Self-similar mixing structures, a novel feature of chaotic mixing, were generated in this study as precursor to an array of mixing microstructures, such as nested layers, elongated fibrils, droplets and their combinations in the blending of two immiscible polymers, polypropylene (PP) and polyamide-6 (PA6). Simulations based on Newtonian flow model were used to compute Poincare maps and stretching distribution as the tools for investigation of the effect of shear gap and chaotic mixing parameter, such as angular displacement per period (0) of rotors, on the degree of mixing and morphology development in a batch chaotic mixing device. It was found that a value of 0 = 1440degrees provided the conditions for fastest conversion of the PP-phase into droplets for the same total strain. A 25% reduction in shear gap from 0.0127 in to 0.0095 in gave rise to much more uniform mixing of the components and led to faster conversion of the PP-phase into droplets for the same value of 0 and the same total strain. A very large fraction (> 90%) of the droplets generated fell below the equilibrium size and were found to be much smaller than those produced by twin-screw extrusion method.