Detection of adulteration in acetonitrile using near infrared spectroscopy coupled with pattern recognition techniques

• NIR and chemometrics were used to determinate the adulteration of acetonitrile.
• NIR spectroscopy data were subjected to analysis by PCA and SVM model.
• SVM has more dominance in determining the impurities of acetonitrile than PCA.
• Acetonitrile with 1% impurities can be recognized by NIR coupled with chemometrics.

In this paper, near infrared spectroscopy (NIR) in cooperation with the pattern recognition techniques were used to determine the type of neat acetonitrile and the adulteration in acetonitrile. NIR spectra were collected between 400 nm and 2498 nm. The experimental data were first subjected to analysis of principal component analysis (PCA) to reveal significant differences and potential patterns between samples. Then support vector machine (SVM) were applied to develop classification models and the best parameter combination was selected by grid search. Under the best parameter combination, the classification accuracy rates of three types of neat acetonitrile reached 87.5%, and 100% for the adulteration with different concentration levels. The results showed that NIR spectroscopy combined with SVM could be utilized for determining the potential adulterants including water, ethanol, isopropyl alcohol, acrylonitrile, methanol, and by-products associated with the production of acetonitrile.

Figure optionsDownload as PowerPoint slide