Colloidal silica slurries are typically selected because their continuous, mild hydration of the silicon surface prevents the direct mechanical fracturing that would otherwise inject new crystallographic defects .
In the fabrication of Backside Illuminated (BSI) CMOS image sensors, the bulk sili
con handle must be drastically thinned to allow direct backside illumination, which maximizes the fill factor and light absorption efficiency . Following mechanical grinding and the primary CMP1 step, which removes severe mechanically induced subsurface damage, the CIS Backside Si CMP2 step acts as the final, highly precise target thickness adjustment . This secondary polishing step is critical for achieving a stringent Remaining Silicon Thickness (RST) and minimizing Total Thickness Variation (TTV) across the wafer . Uniform substrate height is absolutely essential for BSI sensors because the silicon thickness directly defines the optical absorption path length for incident photons . If the RST varies across the wafer, the optoelectronic conversion efficiency—particularly for longer-wavelength red light that penetrates deeper into the silicon crystal—will exhibit severe non-uniformity, degrading image quality . Furthermore, this step prepares the pristine, ultra-smooth surface required for the subsequent oxide hard mask deposition and backside lithography steps (Engineering Practice). The material removal mechanism during this CMP step relies on a microscale plastic removal process driven by the strict synergy of chemical softening and mechanical abrasion . In a specifically formulated alkaline slurry environment, chemical components react with the silicon surface to form a hydrated or passivated layer with significantly reduced mechanical properties, typically characterized as a lower dynamic hardness . Subsequently, nanoscale abrasive particles suspended in the slurry, supported by the polishing pad asperities, engage the wafer in sliding or rolling contact to gently shear away the softened layer . The overall removal rate is macroscopically governed by Preston's equation, which dictates a linear dependence on down pressure and the relative velocity between the pad and the wafer . To achieve the extreme precision required for CMP2, the process shifts focus from bulk removal to atomic-level surface smoothing, necessitating highly uniform load distribution across all active abrasives . Real-time control of this precise thinning is facilitated by in-situ optical spectral measurements, which analyze thin-film interference reflections to invert the current film thickness without interrupting the polishing process . The selection of specific slurry chemistries and abrasive morphologies for CMP2 is intentionally tailored to favor extraordinary surface quality over raw material removal rate . Colloidal silica slurries are typically selected because their continuous, mild hydration of the silicon surface prevents the direct mechanical fracturing that would otherwise inject new crystallographic defects . Additionally, tuning the abrasive particle size distribution—such as utilizing precisely engineered mixed abrasive slurries—can optimize the interface contact state, shifting the wear mechanism from unpredictable three-body rolling to highly efficient two-body sliding wear . This controlled abrasive interaction minimizes the ineffective frictional energy dissipation that causes micro-scratches . To combat cross-wafer non-uniformity and temporal pad wear effects, advanced pad conditioning techniques are employed to maintain consistent contact area and pressure uniformity throughout the run . Integrating an equivalent light source spectrum methodology for in-situ thickness measurement isolates the true wafer reflection from the complex optical interference of the intermediate slurry and pad layers, guaranteeing the target RST is achieved with nanometer precision .
Sign in to continue through all 417 steps