the doping profile must be designed to avoid extreme local electric fields at the junction, which would otherwise exacerbate band-to-band tunneling and compromise the sensor's dynamic range and dark current performance .
Following the formation of the sidewall spacers (SWS) and subsequent cleanin
g, the NMOS Source/Drain (S/D) and Floating Diffusion (FD) Ion Implantation (IIP) step establishes the heavily doped N+ regions . This process step is fundamentally required to provide low-resistance contact regions for NMOS logic transistors and to form the charge-to-voltage conversion node—the floating diffusion—within the pixel architecture . Performing this high-dose implantation after the spacer formation ensures that the peak dopant profile is laterally offset from the gate edges . This spatial offset is critical because it mitigates short-channel effects in logic devices and restricts gate overlap capacitance at the transfer gate, which is essential for maximizing the conversion gain of the sensor . The subsequent ashing and cleaning steps will remove the photoresist mask used to selectively define these N+ regions, preparing the wafer for specialized N-type FD tuning implants . The implantation process operates by ionizing dopant atoms, accelerating them through an applied electric field, and directing them into the unmasked regions of the silicon lattice . As the high-energy ions penetrate the substrate, they undergo a series of elastic and inelastic collisions with the host silicon atoms, ultimately coming to rest at a projected range determined by the initial implant energy . The resulting spatial distribution of the implanted impurities can generally be approximated by a Gaussian profile, where the total dose dictates the peak concentration . Because the silicon substrate possesses strict crystalline symmetry , precise control over the incident tilt and twist angles is required to prevent channeling . Channeling occurs when ions align with open crystallographic planes and travel unusually deep into the substrate, distorting the intended shallow junction profile . To further randomize ion trajectories and capture surface contaminants, a thin amorphous screen oxide is often employed over the silicon surface during the implant . Arsenic is frequently selected as the dopant species for NMOS S/D implants because its high atomic mass and low diffusivity facilitate the formation of the shallow, abrupt junctions required for highly scaled transistors . For the floating diffusion region, the implant energy and dose must be meticulously optimized to govern both the p-n junction depth and the lateral diffusion spread . Since the conversion gain of the CMOS image sensor is inversely proportional to the total FD capacitance, restricting both the junction capacitance and the gate overlap capacitance via precise doping boundaries is a primary design objective . However, the heavy mass of arsenic and the high required dose inherently inflict substantial lattice damage by displacing silicon atoms from their equilibrium sites . To suppress excessive damage accumulation, the beam current density and irradiation time must be carefully modulated, as unmitigated crystal defects act as generation-recombination centers that elevate dark current . Furthermore, to protect the ultra-sensitive pixel regions from metallic impurities originating from the implanter's plasma source, mass-analyzing electrostatic angle-energy filters and RF plasma showers are rigorously utilized . In the context of a nanoscale Back-Illuminated (BSI) CMOS Image Sensor, scaling the pixel pitch demands highly engineered FD capacitance to achieve photon-countable sensitivity and sub-electron read noise . The S/D implantation must be perfectly balanced with the subsequent thermal budget to prevent Transient Enhanced Diffusion (TED) from driving the dopants too deeply under the transfer gate . If the N+ profile diffuses too far laterally, the ensuing parasitic overlap capacitance severely degrades the input-referred conversion gain . Concurrently, the doping profile must be designed to avoid extreme local electric fields at the junction, which would otherwise exacerbate band-to-band tunneling and compromise the sensor's dynamic range and dark current performance .
Sign in to continue through all 417 steps