A novel technology based on Raman spectroscopy for effective differentiation of alanine chiral isomers

Abstract

In this study, Raman time-domain spectroscopy (R-TDS) was used to record the spectra of β-alanine, D-alanine, L-alanine, and DL-alanine molecules in the Raman shift range of 250–2000 cm−1. Baseline correction and smoothing of the original spectra yielded considerably different scattering peaks for β-alanine than the other three alanine isomers. The positions and intensities of the peaks in the spectra of D-alanine, L-alanine and DL-alanine were very similar, enabling β-alanine to be easily distinguished from the aforementioned isomers based on the Raman spectra. The Raman spectral data of D-alanine, L-alanine and DL-alanine were subjected to variational mode decomposition (VMD), whereby each spectral line was decomposed into four intrinsic mode functions (IMFs). Finally, a grid search-support vector machine (GS-SVM) classification recognition model was established for Raman spectral analysis and classification analysis using the first four IMFs as datasets. VMD-GS-SVM classification had a 96.67 % discriminative accuracy for predicted samples of the three isomers of D-alanine, L-alanine and DL-alanine. The classification recognition accuracy gradually trended downward as the number of mode components increased. The results show that the fast identification model of alanine isomers based on R-TDS and VMD-GS-SVM is effective and feasible, providing an accurate and fast optical measurement solution for the development of novel drugs and quality monitoring of fine chemicals.