In chemical mechanical polishing (CMP), the injection position of the polishing slurry significantly affects the interfacial hydrodynamics, abrasive transport, removal efficiency, and overall planarization. This study systematically investigates the influence mechanism of the slurry injection position in the CMP process of 12-inch wafers via a multiphase flow–discrete phase coupling computational fluid dynamics (CFD) model combined with a user-defined function (UDF) to constrain abrasives. The results show that the injection position directly determines the distribution of slurry between the wafer and the polishing pad. At 45 mm from the pad center, the slurry effectively fills the gap, achieving the highest material removal rate (MRR). At 105 mm, the slurry is distributed most uniformly beneath the wafer, resulting in optimal planarization. However, at 165 mm, the slurry flow extends beyond the wafer center, causing abrasive agglomeration and localized overpolishing, which significantly decreases the surface uniformity. Dye visualization and CMP experiments with 12-inch copper wafers validate the accuracy of the model. The findings suggest that the slurry injection position should balance the material removal rate and planarity to optimize the slurry distribution system, providing a theoretical basis for future optimization efforts.

Chemical mechanical polishing (CMP) and post-CMP cleaning are critical steps in the current semiconductor manufacturing process. These processes require ensuring atomic-scale flatness and complete removal of contaminants. This review examines the use of molecular dynamics (MD) simulations to elucidate the atomic-scale mechanisms underlying CMP and postcleaning, focusing on four major MD methodologies: classical MD, reactive force field MD (ReaxFF), tight-binding quantum chemical MD (TB-QC MD), and ab initio MD (AIMD). Classical MD provides a foundation for simulating large-scale systems but lacks accuracy for modeling chemical reactions. ReaxFF allows real-time bond breaking and formation simulations during CMP. TB-QC MD combines quantum accuracy with classical efficiency, enabling exploration of the effects of chemical reactions on friction and material removal. AIMD directly calculates atomic interactions for precise depictions of chemical processes, although it is computationally expensive. MD simulations act as a “computational microscope”, enhancing CMP and postcleaning processes by quantifying interactions, material removal pathways, and contaminant desorption. Future research should address multiscale modeling challenges, improve AIMD efficiency, and develop accurate potential functions to propel semiconductor manufacturing toward greater precision and efficiency.