Finite Element-Based Calibration of Effective Interphase Properties in Composites Reinforced with Randomly Distributed Spherical Particles

Composite materials generally consist of a matrix reinforced by one or more fillers each having unique material properties. Fillers of sub-micron length scales often induce an interphase region between a matrix and filler with unknown distinct physio-chemical properties. Determination of the mechanical properties of this region is important for the tailoring of composite materials through advanced structure-property engineering. However, attempts to experimentally measure the interphase properties directly are challenging. In this study a novel framework is proposed to model and calibrate the stiffness and thickness of an interphase region against experimental measurements of the bulk composite stiffness, a property which can be readily measured. This framework is demonstrated for a matrix reinforced by randomly distributed spherical particles but is generalizable to any composite system. Using finite element analysis, representative volume elements (RVE) are loaded with far-field strains and the resulting stresses are used to determine the effective stiffness of the simulated composite. Gaussian process regression is used to generate computationally cheap, surrogate models of the finite element outputs. The surrogate models are then explored to find suitable interphase parameters that best match experimental data. The results indicate a suitable alternative for the mechanical characterization of interphase regions.