Reports: DNI654240-DNI6: Mesoscale Simulation and Machine Learning of Asphaltene Aggregation

Andrew L. Ferguson, PhD, University of Illinois at Urbana-Champaign

Doctoral research student, Jiang Wang (Physics, UIUC) was recruited to perform the proposed work, and commenced this research position at the beginning of the Fall semester on 08/16/14. Over the course of the last month, Jiang has been familiarizing himself with the asphaltene literature, and – under my supervision and guidance – is currently constructing atomically-detailed molecular models of a prototypical asphaltene molecule – C42H47NS, MW = 597 Da (E.S. Boek  et al. Energ. Fuel. 23 1209 (2009)) – and the canonical good and poor asphaltene solvents, toluene and n-heptane, and developing coarse-grained representations of these molecules under the MARTINI force field (S.J. Marrink et al. J. Phys. Chem. B 111 7812 (2007); S.J. Marrink & D.P. Tieleman Chem. Soc. Reviews 42 6801 (2013)). We will commence preliminary mesoscale molecular dynamics simulations of asphaltene assembly using the GROMACS simulations suite (S. Pronk et al. Bioinf. 29 845 (2013)) within the coming weeks. We will first validate our coarse-grained models by comparing the dimerization potential of mean force against that computed from fully atomistic simulations (M. Sedghi et al. J. Phys. Chem. B 117 5765 (2013)), and adjust the coarse-grained model as necessary. As a contingency in the event of unacceptably poor atomistic agreement, we will instead employ a less severely coarse-grained united-atom model (J.S. Hansen et al. J. Chem. Phys. 138 094508 (2013)). We will subsequently embark upon the first aim of this work to perform large mesoscale simulations of the aggregation of ~100 asphaltene molecules at 7 wt% concentration using the UIUC campus supercomputing facilities.