40nm BSI CMOS Image SensorPreview

Metal 5 Cu deposition

239/ 417

Cu CMP

Ta-based liner CMP
236Ta-based liner deposition237Cu Seed deposition238Metal 5 Cu deposition239Cu CMP240Ta-based liner CMP241Post CMP Cleaning242ILD 6-1 Deposition243ILD 6-2 Deposition244Pre Litho Cleaning

Process Cross-Section

MET6 · Cu CMP (stop on Ta)IO/HV gate ox (SiO2; relative thickness shown schematically; IO device not shown in this cross-section)TiSi (low-temp anneal)n- LDD (31P+, self-aligned)P-pinning (11B+)P+ VSS contact (11B+)FD node (31P+)VT adjust (11B+, periphery channel)SiO2 liner (SACVD)SiO2 (SACVD bulk fill)n+ S/D (31P+)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+)N-well (periphery, 31P+)SiP-well (implanted region)CuTaSiO2CESLPMD 5 (SiO2)PMD 4 (SiO2)MET0 (W)PMD 3 (SiO2 · CMP overburden)W (contact fill)PMD 2 (SiO2 · body segment)TiN (barrier)Ti (adhesion)Ti/TiN linerPMD 1 (SiO2 · bottom segment)CESL 2 (SiNO)SiNCESL 1 (SiN)PolySWS pad ox (SiO2, PECVD)gate ox (SiO2, thermal)

Step highlight

inhibitors like triazole compounds are utilized to form a passivating protective film on the Ta-based liner, ensuring that the polishing process abruptly slows down once the bulk copper is cleared .

In depth

The Cu CMP step in the MET6 module serves to remove the electroplated bulk copper overburden and a

chieve global planarization, strictly stopping on the underlying Ta-based liner . As the upper metallization layer in a 40nm BSI CMOS Image Sensor flow, MET6 typically features wider global routing lines and optical shielding structures compared to the finer M1-M3 interconnects . Consequently, this specific CMP step must handle larger localized volumes of copper removal while heavily mitigating layout-dependent topography variations over wide trenches . This step isolates the Cu interconnects within the dielectric trenches and prepares the wafer for the subsequent dedicated Ta-based liner CMP step, which will ultimately expose the underlying ILD . The fundamental mechanism of Cu CMP is a highly coupled process of surface electrochemical oxidation and nanoscale mechanical abrasion . Oxidizers in the slurry chemically convert the metallic copper surface into a softer, removable oxidized state or soluble complex . Macroscopically, the material removal rate is often modeled by the Preston equation, scaling proportionally with applied pad pressure and relative sliding velocity . Microscopically, material removal is driven by selective plastic deformation; the sliding friction of the abrasives significantly reduces the Hertzian contact pressure required to induce plasticity in the copper substrate . Furthermore, repeated abrasive scratching and chemical corrosion introduce surface roughening and crystal defects that lower the local yield threshold, allowing the abrasives to micro-cut the weakened copper efficiently . Slurry formulation is the most critical control vector for this step, as it must provide extremely high Cu-to-barrier selectivity to prevent premature barrier breakthrough . Complexing agents, such as amino acids, are selected to form stable, soluble complexes with the oxidized copper to enhance the chemical removal rate at a neutral or slightly acidic pH . Simultaneously, inhibitors like triazole compounds are utilized to form a passivating protective film on the Ta-based liner, ensuring that the polishing process abruptly slows down once the bulk copper is cleared . To prevent mechanical defect generation, nonionic dispersants may be added to the slurry to improve the surface wettability of the silica abrasives and suppress large particle agglomeration via steric hindrance . At the 40nm node, interconnect structures exhibit pronounced pattern density effects, making the CMP process highly sensitive to local contact mechanics and non-uniform mass transport . Because copper is physically softer and chemically more reactive than the surrounding barrier and dielectric materials, wide structural lines are particularly vulnerable to metal dishing . This layout-dependent dishing effect directly reduces the conductive cross-sectional area, which subsequently increases line resistance and degrades the electrical RC performance of the global routing network .

Risks & Challenges

  • [High] Metal Dishing: Because copper possesses a lower hardness and higher chemical reactivity compared to the Ta-based liner and dielectric, it experiences a higher localized removal rate under identical pad pressure and slurry conditions . This differential removal forms a cylindrical depression in wide interconnects, which reduces the cross-sectional area and increases interconnect resistance .
  • [High] Micro-scratching: The aggregation of colloidal silica abrasives into large particles under Brownian motion and shear conditions can introduce localized high contact stresses at the pad-wafer interface . These hard, sharp agglomerates dynamically exceed the plastic yield threshold of the soft copper, creating deep micro-scratches that degrade device yield and reliability .
  • [Medium] Loss of Liner Selectivity: If the dynamic concentration of chemical inhibitors in the slurry is insufficient to maintain a dense passivating film on the exposed Ta-based liner, the chemical-mechanical action will prematurely erode the barrier . This degrades the barrier integrity and alters the local stress distribution, potentially leading to subsequent dielectric erosion .
  • [Low] Copper Residue (Underpolishing): Variations in local pattern density can alter the slurry transport and contact pressure distribution, locally reducing the mechanical abrasion and chemical reaction rates . This uneven polishing can result in the incomplete removal of the copper overburden on the barrier layer, causing fatal electrical shorts between adjacent interconnect lines .

Sign in to continue through all 417 steps

Sign up with emailLog in

Related steps

  • METAL 6 TRENCH - Photo
  • ILD 5-2 Oxide Etch
  • Ashing
  • Ta-based liner deposition
  • Cu Seed deposition
  • Metal 5 Cu deposition