Combining hierarchical clustering analysis with a simplex lattice mixture design in rapidly identifying the pyrolytic interactions and predicting the product yields during the co-pyrolysis of cellulose, xylan, and milled wood lignin

The pyrolysis behavior of lignocellulosic biomass is complex owing to the different combinations of the three main components, and the diversity of the pyrolyzate distribution renders the identification of the pyrolytic synergy of lignocellulosic components challenging. Herein, the co-pyrolysis of cellulose, xylan (a representative example of hemicellulose), and milled wood lignin (MWL) was investigated using hierarchical clustering analysis (HCA) combined with a simplex lattice mixture design. HCA revealed that MWL enhanced cellulose-derived levoglucosan (LG) production, with a 2.1-fold higher theoretical yield, owing to the inhibition of LG repolymerization. The cellulose-xylan-MWL interactions resulted in yields of 1,4:3,6-dianhydro-α-d-glucopyranose, glycolaldehyde, and furfural that were > 1.2-fold higher than those theoretically calculated. The ternary interactions also increased the yields of phenol, p-cresol, 3,5-dimethylphenol, 3-methoxy-1,2-benzenediol, and methylphenol, which is attributed to H-donation by cellulose and xylan pyrolyzates. The yield prediction models of the pyrolyzates were established based on the HCA-identified interactions. The cubic model fitted the yields of gas, LG, glycolaldehyde, furfural, and char that were influenced by ternary interactions. The quadratic model predicted the yields of CO, CO2, CH4, and C2H6, which were primarily influenced by binary interactions. This study elucidates the complex pyrolytic interactions between lignocellulosic components and predicts the pyrolyzate distribution based on the biomass composition. This is beneficial for the targeted recovery of desirable chemicals from various biomass resources.