In the 40nm BSI CMOS Image Sensor process flow, the Oxidation Pre-Cleaning step serves as the critical bridge between physical particle removal and the growth of high-quality thermal oxides .Its primary objective is to strip native oxides, metallic contaminants, and trace organic residues to expose a pristine, crystallographically ordered bare silicon surface .If native oxide or contaminants remain, they will incorporate into the subsequent thermal oxide layer, introducing electrically active interface states and degrading gate oxide integrity (GOI) .Because the subsequent oxidation step relies on the diffusion of oxygen species through the growing film to react with silicon atoms at the interface, a highly uniform starting surface is absolutely required to ensure uniform oxide thickness and reliable device electrostatics .The chemical mechanism of this step relies heavily on hydrofluoric acid (HF) based solutions or vapor to etch the native silicon dioxide layer .HF attacks the oxide by breaking Si–O bonds and forming soluble fluorosilicate species that are carried away in the rinse .In advanced nodes, dilute HF with the addition of HCl is often utilized to control the atomic-scale termination of the silicon surface .This specific chemistry favors hydrogen termination, where the surface silicon dangling bonds are saturated by Si–H bonds immediately after oxide removal .This hydrogen-passivated surface is metastable, providing a temporary chemical shield against re-oxidation in ambient air while the wafer is transferred to the oxidation furnace .Alternatively, HF vapor pre-cleaning can be employed to bypass dissolved oxygen and liquid surface tension effects, yielding a highly hydrophobic and smooth silicon surface free from watermarks .The selection of dilute HF rather than concentrated HF is driven by the thermodynamics of surface bond formation .While concentrated HF rapidly removes oxide, it promotes stronger Si–F bonding, which leads to non-uniform direct silicon etching, nanoscale surface roughening, and heterogeneous termination .Surface roughness directly degrades carrier mobility by increasing quantum mechanical scattering at the Si/SiO2 interface, an effect that becomes dominant in highly scaled inversion layers .Furthermore, the addition of oxidizing agents such as ozone-rich water prior to the HF dip is often used to remove residual carbon contaminants by oxidizing them into soluble forms, after which the chemical oxide is stripped to reveal the bare lattice .Removing metal and carbon contaminants prevents the formation of deep-level traps and carbon clusters at the interface, which would otherwise act as severe generation-recombination centers .At the 40nm technology node, specifically for BSI CMOS Image Sensors, this pre-cleaning step is critical for minimizing dark current .Interface trap density (Dit) arising from broken bonds, chemical disorder, or metallic impurities at the Si/SiO2 interface acts as the primary source of Shockley–Read–Hall thermal generation, directly manifesting as dark current and white pixel defects in the image sensor array .Furthermore, as physical oxide thicknesses scale down, equivalent oxide thickness (EOT) and tunneling leakage currents become extremely sensitive to the interfacial layer's exact thickness and composition .Therefore, achieving an atomically flat, perfectly hydrogen-terminated surface via meticulously controlled pre-cleaning is not just a yield enhancement, but a fundamental requirement to maintain the flat-band voltage stability and low-leakage performance of the 40nm device .
Multi-Layer High-K Gate Stack Materials for Low Dit 4H-SiC Based MOSFETs
2022
Atomic level termination for passivation and functionalisation of silicon surfaces.
2020
Integration of high-k/metal gate stacks for CMOS application
2008
CMOS compatible self-aligned S/D regions for implant-free InGaAs MOSFETs
2012
Physics of Semiconductor Devices - Full
2006
Substrate processing method
2023
Method of preparing a structured substrate for direct bonding
2024
Method of forming a sic/gate dielectric interface layer in a semiconductor device
2024
Questions about this step? Ask AI
AI process assistant backed by 10,000+ papers & patents