Revealing global reaction mechanisms of polypropylene pyrolysis by reactive molecular dynamic simulation and reaction class prediction

Polypropylene (PP) pyrolysis offers a promising solution for transforming waste plastics into high-value chemicals. This work presents a combination approach of large-scale reactive molecular dynamic simulation (ReaxFF MD) and automatic reaction class prediction with machine learning method for investigating PP pyrolysis reaction mechanism. The reasonableness of the global reaction understanding on PP pyrolysis is supported by both the consistent representative pyrolyzate detections between experiments and simulation, as well as the same yield ranking of C₃H₆>C₂H₄>CH₄>C₂H₆ at the initial decomposition dominant stage between experiments and simulations. The comprehension of the dynamic product profiles and their relevant reaction classes in polypropylene pyrolysis was revealed, which indicates that the major species distribution in PP pyrolysis detectable experimentally become explainable by the major reaction classes with the aid of automatic reaction classification approach of SRG-Reax, The dominant homolysis and β-scission reaction classes accounting for 30 %∼50 % in the initial PP pyrolysis corresponds to the rapid increasing yields of C₂H₄ and C₃H₆. The increasing reaction classes of ring opening & recombination and chain cyclization as well as of intra-molecular H-shift, H detachment and inter-molecular H-abstraction should be responsible for the increasing ring formation of 5-membered, 6-membered, 7-membered rings, particularly for formation of aryl rings and naphthalene with temperature. The formation pathways of benzene, naphthalene and anthracene were unraveled. The mechanism understanding of PP pyrolysis demonstrate the potential of the combined simulation strategy of large-scale simulation and reaction class prediction in deepening understanding polymer degradation mechanism.