Infrared spectroscopy-assisted predicting of impurities in chemicals using machine learning: towards smart self-driving laboratories

Abstract

The changeable nature of chemicals requires constant monitoring. In a self-driving laboratory, manual revision of chemicals should be replaced by automated compliance check. Infrared spectroscopy is an SDL compatible tool, that can be integrated in a robotic arm to record a spectrum of each compound scheduled to use in a synthetic trasformation. In a case of any changes happened with a chemical, the IR-spectrum would include both signals of the main compound and an impurity, resulting in fail in decoding and further synthesis. Here, a series of complicated IR-spectra of mixtures of compounds was generated to train a neural network. The trained model was tested using IR-spectra of the real mixtures of compounds. F1-score of 0.94 for purity determination and 0.97 for functional group determination was achieved for generated spectra. F1-score 0.99 for functional group determination was achieved for the real spectra and 0.95 for purity determination task was observed. If fact, the model was able to determine, whether a compound was pure enough for further use or contaminated with the identified substance and should be purified.