Multispectral Fusion Approach for Efficient Quantitative Detection of Low Lignin Content in Lignocellulose

In this study, poplar wood was pretreated with a deep eutectic solvent (DES) with the goal of efficiently removing lignin and retaining cellulose. To evaluate the lignin removal efficiency, the treated samples were analyzed by attenuated total reflectance Fourier transform infrared (ATR-FTIR) and near-infrared (NIR) spectroscopic techniques, and a rapid determination of the lignin residual content was achieved by building a spectral prediction model. The models were constructed and evaluated using raw spectra and various preprocessing tools, including first-order derivative, second-order derivative, multiple scattering correction, standard normal variate transformation, and their combination strategies, respectively. The partial least-squares regression method was used for spectral data modeling, and the Monte Carlo uninformative variable elimination (MC-UVE) algorithm was introduced for feature selection to optimize the modeling variables. The results show that both mid-infrared (MIR) and NIR exhibit good modeling performance under different preprocessing conditions. Among them, the fused spectral model with second-order derivatives combined with the MC-UVE algorithm performs optimally, achieving the highest coefficient of determination (Rc2 = 0.9687, Rp2 = 0.9601) in the joint MIR-NIR modeling while having the lowest modeling and prediction errors (RMSEC = 0.8633, RMSEP = 0.9774). In contrast, full-spectrum models without variable screening generally have redundant information and a slightly lower prediction accuracy. Overall, MC-UVE was effective in improving model stability and reducing noise interference, while the fusion MIR(2d)+NIR(smooth) preprocessing strategy further enhanced the extraction of lignin information. The spectral modeling approach based on the fusion of mid- and near-infrared, supplemented with feature screening and appropriate preprocessing, significantly improves the prediction accuracy. This strategy also provides a reliable and economically valuable way to monitor lignin residues.