Prediction of yerba mate caffeine content using near infrared spectroscopy

There is a commercial and beneficial interest of producing yerba mate leaves into different grades of caffeine. This work uses a handheld and bench near-infrared (NIR) spectroscopy to compare and predict, using partial least squares (PLS) regression, the amount of caffeine in yerba mate leaves. Standards of pure caffeine were compared, using high-performance liquid chromatography (HPLC), with extracts of yerba mate. The bench spectroscopy gave a strong confidence model of caffeine prediction, whereas the handheld related to a fair model. For first detection and initial separation of yerba mate in the field, the modeling proposed can be used to predict caffeine intensity.