Abstract
The aim of this study was to set up a chemometric procedure using near infrared spectra acquired with a low-cost handheld spectrometer (SCiO), to quantify the main chemical components of maritime pine (Pinus pinaster) resin, in view of using the SCiO as a quality control tool for the tapping industry. This study was carried out on samples of resin harvested during the summer of 2018, in Biscarosse, France. Spectral data were collected using both an SCiO, and a benchtop spectrometer (MultiPurpose Analyzer I) for baseline reference . The rates of turpentine and rosin were quantified by gas chromatography (turpentine composition), liquid chromatography (rosin composition), and a ventilated oven . The chemometric procedure involved spectra preprocessing and relevant subset selection with the DUPLEX algorithm. Lastly, Partial Least Squares (PLS) regression was used to calibrate the models. The quantitative predictive ability of the resulting PLS regression models was evaluated via Ratio of standard error of Performance to standard Deviation (RPD) statistics. The results show that spectra preprocessing enhanced the quantitative predictive ability. For MPA I, RPD > 3.5, which expresses some very good to excellent quantitative predictions of the models. For SCiO, RPD > 2.5, which expresses a good quantitative predictive ability for quality control purposes. Thus, RPD statistics confirm that an SCiO could be used as a quality control tool.
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have