Near infrared (NIR) spectroscopy provides rapid determination method for biomass characterization. The objective of this study is to use NIR to model and predict contents of monosaccharides in biomass hydrolysates. A uniform distribution of pretreatment conditions was used to generate representative samples that cover wide ranges of sugar concentration for model development. The formation of glucose, xylose, and arabinose was studied via response surface methodology and 3D models were provided to show the effects of pretreatment conditions. The NIR models developed with partial least squares are able to provide excellent and good prediction for glucose and xylose concentration in biomass hydrolysates, respectively. Data transformation did not increase model performance, but the reduced wavelength range improved model prediction for all the sugar contents. The NIR method significantly reduced the time and cost of sugar determination.
Keywords: Biomass hydrolysates; High-throughput analysis; Near-infrared; Partial least square.
Published by Elsevier Ltd.