Revealing the initial pyrolysis behavior of decalin in an experimental study coupled with neural network-assisted molecular dynamics

Decalin, as a typical polycyclic alkane, has broad application prospects as an essential component in aviation fuel surrogates. However, the initial pyrolysis behavior of decalin, especially the possible role of rearrangement reactions, which is necessary for subsequent pyrolysis and combustion, is still under investigation. In this work, experiments in a chemical micro reactor with a characteristic residence time of microsecond (mu s) are designed to explore the initial pyrolysis behavior of decalin, coupled with a photoionization molecular-beam mass spectrometer (PI-MBMS) for diagnosis. At the same time, the deep potential molecular dynamics (DPMD) method is used to assist in interpreting the experimental results. Comparing the initial decomposition pathways of decalin obtained from DPMD simulation with previous literature results, it is confirmed that despite the different pyrolysis conditions of the fuel in experiments and DPMD simulations, the specific initial pyrolysis reactions remain unchanged. Combining the DPMD results to diagnose the experimental pyrolysis products, there exist species that had not been reported in previous work, including intermediate components, such as 1,4-cyclohexadiene and 2-methyl-1-buten-3-yne, as well as cycloalkanes with low unsaturation, such as cyclohexane, methylcyclopentane, and methylcyclohexane. By extracting the pathways of the corresponding species generation from MD results, it is found that rearrangement reactions and intermolecular H-atom reactions in the system play a key role in generating these species. These reactions jointly lead to changes in the structures of intermediates, altering the afterward decomposition reactions and further affecting the distribution of the pyrolysis products. Our work reveals the effects of rearrangement reactions and intermolecular H-atom reactions in the pyrolysis of decalin for the first time at the experimental level. Also, the benefits of MD help in interpreting experiments are demonstrated.