Skip to content

Pixel Design Playground

The comprehensive pixel design tool. Configure every parameter of a BSI CMOS pixel — from silicon thickness to BARL sublayers — and see the combined effect on QE, reflectance, energy budget, and stack geometry.

Pixel Design Playground

Configure all pixel parameters and explore multi-panel results: QE spectra, layer stack visualization, and energy budget analysis powered by TMM.

Pixel Parameters
Color Filter
Simulation Settings
0%20%40%60%80%400450500550600650700750Wavelength (nm)QE (%)BlueGreenRed
Peak QE R
65.5%
Peak QE G
85.1%
Peak QE B
68.2%
Avg QE (vis)
22.0%
Avg Reflectance
7.3%
Stack Height
4.58 um

Model scope

Use this browser tool for intuition, relative trends, and design-space exploration. Its local simplified model is not a substitute for RCWA/FDTD sign-off, silicon calibration, or vendor process data.

Physics Notes

Coupled Pixel Stack Model

Plain-English Intuition

This is a what-if sandbox for pixel design. Drag a slider — pixel pitch, filter thickness, silicon depth — and watch several metrics move at once because everything in a pixel is coupled. It is a quick first-pass tool to build intuition before you spend hours on a full RCWA or FDTD simulation.

The playground combines pixel pitch, color-filter thickness, BARL layers, silicon thickness, and incidence angle into one simplified stack-level optical response.

Assumptions

  • The playground combines simplified optical, geometry, and sensor-metric models into one exploratory view.
  • Interactions between microlens, CFA, BARL, DTI, silicon, and noise are approximate and should be validated in dedicated tools.
  • The goal is relative trend ranking, not final quantitative device prediction.

Outputs

  • Multi-panel summaries for optical efficiency, crosstalk tendency, full well, SNR, dynamic range, and scaling tradeoffs.
  • A first-pass parameter screening table before running slower RCWA/FDTD or calibrated compact-model studies.

Validation Example

  • Increasing pitch should generally raise photon collection and full well while easing diffraction pressure.
  • Turning off DTI or reducing isolation should increase crosstalk indicators rather than improve every metric simultaneously.

Core Equations

Stack response
$$QE_c(\lambda) \approx T_{\text{CF},c}(\lambda) \cdot A_{\text{Si}}(\lambda; \text{stack})$$
  • \(QE_c\): Quantum efficiency for color $c$
  • \(T_{\text{CF},c}\): Color filter transmittance
  • \(A_{\text{Si}}\): Absorbance in the silicon photodiode

The color channel response is approximated by filter transmission times silicon absorption.

Pixel area scaling
$$A_{\text{px}} = p^2$$
  • \(A_{\text{px}}\): Geometric pixel area
  • \(p\): Pixel pitch

Area changes photon count and full-well trends even if spectral QE is unchanged.

Energy budget
$$R + T + \sum A_{\text{layer}} = 1$$
  • \(R, T\): Reflectance and transmittance
  • \(A_{\text{layer}}\): Per-layer absorbance

The same conservation check is used to interpret stack losses.

Model Interpretation

  • Changing one slider can shift multiple metrics because optical stacks are coupled.
  • Use this for design-space triage before running RCWA or FDTD on a detailed geometry.
  • The model omits lateral field maps, carrier transport, and process variation.

Coupled Knobs

  • Reducing pitch lowers photon count through $A_{\text{px}}=p^2$ even when optical QE stays constant.
  • Changing color-filter thickness shifts both spectral separation and parasitic absorption.
  • Silicon thickness improves long-wavelength absorption but can increase crosstalk or carrier-collection burden in a real pixel.

Use As A Triage Tool

  • Use large metric changes to identify promising regions, then move those candidates into TMM, ray tracing, RCWA, or FDTD.
  • Compare relative trends rather than absolute numbers because the simplified browser model uses compact material spectra.
  • Treat any optimum at a slider boundary as a sign that the explored design range is too narrow.

Known Missing Physics

  • The playground does not solve lateral diffraction, microlens focusing, DTI confinement, or charge diffusion.
  • Process variation, thickness tolerance, and material dispersion uncertainty are not sampled statistically.
  • Use it to choose simulations, not to sign off a pixel stack.