Abstract

Spectral features in Raman spectra of organic molecules can be attributed to certain functional groups. A library of 1222 Raman spectra was used to train an artificial neural network (ANN) for predicting the presence of 13 functional groups. Sensitivity analysis was applied to the ANN models to determine a sensitivity factor or feature spectrum for each functional group. The feature spectra could then be used to predict the presence of specific groups based on Bayes' theorem. Once a model is constructed for each functional group, it can be applied directly to measured spectra of structurally unknown molecules and provide real-time predictions. Prediction accuracies of greater than 90% were obtained for aromatic, alkene, aldehyde, ketone, ester, nitro, and nitrile linkages. Accuracies for alcohols and amines were in the 80% range.

PDF Article

Cited By

You do not have subscription access to this journal. Cited by links are available to subscribers only. You may subscribe either as an OSA member, or as an authorized user of your institution.

Contact your librarian or system administrator
or
Login to access OSA Member Subscription