Several chemical–mechanical planarization characterization test wafers were polished to understand the polishing mechanism of the fixed abrasive process. Oxide thickness removal in the “active” (up) and the “recessed” (down) regions of the wafer was monitored for different times of polish. It was found that there was no significant removal in the recessed areas until the step height was reduced to about 100 Å, and the polish rate in the active area decreased rapidly once this critical step height had been attained. At this critical step height, the polish rate of the down areas started to increase and approached that of the up area, with both eventually reaching the negligibly low removal rate of the blanket wafer. The drop in the polish rate of the up area, after planarity had been attained, was fitted to an exponential model.