Fast biomass compositional analysis using Fourier Transform Near-infrared Technique

The objectives of this research were to determine the variation of chemical composition across botanical fractions of cornstover, and to use Fourier Transform Near-infrared (FT-NIR) techniques to qualitatively classify separated cornstover fractions, and develop calibration model for the quantitative analysis of chemical compositions of cornstover. Large variations of biomass chemical composition for wide calibration ranges were achieved by manually separating the cornstover samples into six botanical fractions, and their chemical compositions were determined by conventional wet chemical analyses, which proved that chemical composition varies significantly among different botanical fractions of cornstover. Husk, followed by rind and pith, has the highest sugar (glucan+xylan) content; node has the lowest sugar content. Based on FT-NIR spectra acquired on the biomass, classification by Soft Independent Modeling of Class Analogy (SIMCA) was employed to conduct qualitative classification of cornstover and Partial Least Square (PLS) regression was used for quantitative chemical composition analysis. SIMCA was demonstrated successfully in classifying botanical fractions of cornstover. The developed PLS models yielded root mean square error of prediction (RMSEP) of 1.058, 1.539, 0.987, and 1.435 for glucan, xylan, lignin, and ash, respectively. The FT-NIR techniques in combination with multivariate analysis are very useful to biomass feedstock suppliers, bioethanol manufacturers, and bio-power producers.