40nm BSI CMOS Image SensorPreview

Ashing & Strip/Clean

68/ 417

Implant Oxide Removal

Sacrificial Oxidation
65NMOS VT Adjust Implant Mask Lithography66NMOS VT Adjust Ion Implantation67Ashing & Strip/Clean68Implant Oxide Removal

Process Cross-Section

NMOS · N4 · Implant Oxide RemovalVT adjust (11B+, periphery channel)SiO2 liner (SACVD)SiO2 (SACVD bulk fill)n+ N-well contact (31P+)p+ surface passivation (10B+)Liner SiO2 (RTP thermal)P-well (pixel array, 11B+)P-well (periphery, 11B+)PD N-well (31P+)SiNN-well (periphery, 31P+)SiP-well (implanted region)

Step highlight

HF breaks Si-O bonds to dissolve the oxide layer, exposing a hydrophobic silicon surface that prevents contamination during subsequent implant steps .

In depth

After the NMOS threshold voltage (VT) adjust ion implantation and the subsequent photoresist ashing and cleaning, the thin implant screen oxide mu

st be removed . This oxide was originally placed over the channel region to randomize the incoming ion trajectories, prevent channeling effects, and protect the bare silicon surface from metallic and organic contamination during the implant process . Because the high-energy ion bombardment physically damages the oxide network and embeds contaminants, leaving it in place would severely degrade the quality of the final gate dielectric . Therefore, a complete chemical removal of this compromised layer is required to expose a pristine silicon surface before growing the sacrificial oxide (SACOX) (Engineering Practice). Unlike a thick oxide hard mask removal, which requires aggressive etching to remove a structural blocking layer , this step targets a very thin, damaged screen oxide where minimizing substrate silicon loss and preventing isolation structure degradation are the primary concerns (Engineering Practice). The removal process relies on the chemical reaction between hydrofluoric acid (HF) and silicon dioxide . The HF effectively breaks the strong Si-O bonds, forming water-soluble complex ions that are easily rinsed away . Interestingly, the etch rate of this implant oxide is heavily influenced by the preceding ion implantation step; the physical collisions of the implanted dopants break local Si-O-Si bonds and introduce structural defects, rendering the oxide significantly more susceptible to chemical attack compared to unimplanted thermal oxide . This damage-enhanced etch rate means the oxide can be cleared very rapidly under controlled conditions (Engineering Practice). However, as the oxide clears, the bare silicon surface becomes exposed to the processing environment, rapidly transitioning from a hydrophilic oxide surface to a highly hydrophobic, hydrogen-terminated silicon surface . Dilute hydrofluoric acid (DHF) is universally selected for this step because it provides near-infinite selectivity to the underlying silicon substrate, ensuring that the critical channel silicon is not consumed . Although wet HF dipping is standard, careful control of dissolved oxygen in the bath is required to prevent concurrent re-oxidation and subsequent roughening of the fresh silicon surface . Alternatively, HF vapor cleaning can be employed to minimize dissolved oxygen issues, avoid water marks, and reduce metal contamination, ultimately producing an extremely smooth surface . The primary parameter interaction involves balancing the etchant concentration and exposure time: too short an exposure leaves residual damaged oxide and implant-induced contaminants, while excessive over-etching risks attacking the adjacent shallow trench isolation (STI) structures and causing unwanted topography . In a 40nm CMOS image sensor process, precise control of the NMOS threshold voltage is paramount for pixel readout and logic performance, necessitating exceptionally low defect densities at the surface . Because the subsequent steps involve growing and removing a sacrificial oxide to heal surface damage before the final thin gate oxide growth, any micromasking by residual implant oxide during this step will propagate roughness through the entire gate stack . Furthermore, the 40nm node's aggressive equivalent oxide thickness (EOT) targets leave zero tolerance for the mobility degradation that would result from a compromised initial silicon surface .

Risks & Challenges

  • [High] Incomplete Oxide Removal (Micromasking): If the damaged screen oxide is not fully cleared, residual patches will act as masks during the subsequent sacrificial oxidation step . This leads to non-uniform SACOX growth and highly localized surface roughness, which ultimately degrades the final gate oxide integrity and shifts the threshold voltage .
  • [Medium] Excessive STI Oxide Recess: The HF chemistry used to strip the implant oxide also inevitably etches the adjacent shallow trench isolation (STI) dielectric . Prolonged over-etching causes excessive STI recess, exposing the silicon channel edges and creating parasitic corner devices with lowered threshold voltages .
  • [Medium] Surface Roughening via Dissolved Oxygen: If a wet HF process is used without adequate deoxygenation, dissolved oxygen in the chemical bath can simultaneously oxidize the freshly exposed silicon while the HF etches it . This continuous etching-oxidation cycle micro-roughens the silicon surface, which directly translates to channel mobility degradation in the completed MOSFET .
  • [Low] Contamination Re-deposition: The implant screen oxide traps metallic contaminants originating from the ion implanter beamline . If the cleaning or rinsing step following the HF etch is insufficient, these liberated metallic impurities can re-deposit onto the hydrophobic silicon surface, forming deep-level traps that increase junction leakage .

Sign in to continue through all 417 steps

Sign up with emailLog in

Related steps

  • NMOS VT Adjust Implant Mask Lithography
  • NMOS VT Adjust Ion Implantation
  • Ashing & Strip/Clean
  • NMOS LDD Implant Mask Lithography
  • NMOS LDD Ion Implantation
  • Ashing & Strip/Clean