8 February 2026 in ICCK Transactions on Advanced Computing and Systems
The escalating demand for efficient particle separation in microfluidic systems necessitates innovative design solutions. This study presents a simulation-based topology optimization method to passively separate particles within a 2D microfluidic channel, eliminating the need for external forces. Leveraging a coupled Navier-Stokes solver and particle advection simulation, the framework iteratively refines the channel's geometry by minimizing an objective function quantifying particle mis-sorting. Our approach computationally generated optimal, manufacturable topologies, demonstrating a peak sorting efficiency of 0.6667 (66.67%) achieved by the second iteration, which then stabilized in subsequent iterations to 0.6111 (61.11%), significantly surpassing the adaptability and robustness of traditional, manually designed microfluidic channels. This work provides a robust, physics-based framework for exploring complex design spaces, representing a significant advancement in the development of high-performance, next-generation microfluidic devices.