Use of NIRS to predict composition and bioethanol yield from cell wall structural components of sweet sorghum biomass

Abstract

Sweet sorghum biomass is gaining importance as feedstock for second generation bioethanol production. Consequently, breeding programs are seeking to improve the quality of this feedstock in order to increase the productivity, with the generation of a great number of samples to be analyzed. Thus, this paper developed rapid and low cost methods based on partial least squares (PLS) and near infrared reflectance spectroscopy for determining cellulose, hemicellulose, lignin and theoretical ethanol yield (TEY) in sorghum biomass. The models were built with 957 samples, obtained from more than 100 hybrids and inbred strains, in the ranges of 21.4–49.1% w/w, 18.4–34.8% w/w, 1.8–11.5% w/w and 221–412Lt−1 for cellulose, hemicellulose, lignin and TEY, respectively. These models presented root mean square errors of prediction of 1.5%, 1.7%, 0.8% and 12Lt−1 (and ranges of relative errors of prediction between −5.3 and 6.5%, −9.8 and 12.2%, −28.8 and 37.6%, and −5.6 and 6.1%), respectively. The methods were submitted to a complete multivariate analytical validation in accordance with the Brazilian and international guidelines, and considered accurate, linear, sensitive and unbiased. Finally the stability of these methods was monitored for approximately six months by developing appropriate control charts.