As interconnect dimensions shrink, Cu CMP performance metrics—such as removal rate selectivity, dishing, and erosion—become strongly dependent on local pattern density rather than blanket wafer characteristics .
The Chemical Mechanical Planarization (CMP) of Copper at the Metal 2 (MET2) module se
rves to remove the electroplated Cu overburden, achieving global planarization for the nanoscale Backside Illuminated (BSI) CMOS Image Sensor . This step is intentionally positioned after the M2 Cu deposition and prior to the Ta-based liner CMP . By utilizing a specialized two-step approach, the bulk Cu is first cleared while stopping precisely on the Ta-based liner, which prevents excessive damage to the delicate underlying structures . What distinguishes this specific MET2 CMP step from similar Cu CMP steps in the flow is its intermediate routing layer geometry; M2 typically features different line-width/spacing ratios and pattern densities compared to local M1 or thicker global top-metal levels, heavily influencing local contact mechanics and slurry transport . The fundamental mechanism of Cu CMP is a synergistic coupling of chemical surface modification and mechanical abrasion . Oxidizers in the slurry chemically convert the top atomic layers of the copper surface into a softer, passivated oxide or complexed state . Simultaneously, applied normal force and relative sliding velocity between the polishing pad and the wafer drive mechanical removal, generally following the Preston equation behavior where removal rate is proportional to pressure and velocity . However, at the nanoscale, the removal is dominated by mechanical action rather than pure electrochemical dissolution . Sliding friction reduces the Hertzian contact pressure required for plastic deformation on the Cu surface, while chemical oxidation and repeated abrasive scratching introduce crystal defects that lower the local yield threshold . This allows abrasive particles to preferentially induce plastic deformation and micro-cutting in the weakened Cu regions . Slurry formulation is critical for achieving the necessary planarization efficiency and selectivity . Complexing agents, such as amino acids, are often used to form stable, soluble Cu complexes that accelerate removal, while choline salts or specific pH regulators modulate the interfacial electrochemical environment to suppress uncontrolled dissolution , . To address the demanding defectivity requirements, nonionic dispersants can be added to the slurry to induce steric hindrance . This prevents colloidal silica abrasives from agglomerating into large particles under high shear forces, which is essential to inhibit the formation of deep micro-scratches on the soft Cu surface . The choice of a highly selective slurry is dictated by the need to halt the polishing process uniformly across the wafer once the Ta-based liner is reached, minimizing variations prior to the subsequent barrier CMP step . Operating at the 40nm node introduces severe scaling challenges for this process . As interconnect dimensions shrink, Cu CMP performance metrics—such as removal rate selectivity, dishing, and erosion—become strongly dependent on local pattern density rather than blanket wafer characteristics . Uncontrolled dishing leads to substantial fluctuations in the sheet resistance of the M2 lines, which can degrade the signal timing and RC delay characteristics of the semiconductor device . Furthermore, variations in local pad pressure due to tight pitch geometries demand meticulous tuning of both the slurry chemistry and the applied mechanical downforce to ensure uniform planarization across the die .
Sign in to continue through all 417 steps