Predicting aggregation rates of polycyclic aromatics through machine learning

Jacob C. Saldinger,Paolo Elvati,Karam Alrawi,Angela Violi
DOI: https://doi.org/10.1016/j.fuel.2024.131031
IF: 7.4
2024-02-03
Fuel
Abstract:The transition of gas-phase polycyclic aromatic compounds (PACs) into mature soot particles stands as a critical stage in the process of soot formation. However, formulating an accurate estimation of the rates of these processes poses a challenge, stemming from the need to consider the energy barriers controlling the dimerization of PACs. In this study, we underscore the effectiveness of machine learning in predicting the free energy barrier of dimer dissociation, which is fundamental to the dimerization rate, relying solely on the basic molecular features of any PAC. To obtain the free energy surfaces of the dimerization process for a diverse set of PACs at various temperatures, we employed well-tempered Metadynamics simulations accompanied by a gauge correction. Following this, we trained a machine learning model to predict free energy barriers across the said temperatures. We found that this model significantly outperforms the accuracy of commonly used mass correlations. Our machine-learning model also automatically identifies specific molecular descriptors. These descriptors quantify the relation to the barrier and qualitatively link to interpretable properties vital to the dimerization process. The study has far-reaching implications for soot formation research. It introduces an accurate method for estimating PAC dimerization rates, offering potential enhancements in soot formation model precision. Moreover, it pinpoints key molecular characteristics that provide a deeper comprehension of the mechanisms governing PAC clustering. This understanding can fuel the development of novel strategies for controlling soot formation.
energy & fuels,engineering, chemical
What problem does this paper attempt to address?