Micro- and Nano-Mechanical Characterisation and Modelling of the Local Matrix Deformation in Fibrereinforced Epoxy

The prediction of the deformation and failure of fibre-reinforced polymer composites via bottom-up multi-scale models has become standard in the composite community. The development of accurate computational multi-scale models relies on the proper description, and thus characterisation of the individual components of the composite ply, i.e. fibres, matrix, and interfaces and interphases between the matrix and the fibres. However, the determination of the properties of these constituents at the micro/nano-scale remains a challenge. Additionally, the properties of the matrix are usually defined using continuum constitutive laws. Hence, there is a need for micro-/nano-mechanical characterisation methods to establish the matrix material response at the fibre/matrix level. These challenges place a limit on the accuracy of composite model predictions, even for simple unidirectional (UD) composites loaded in transverse compression or shear, where the matrix dominates the macroscopic deformation response of the composite. In this study, a combined experimental and numerical approach is used to characterise the individual constituents of a UD composite composed of carbon fibres and an epoxy resin. Emphasis is placed on the measurement and prediction of the constitutive response at a length scale close to the fibre diameter, where e.g. matrix size effects may exist. First, the local matrix deformation response in resin-rich pockets within UDs is probed by nanoindentation and atomic force microscopy (AFM). The extracted properties are compared with macro- and micro-scale properties of RTM6 from previous studies. Second, transverse compression tests on UD specimens are conducted inside a scanning electron microscope (SEM). The local strain field around the fibres is quantified using nano digital image correlation on a microscale region of interest (ROI). The DIC strain maps on a ROI are compared with those predicted via FEA using an established continuum model for RTM6.