Micromixing is a key step in realizing fast analysis time in many bio-chemical, biological and detection applications of lab-on-a-chip (LOC) devices. The conventional T-mixer design requires longer channel lengths and times to achieve complete mixing owing to its dependence on transverse diffusion. As the surface properties of the microchannel govern the electro-osmotic flow characteristics, surface heterogeneity (non-uniform zeta potentials) can be exploited to generate vortices or specific flow structures to improve the mixing performance. Previous studies have shown that localized circulations or non-axial flow induced due to the presence of heterogeneity augment micromixing performance. However, the effect of heterogeneous charge patterns on mixing performance has not been studied systematically. In this computational study, a binary numerical optimization problem is formulated to achieve best mixing performance by identifying the optimal heterogeneous charge pattern. The resulting optimal design generates the most favorable transverse flow structure to provide optimal mixing performance. Various other configurations (staggered, herringbone, etc.) are examined over a range of operating conditions. The optimal design is found to be superior for all operating conditions with over 3-fold improvement in mixing performance with respect to homogeneous T-mixer. (C) 2012 Elsevier Ltd. All rights reserved.