Neural network potential-based molecular investigation of pollutant formation of ammonia and ammonia-hydrogen combustion

ZH Xing and X Jiang, CHEMICAL ENGINEERING JOURNAL, 489, 151492 (2024).

DOI: 10.1016/j.cej.2024.151492

This study developed neural network potentials (NNPs) specifically designed for ammonia and ammoniahydrogen combustion systems for the first time. The NNPs were employed to perform reactive molecular dynamics (RMD) simulations, combining the precision of density functional theory (DFT) calculations with the efficiency of empirical force fields. NNP-based RMD simulations provide a more detailed and comprehensive understanding of chemical reaction mechanisms. The impacts of different equivalence ratios and hydrogen addition on ammonia combustion as well as NO X and N 2 O formation were analysed. The results indicate that both the addition of hydrogen and the reduction of equivalence ratios contribute to enhancing ammonia combustion. The primary reason is the enhancement of the oxidative environment within the system, leading to a significant increase in the frequency of reactions associated with oxidising radicals such as OH, O, and HO 2 . This is also the cause for the increased production of NO and NO 2 , because the formation of NO depends on the oxidation of NH and NH 2 , while NO 2 is formed through the oxidation of NO. An interesting finding is that the addition of hydrogen decelerates the generation of N 2 O, while reducing the equivalence ratio promotes N 2 O production. This phenomenon is attributed to the additional H radicals generated during hydrogen decomposition, which hinders the binding between NH and NO, resulting in a reduction in the formation of the precursor HN 2 O required for N 2 O formation.

Return to Publications page