Machine learning‐assisted modeling of catalytic hydrogenolysis of lignin into monomeric products over Ni–Cu/ <scp>HZSM</scp> ‐5 bimetallic catalysts using methanol as H‐donating solvent

Abstract BACKGROUND Lignin, an abundant and renewable biopolymer, remains underutilized for the production of high‐value aromatic chemicals due to its complex and recalcitrant structure. This study presents an integrated experimental and machine‐learning‐assisted modeling investigation of alkali lignin hydrogenolysis over cost‐effective Ni–Cu bimetallic catalysts supported on HZSM‐5, using methanol as a hydrogen‐donating solvent. The catalysts were comprehensively characterized using Brunauer–Emmett–Teller, scanning electron microscopy, energy‐dispersive X‐ray spectroscopy, and X‐ray photoelectron spectroscopy to elucidate surface morphology, elemental dispersion, electronic structure, and the synergistic effects between Ni and Cu species. RESULTS Under optimized conditions (250 °C, 4 h), lignin conversion reached 91%, and the monomer yield achieved 41 wt.%, with the liquid products exhibiting a high heating value of 32 MJ kg −1 . Among all tested catalysts, the 30% Ni–20% Cu/HZSM‐5 system delivered the most effective performance, outperforming its monometallic counterparts in both conversion and yield. Product distribution analysis using gas chromatography–mass spectrometry, gas chromatography–flame ionization detection, gel permeation chromatography, Fourier transform infrared spectroscopy, and elemental analysis identified 19 distinct monomeric products. CONCLUSION A proposed mechanism elucidates the cooperative interaction between Ni and Cu in enhancing activity and selectivity. Machine learning models (random forest, XGBoost, and support vector regression) were developed to predict lignin conversion and monomer yield. Support vector regression achieved the highest predictive accuracy ( R 2 = 0.99 for yield and 0.97 for conversion). Feature importance analysis identified catalyst dosage as the dominant factor (0.39), followed by temperature and time. This integrated experimental–computational approach provides a predictive framework for process optimization and catalyst design in lignin valorization. © 2026 Society of Chemical Industry (SCI).