Modeling the effect of particle clustering on the mechanical behavior of SiC particle reinforced Al matrix composites

Abstract

The degree of clustering of particles has a significant influence on the mechanical behavior of particle reinforced metal matrix composites (MMCs). The clustered particles act as crack initiation sites and generally have a negative effect on tensile strength, ductility, toughness, and fatigue strength of the composite [1–10]. Murphy et al. [9] examined the tensile behavior of a 20% SiC particle reinforced Al–Si composite with different degrees of clustering (by controlling the cooling rate during solidification of the composite). It was shown that an increase in particle clustering yielded a higher work hardening rate, with a significant reduction in ductility. It has been suggested that the matrix flow in the particle cluster is significantly constrained, which results in the premature local onset of crack initiation [10–12]. Very few studies have explicitly modeled the effect of particle clustering [10–14]. Segurado et al. [10] recently investigated the effect of particle clustering on stress–strain behavior using the finite element method (FEM). They found that if particle cracking is not considered in the model, the influence of particle clustering on the predicted stress–strain behavior is not significant. While crack propagation was not explicitly modeled, the fraction of fractured particles as a function of applied strain was estimated by incorporating a Weibull distribution in strength of the particles. It was found that the presence of clustering greatly increased the fraction of fractured particles. In this study, we have conducted a two-dimensional FEM simulation to quantify the effect of clustering on local and macroscopic stress–strain behavior of Al–SiCp composites. The models explicitly incorporate cracking of the particles for two levels of particle clustering. Two model Al/SiCp microstructures, consisting of circular SiC particles arranged to obtain very different degrees of clustering, were generated using image analysis (Image J, Bethesda, MD, USA). A detailed description of the image segmentation process is given elsewhere [15]. The SiC particles were represented as circular particles and the area fraction of particles was kept constant at 30%. Several techniques have been used to quantify clustering of particles in a composite [2, 16, 17]. Yang et al. [16] have shown that the coefficient-of-variance of the mean nearneighbor distance (COVd) is particularly sensitive and effective in characterizing clustering. This parameter is also relatively insensitive to particle volume fraction, size, and morphology. COVd can be described by the following equation [16]: