Estimativa de lignina e extrativos em eucalyptus comparando equipamentos NIR portáteis

Abstract

Wood quality control is essential, as it enables more efficient management of this raw material throughout the production process. Near-infrared (NIR) spectroscopy is a promising technique for monitoring these properties; however, further studies are needed to evaluate the accuracy of its estimations. The objective of this study was to evaluate the performance of portable NIR spectrometers (MicroNIR, Viavi Solutions Inc., CA United States), considering differences in equipment configurations, methodology and operators, in the estimation of lignin and extractive content in Eucalyptus wood. The samples were obtained from Eucalyptus clones supplied by two companies. Principal component analysis (PCA) was applied to evaluate spectral dispersion, partial least squares regression (PLS-R) was used to build predictive models, which were subjected to four validation methods: (1) validation with external data not used in the model calibration; (2) model calibration using 70% of the data and validations with the remaining 30%; (3) calibration using equipment 1 and validation with equipment 2; and (4) calibration using equipment 2 and validation with equipment 1. Regression coefficients were also analyzed to identify the most relevant spectral bands for model construction. The models were selected based on the highest coefficient of determination (R2 ), lowest root mean square error (RMSE) and highest residual predictive deviations (RPD). For lignin content prediction, the 70%/30% validation using equipment 1 showed the best performance compared to the others methods (R2= 0,71, RMSEp= 0,44, and RPD=3,936), in the case of extractive content, the external validation showed high R 2 and RMSEp values, whereas the 70%/30% validation yielded both low RMSEp and low R 2 values, also in equipment 1. Analysis of regression coefficients allowed the identification of bands that differentiate the two components, especially in the 1.000 a 1.300 nm region, corresponding to lignin molecular bands. Therefore, NIR using a portable device demonstrates good potential for predicting the chemical composition of wood.