At the nanoscale, the mechanical removal is dominated by abrasive sliding friction, which significantly reduces the Hertzian contact pressure required to induce plastic deformation in the copper .
Following the deposition of the Ta-based liner, Cu seed, and the bulk Metal 1 (M1) Cu electroplating
, the Cu CMP step is deployed to remove the bulk Cu overburden and define the isolated M1 interconnect lines . This specific step focuses exclusively on clearing the excess copper while deliberately stopping on the underlying Ta-based liner, which constitutes the first phase of the two-step interconnect planarization process . By doing so, it planarizes the wafer surface and prepares the structure for the subsequent Ta-based liner CMP and Post-CMP cleaning modules . As the first metal routing layer (MET1) in the 40nm BSI CMOS Image Sensor flow, this step faces the tightest pattern densities and pitch requirements compared to higher-level BEOL Cu CMP steps, demanding extremely precise planarization to prevent short circuits while maintaining structural integrity . The material removal mechanism during Cu CMP is a synergistic coupling of chemical oxidation and mechanical abrasion . In the slurry, oxidizers such as H2O2 react with the copper surface to form a passivating oxide layer . Mechanically, material removal follows Preston's equation, where the removal rate is proportional to the applied downward pressure and the relative velocity between the polishing pad and the wafer . At the nanoscale, the mechanical removal is dominated by abrasive sliding friction, which significantly reduces the Hertzian contact pressure required to induce plastic deformation in the copper . Simultaneously, the chemical oxidation and repeated mechanical scratching introduce crystal defects and surface roughening, which lower the local yield threshold of the copper surface . This allows the abrasives to remove the passivated copper through selective plastic deformation and micro-cutting, generating debris that is subsequently dissolved by complexing agents in the slurry . The selection of slurry chemistry and process parameters is driven by the need for high planarization efficiency and strict defect control . Because copper is inherently softer and chemically more active than the surrounding dielectric, it is highly susceptible to localized over-polishing, known as metal dishing . To mitigate this, specialized slurry formulations often incorporate dispersants, such as ethoxylated decyl alcohol (EDA), to introduce steric hindrance and prevent the agglomeration of colloidal silica abrasives . By improving colloidal stability, the concentration of large particles is minimized, which effectively suppresses the high localized contact stresses that cause detrimental micro-scratches . Additionally, surfactants may be utilized to selectively adsorb in recessed regions of the wafer, forming a protective boundary layer that inhibits further chemical and mechanical interaction, thereby enabling self-limiting planarization without excessive dishing . Operating at the 40nm node introduces severe pattern-dependent challenges that cannot be predicted by blanket wafer removal rates . As interconnect dimensions shrink, local contact mechanics and slurry transport become highly non-uniform, drastically altering the Cu-to-liner selectivity on high-density patterned structures . If the effective dishing radius increases due to these pattern effects, the resulting loss in metal thickness directly reduces the conductive cross-sectional area, degrading electrical performance through increased resistance . To extend the CMP process window and prevent interconnect structural degradation, integration schemes must rely on advanced slurry formulations and strict endpoint detection, sometimes incorporating planarization stop layers with distinct material selectivities to halt the polishing process precisely .
Sign in to continue through all 417 steps