Identifcation of the optical isomers using laser induced breakdown spectroscopy combined with machine learning

We explore the possibility of extending the LIBS application combined with machine learning for the identification of optical isomers. Three sets of optical isomers, viz., Tryptophan, Alanine, and Leucine, were chosen to demonstrate this feasibility. Each set comprises D, L, and DL isomers. Identifcation of specifc optical isomers is crucial in the feld of pharmaceuticals as they account for more than 50% of drugs in the current market. Desired benefcial efects can only be obtained from using a particular isomer drug, while the intake of others may result in adverse side efects. Herein, for the frst time, to the best of our knowledge, we report concurrent identifcation and separation of optical isomers using laser-induced breakdown spectroscopy (LIBS). The plasma diagnostic studies revealed that the plasma of DL isomer has a higher temperature compared to the remaining two types. The time-resolved studies also supported this observation, which demonstrated a higher decay time for the spectral lines of the DL isomer. The Principal Component Analysis (PCA) has revealed that optical isomers can be well separated. Further, Logistic Regression (LR) and Support Vector Machine (SVM) analysis quantitatively measured the classifcation accuracies. The results demonstrated that isomers can be recognized with ~ 93 – 100% accuracy. In order to avoid the infuence of contaminates, the analysis is repeated by removing the spectral lines of containments and achieved accuracies of more than 90%. These results indicate that LIBS can be utilized as a promising technique for separating optical isomers, which immediately impacts various felds such as pharmaceuticals, agrochemicals, and food additives.