Predicting bleachability of Eucalyptus mechanical pulp by moisture content-dependent near-infrared spectroscopy

Abstract

As the main raw materials in the pulp and paper industry, Eucalyptus wood has different bleachability due to the variety of species. In this study, moisture content-dependent near-infrared (NIR) spectroscopy was developed to predict the bleachability of two different Eucalyptus mechanical pulp from Eucalyptus globulus (EG) and Eucalyptus urophylla (EU). Two-dimensional (2D) correlation analysis was applied to explain the distinctive spectral change of EG and EU mechanical pulps under moisture content perturbation, the susceptible extent and sequential order of spectral changes were displayed in the synchronous and asynchronous maps, respectively. In the synchronous map of EU pulp, the spectral intensity at 1685 nm related to lignin changed significantly with moisture content, and the spectral change at 1432 nm related to cellulose was stronger than that at 1373 nm related to holocellulose, while the opposite behaviors were observed in the synchronous map of EG pulp. Meanwhile, the two Eucalyptus species had different sequential orders of spectral change occurring at the lignin-related band (1164 nm) and carbohydrate-related bands (1432 nm and 1590 nm) in the asynchronous maps. Then the correlation between moisture content-dependent spectral features and bleachability of pulps was confirmed by principal component analysis (PCA). Finally, a radial basis function neural network (RBFNN) predictive model was built using moisture content-dependent spectral features to describe the brightness response of Eucalyptus mechanical pulp under different bleaching conditions and exhibited excellent prediction accuracy and generalization performance. Therefore, moisture content-dependent NIR spectroscopy may be a promising technique for predicting the bleachability of raw material in pulping and papermaking processes.