Klason lignin estimation in Leucaena leucocephala by near infrared spectroscopy for selection of superior material for pulp and paper

Abstract

Near infrared spectroscopy partial least squares (PLS) regression calibrations were developed for estimation of Klason lignin content in Leucaena leucocephala (Lam.) de Wit (commonly called Subabul in India) belonging to the family Fabaceae and sub-family Mimosoideae with solid radial-longitudinal strip and milled samples for trees having diameter ranging from 3.5 to 34.1 cm and age between 4 and 20 years. Samples were conditioned to 10–12 % moisture content. Two separate models were developed for milled samples having particle sizes of 250–400 μm (screened-1) and 420–1,000 μm (screened-2). Regression model using PLS regression with full cross validation were test validated on a separate set of samples. The three multisite calibrations gave RPD (ratio of performance to deviation) between 2.24 and 2.97 with lowest for solid radial strip samples (2.24) and highest for screened-1 powder (2.97). Screened-1 milled sample gave highest rp2 of prediction (0.92) with four factors compared to four that were used for solid strips and eight for screened-2 samples. Model with solid strips gave lowest rp2 of prediction (0.86). Results indicate that properly mixed milled samples having particle size 250–400 μm (screened-1) help in construction of PLS models with higher RPD and less number of factors compared to screened-2 samples (2.48) and solid strips (2.24). Particle size played an important role in construction of the model. Lower RPD with solid strips could be due to variation of chemical constituents from pith to bark. The standard error of calibration (SEC), standard error of cross validation (SECV) and standard error of prediction (SEP) were less than 1 % for Screened-1 samples. For screened-2 samples SEC and SECV were less than 1 % while SEP was 1 %. In case of radial solid strip SEP was slightly above 1 % (1.16) while SEC and SECV were less than 1 %. For constructing models of high accuracy with radial solid strips separate models may be constructed for juvenile and mature wood.