1. Coupling of soft-modeling methods with multivariate pattern recognition technique for the identification of nitroaniline isomers
- Author
-
Masoumeh Hasani and Fereshteh Emami
- Subjects
Multivariate statistics ,Rank (linear algebra) ,business.industry ,Chemistry ,Applied Mathematics ,NQS ,Pattern recognition ,Linear discriminant analysis ,Least squares ,Analytical Chemistry ,Nitroaniline ,Pattern recognition (psychology) ,Principal component analysis ,Artificial intelligence ,business - Abstract
A new approach that takes advantage of both soft-modeling and pattern recognition methods is proposed to analyze kinetic data monitored spectrometrically to classify structurally similar nitroaniline isomers. The colorimetric condensation reaction of 1,2-naphthoquinone-4-sulfonate (NQS) with amines has been used to classify 2-, 3-, and 4-nitroaniline on the basis of their different kinetic properties. These nitroanilines react differentially with NQS at pH 7 to produce a colored product. Soft-modeling approaches such as multivariate curve resolution–alternating least squares (MCR–ALS) and simple-to-use interactive self-modeling mixture analysis (SIMPLISMA) followed by multiplicative signal correction have been used as preprocessing procedures to obtain patterns diagnostic for the analytes. Rank analysis was initially performed using determination of rank by augmentation to detect any species present during the course of the reaction in the system under investigation. The spectra and kinetic profiles of all absorbing species present in the raw measurements were obtained from SIMPLISMA and MCR–ALS resolution. The kinetic profiles were then used as input of the pattern recognition algorithm. The response patterns were systematically classified using linear discriminant analysis with high classification accuracy. Either all or one of the resolved profiles related to the components involved in the reaction can be used for identification of nitroaniline isomers. Copyright © 2013 John Wiley & Sons, Ltd.
- Published
- 2013
- Full Text
- View/download PDF