Large-Scale Atomistic Modeling of Bio-Oils Informed by Multimodal Characterization Data and Molecular Dynamics

Abstract Understanding the molecular structure of bio-oil is essential for improving upgrading strategies, predicting physicochemical properties, and designing efficient production pathways. This work introduces an integrated experimental-analytical-modeling framework that translates multimodal characterization data into large-scale, structure-resolved molecular representations of bio-oil. The approach combines elemental analysis, quantitative 13C NMR, GC-MS, FT-ICR MS/Orbitrap MS petroleomics, ESI(−)-MS molecular weight distribution, and bulk property measurements, including density and viscosity, to constrain the model composition and phase heterogeneity. A database of nonvolatile structures was generated using reactive molecular dynamics simulations and subsequently expanded through chemically consistent aging reactions, enabling broad coverage of experimentally observed molecular formulas. Two modeling strategies were evaluated: (i) treating the entire nonvolatile fraction as a single data set and (ii) using a subfraction-resolved formulation that incorporates the experimentally measured contributions of dehydrated sugars and pyrolytic lignin (dimer, trimers, tetramers, humins). Both approaches reproduced independent experimental constraints, including elemental composition, functional-group and molecular weight distributions, with errors below 10%. The resulting molecular ensembles captured key features of bio-oil behavior, including multiphase organization, micelle aggregation, and the role of hydrogen bonding in structural evolution. Additional validation through predicted 1H NMR and HSQC-NMR spectra showed strong agreement with experimentally observed chemical-shift regions, supporting the fidelity of the reconstructed molecular ensembles. This framework provides a chemically consistent pathway for resolving bio-oil complexity and generating predictive atomistic representations for downstream simulation and process-design applications.