(531e) Quantifying Material Structural Component Evolution Using Scattering Data and Mixed-Integer Network Component Analysis

Tolle, I. - Presenter, Rensselaer Polytechnic Institute
Martin, L. L. - Presenter, Rensselaer Polytechnic Institute

In this work, we present an optimization-based methodology focused on extracting process-dependent structural component evolution in polymer materials. Through the decomposition of multivariate X-ray (SAXS/WAXS) or neutron scattering datasets using mixed-integer network component analysis (NCA), a unique set of component scattering signatures and corresponding component fraction evolution can be obtained without assumptions on polymer structural geometry. Analysis of the resulting decomposition can thus aid in scattering model selection and identifying structural evolution mechanisms. The methodology is applied systematically using X-ray data for the study of ethylene/alpha-olefin copolymer isothermal crystallization. The validity of the decomposition is first confirmed through extraction of amorphous and crystalline component signatures present in WAXS data. It is then applied in the decomposition of SAXS data to study nanometer scale structural evolution, where significant ambiguity in the data interpretation can exist. The decomposition generates a structural evolution mechanism that can be correlated to structural components composed of ordered and disordered crystalline lamellae. Application of this methodology to the study multi-phase materials using neutron scattering and contrast variation is also presented. The components generated correspond to the contributions from different phases, thus aiding in model selection when minimal a priori knowledge is available.