On the interpretation of reflectivity data from lipid bilayers in terms of molecular-dynamics models

Neutron and X-ray reflectivity of model membranes is increasingly used as a tool for the study of membrane structures and dynamics. As the systems under study become more complex, and as long, all-atom molecular-dynamics (MD) simulations of membranes become more available, there is increasing interest in the use of MD simulations in the analysis of reflectometry data from membranes. In order to perform this, it is necessary to produce a model of the complete interface, including not only the MD-derived structure of the membrane, but also the supporting substrate and any other interfacial layers that may be present. Here, it is shown that this is best performed by first producing a model of the occupied volume across the entire interface, and then converting this into a scattering length density (SLD) profile, rather than by splicing together the separate SLD profiles from the substrate layers and the membrane, since the latter approach can lead to discontinuities in the SLD profile and subsequent artefacts in the reflectivity calculation. It is also shown how the MD-derived membrane structure should be corrected to account for lower than optimal coverage and out-of-plane membrane fluctuations. Finally, the method of including the entire membrane structure in the reflectivity calculation is compared with an alternative approach in which the membrane components are approximated by functional forms, with only the component volumes being extracted from the simulation. It is shown that using only the fragment volumes is insufficient for a typical neutron data set of a single deuteration measured at several water contrasts, and that either weighting the model by including more structural information from the fit, or a larger data set involving a range of deuterations, are required to satisfactorily define the problem.