Developing the next generation of analysis

X-ray and neutron reflectometry techniques probe the nanoscale structure of interfaces. These measurements may be performed at large scale facilities, such as Diamond Light Source (Beamline I07) or ISIS Neutron and Muon Source. In order to grow the use of this technique, in particular to engage non-traditional users, it is important to create methods to easily, and reproducibily, analyse the resulting data.

This work involves assisting users of the large scale facilities by helping them to develop data analysis strategies before their beamtime, such that the data may be analysed quickly, and in an automated fashion, as it is collected. For example, we have produced “chemically-consistent” analysis models for lipid monolayer reflectometry [1]. If you are using the large scale facilities and are interested in collaborating to make use of this “data analysis as a service”-like system, please get in touch.

Alongside this work, we are investigating the use of convolutional neural networks to classify reflectometry profiles into a given analytical model types. The belief is that this will allow for a suggested analytical model to be used, in effect automating the role of the data analysis as a service. Furthermore, this may offer greater insight into the structures that may be observed by reflectometry as well as reduce the time taken to obtain publishible results.

Finally, through the Ada Lovelace Centre funded project we are aiming to develop a data standard to reduced, experimental reflectometry data. The hope is that this standard will allow for straightforward tagging of the experimental data in order to produce extensive data catalogues that may be subject to data mining and machine learning applications.


  1. A. R. McCluskey, A. Sanchez-Fernandez, K. J. Edler, S. C. Parker, A. J. Jackson, R. A. Campbell, & T. Arnold. Phys. Chem. Chem. Phys., 21(11), 6133-6141, 2019. DOI: 10.1039/C9CP00203K.