(656d) Simulation of Neutron Dark Field Interferometry Data in Hierarchical Materials Using Small Angle Scattering Models

Hierarchical materials can be found across many fields, including electrodes for alternative energy, pharmacology, biology, colloidal science, geology, construction, additive manufacturing, polymer science, and more. While small-angle neutron scattering (SANS/USANS) is a useful measurement tool for characterizing the structure in both hard and soft matter at length scales between 1 nm and 10 µm, it only provides a beam-averaged view of the structure, making the study of inherently heterogeneous materials difficult and limiting our understanding of the structure-function relationship that can span many length scales. In response, a new neutron far field interferometer (dubbed “INFER”) is currently under development at the National Institute of Standards and Technology (NIST) that will enable the collection of spatially resolved, in three dimensions, structural information at the same length scales as SANS and USANS. This instrument will generate tomographic reconstructions with voxels on the order of 50 µm that each captures the local structure of the sample through the dark field intensity, which is related to the small angle scattering intensity through a single Hankel transformation. To analyze the large amounts of data expected with this approach (~ 10⁵ – 10⁶ correlograms or a terabyte of raw data per day), machine learning and data science approaches will be critical to segmenting our sample into structurally-similar regions of interest. In this work, we discuss our recent progress in developing physics based INFER data simulation tools that enable us to generate the large amounts of training data required for these models during this time of instrument development. We make use of the existing libraries of form factors and structure factors in SasView that model a wide range of structural systems and materials in the SANS/USANS space and convert the data into the INFER space using the Hankel transformation. Moreover, this approach allows us to use our understanding of structural systems in the SANS or Fourier space to understand how structures appear in the INFER or correlation space and further define the applications and limitations of our instrument.