Vibrational excitation and dissociation processes in high-temperature N2-O2 state-to-state collisions based on neural network and dynamic simulation

The scattering cross-sections and reaction rate coefficients are crucial parameters for elucidating the energy transfer mechanism of state-to-state collisions between molecular gases and also serve as a fundamental basis for modeling the non-equilibrium flow field. However, the database of kinetic processes related to nitrogen shock flows is still being developed. In this work, a detailed kinetic study of the N2 + O2 collision is carried out by combining the quasi-classical trajectory method (QCT) and neural network model (NN). Firstly, QCT is used to calculate 90 N2(v) + O2(w) processes with various initial vibrational states (v,w), and the contributions of all vibrational excitation and dissociation reaction channels are discussed. The following conclusions are drawn: 1) The contributions of the vibration-vibration (VV) energy exchange channel of O2 and N2 are similar, while the vibration-translational (VT) transition mainly occurs on O2; 2) The total dissociation cross-section primarily results from the O2 single-dissociation channel, followed by the exchange-dissociation channel, with relatively minor contributions from the N2 single- and double-dissociation channels. Then, based on the QCT dataset, a high-performance NN model (R-value of 0.99) is trained to predict the total dissociation cross-section caused by N2(v) + O2(w) collisions. Compared with the method that only uses QCT, the method that jointly uses OCT and NN model can achieve an approximately 91.94% reduction in computational cost. Finally, to facilitate use in kinetic modeling, Arrhenius-type fits for the VV/VT rate coefficients are provided over the temperature range of 5000–30000 K, and an exponential form related to the translational energy Et is used to fit the total dissociation cross-section.