Imaging the invisible: resolving polymer brush structure through a freeform bayesian analysis of neutron reflectometry data
Loading...
Date
2020-11-11
Journal Title
Journal ISSN
Volume Title
Publisher
Australian Institute of Nuclear Science and Engineering (AINSE)
Abstract
Surfaces covered with densely tethered polymer chains possess desirable properties and are ubiquitous in
natural and human-made systems. These properties stem from the diffuse structure of these polymer brush
interfaces; consequently, resolving their structure is key to better understanding and designing polymer brush
systems. We have been using the PLATYPUS neutron reflectometer at the ACNS to achieve this structural
resolution over the past six years, contributing to our understanding of brush structure, as well as fundamental
polymer physics.
However, the analysis of collected reflectometry data is not without significant challenges; Inflexible models
preclude viable structures and the uncertainty around accepted profiles (known as spread) is challenging to
quantify. Furthermore, there is no guarantee of profile uniqueness in reflectivity analysis - multiple structures
may match the data equally well. Quantifying profile uniqueness and determining the structures that agree
with collected data (known as multimodality) has not been previously attempted on brush systems. Historically, data analyses have used least-squares approaches, which do not satisfactorily determine profile spread
and bypass the possibility of c.
Here we will briefly document our journey in modelling neutron reflectometry data collected from polymer
brush systems, culminating in the presentation of our developed methodology. In this methodology, we model
our brush with a freeform profile that minimises assumptions regarding polymer conformation while only
producing physically reasonable structures. This model is built within refnx‘s Bayesian statistical framework,
which enables the characterisation of structural uncertainty and multimodality through Markov Chain Monte
Carlo sampling. We demonstrate the rigour of our approach via a round-trip analysis of a simulated system
before applying it to real data, examining the well-characterised collapse of a thermoresponsive brush. The
method we describe is directly applicable to reflectometry experiments on soft and diffuse systems, but may
also be generalised to other instruments where the “inverse problem” hampers data analysis.
Description
Keywords
Reflectivity, Bayesian statistics, Surfaces, Neutrons, Physics, Data, ANSTO
Citation
Gresham, I., Murdoch, T., Johnson, E., Grant, W., Wanless, E., Prescott, S., & Nelson, A. (2020). Imaging the invisible: resolving polymer brush structure through a freeform Bayesian analysis of neutron reflectometry data. Paper presented to the ANBUG-AINSE Neutron Scattering Symposium, AANSS 2020, Virtual Meeting, 11th - 13th November 2020. (pp. 77). Retrieved from: https://events01.synchrotron.org.au/event/125/attachments/725/1149/AANSS_Abstract_Booklet_Complete_-_1_Page_Reduced.pdf