Using molecular simulation to aid in the analysis of neutron reflectometry measurements is commonplace. However, reflectometry is a tool to probe large-scale structures, and therefore the use of all-atom simulation may be irrelevant. This work presents the first direct comparison between the reflectometry profiles obtained from different all-atom and coarse-grained molecular dynamics simulations. These are compared with a traditional model layer structure analysis method to determine the minimum simulation resolution required to accurately reproduce experimental data. We find that systematic limits reduce the efficacy of the MARTINI potential model, while the Berger united-atom and Slipids all-atom potential models agree similarly well with the experimental data. The model layer structure gives the best agreement, however, the higher resolution simulation-dependent methods produce an agreement that is comparable. Finally, we use the atomistic simulation to advise on possible improvements that may be offered to the model layer structures, creating a more realistic monolayer model. Usage: Electronic Supplementary Information (ESI) including all analysis/plotting scripts and figure files, allowing for a fully reproducible, and automated, analysis workflow for the work presented is available at https://github.com/arm61/sim_vs_trad (DOI: 10.5281/zenodo.3254719) under a CC BY-SA 4.0 license. Reduced experimental datasets are available at DOI: 10.15125/BATH-00586, under a CC-BY 4.0 license.
Read full abstract