. The same case occurred when full-cross validation was conducted. Besides, two enhanced methods were developed to account for the systematic variation. One of which normalized the spectra with respect to the total concentration of enantiomeric, along with spectral data, as a variable in the statistical analysis. The other one ignored variations in total concentration, relying on the specific band normalization to sort out any variations due to total concentration differences. The results clearly demonstrated that the spectra according to concentration provided the acceptable predictive ability in determining enantiomeric composition.
Graphical abstractFigure optionsDownload full-size imageDownload as PowerPoint slideHighlights► We use sucrose as a chiral selector to determine the D-met by UV–vis spectral data with PLS models. ► We develop the model by concentration normalization to eliminate the variations the differences. ► We introduce a convenient normalization method–RBN to standardize the spectral of all samples. ► We use the model developed by the medium one to predict the high and low levels and the results are well.
