Complementary Raman and FTIR spectroscopic insights into lignin removal from poplar using deep eutectic solvents

Accurate and rapid quantitative determination of lignin is crucial for evaluating biomass processing and value-added applications. This study systematically investigated the applicability of variable selection methods-SCARS, CARS, MC-UVE, iMWPLS for predicting lignin content. Results indicate that full-spectrum Raman modeling exhibits limited predictive accuracy (R_p²=0.6221, RMSEP=1.8779). The fused spectra exhibited stronger complementary information effects. Under full-spectrum modeling, the R_p² value of the fused spectrum reached 0.7585-22.0% higher than Raman and 8.8% higher than ATR-FTIR-while RMSEP was 1.2923, significantly lower than Raman (31.2% reduction) and marginally better than ATR-FTIR. Under fusion conditions, MC-UVE achieved optimal results (R_p²=0.8657, RMSEP=1.0561), representing a 14.1% improvement and 18.3% reduction compared to the full spectrum, while also outperforming the optimal models from single spectra. This study revealed the differences and complementarity in feature variable distributions between Raman and infrared spectroscopy, emphasizing the critical role of variable selection methods in enhancing modeling accuracy. It provides a novel technical pathway for rapid quantitative detection of lignin. The application of spectral techniques significantly reduces reliance on wet chemical analysis, enabling real-time, non-destructive monitoring of the lignin removal process. This methodological advancement lays the groundwork for subsequent life cycle assessments-where the adoption of spectral monitoring is expected to decrease analytical energy consumption and chemical usage.