New chemical average model for molecular simulations of the asphalt system

This study refined the Brown-Ladner (B-L) method and combined it with elemental analysis to construct average molecular structure models of virgin asphalt, aiming to achieve more precise molecular simulations of asphalt materials. The chemical structure accuracy of the proposed average molecular models was validated using proton nuclear magnetic resonance (¹H NMR) and Fourier transform infrared spectroscopy (FTIR). Furthermore, the electronic structural properties of these models were calculated <i>via</i> density functional theory (DFT) to predict potential reaction sites. Additionally, weak interaction mechanisms within the average molecular models of virgin asphalt were revealed by combining the aRDG approach with molecular dynamics (MD) simulations, and their stability was analyzed through the thermal fluctuation index (TFI). The effectiveness of the improved average molecular models of virgin asphalt was verified from multiple dimensions including thermodynamics, structural properties, transport characteristics, and interfacial adsorption behavior based on MD simulation. Notably, the LYM2 model accurately reflects hydrogen distribution and aromatic-to-aliphatic hydrogen ratios in real samples, identifies reactive sites, and predicts material properties like density, solubility parameters, glass transition temperature, and viscosity effectively, especially excelling in viscosity predictions based on the Rouse model. Moreover, significant π-π stacking effects and steric hindrance effects among average molecules of virgin asphalt were found, which show high kinetic stability due to the negligible influence from thermal motion. The new model of asphalt proposed in this study provides more efficient and accurate molecular structures for <i>ab initio</i> molecular dynamics (AIMD) simulations and machine learning force field training of asphalt materials for future research.