Quantification of Antioxidants in Polyethylene by Near Infrared (NIR) Analysis and Partial Least Squares (PLS) Regression

Abstract

Partial least squares (PLS) regression was used to generate a calibration model that can be used for the prediction of the content of two antioxidants, Irganox 1010 and Irgafos 168, in high density polyethylene (HDPE). The samples containing levels of either one or both antioxidants in a range of 0-4500 ppm were analyzed by near infrared (NIR) in the diffuse reflectance mode. The samples were compounded in an extruder and subsequently analyzed as pellets; therefore, sample preparation was not required. High performance liquid chromatography (HPLC) was used as a reference method and the extraction of the antioxidants was performed either by microwave-assisted extraction (MAE) or ultrasonication. Data pretreatment of the raw NIR-data was necessary in order to eliminate the physical differences of the samples, e.g., size and shape. Multiple scattering correction (MSC) and second derivative of the raw data were used for this purpose. Root-mean-square error of prediction (RMSEP) for Irganox 1010 and Irgafos 168 was 46 and 97 ppm, respectively, when derived raw data was used; similar results were obtained when calibration was performed on MSC data. The number of principal components was determined by cross-validation; in addition, the calibration model was validated with a test set.