GPGPU-assisted polymer nanocomposite modelling and characterisation

Abstract

Development of the hybrid materials with predefined properties by addition of inorganic nanoinclusions to a polymer material constitutes a hard challenge due to significant properties’ variations depending on inclusion’s distribution and interaction. To understand structure-property relations in such materials optical image analysis and numeric modeling are widely used, however matching such data with properties’ measurements for industrial nanocomposites requires a link to be established between experimental and modeling length scales. In this work a computer code was developed to create a model composite structure with a predefined distribution probability of inclusions using NVIDIA CUDA GPGPU approach. The code is capable of randomly populating and analyzing samples of the typical size of microphotographs used for experimental characterization and typical nanoinclusions’ concentrations avoiding unphysical intersections and thus allow correlating the results of both optical characterization and statistical computer modeling. The initial probability distribution can be taken from experimental samples and further varied to investigate the effect of distribution on a desired property. Application to study the effect of carbon nanotubes and carbon nanofibers in a polymer matrix on the composite electrical and mechanical properties is discussed.