Skip to content

Angular Response Simulator

In real cameras, light hits pixels at oblique angles depending on sensor position and lens design. Chief ray angle (CRA) can reach 20–30° at the sensor edge, significantly affecting QE.

Angular Response Simulator

Explore how quantum efficiency changes with angle of incidence — critical for understanding Chief Ray Angle (CRA) effects in image sensors.

QE @ 0°45.3%
QE @ 15°48.8%
QE @ 30°63.1%
Half-power angle60°
0%20%40%60%10°20°30°40°50°60°Angle of Incidence (°)QE (%)550 nm

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

Chief-Ray Angle and Polarization Response

Plain-English Intuition

A tilted ray does not simply travel through the same stack sideways. It refracts at each interface, sees different effective thicknesses, and splits into s and p polarization responses. At the edge of a camera field, that planar-film angular rolloff is only one part of the story; microlens shift and finite-cone averaging decide whether the focused spot still lands on the photodiode.

Angular response explains how QE changes when light enters a planar pixel stack away from normal incidence. It combines Snell refraction, s/p Fresnel behavior, TMM phase changes, longer absorption path length, and normalization against the on-axis response.

Assumptions

  • The page evaluates planar-stack angular response, not a complete off-axis pixel with shifted microlenses and finite aperture.
  • s and p polarization are computed separately and can be averaged for unpolarized illumination.
  • CRA is represented as incidence angle at the stack entrance; lens-pupil cone averaging is handled by separate cone-illumination tools.

Outputs

  • Normalized QE or transmission versus angle, with separate s, p, and unpolarized trends.
  • Screening information for angular rolloff, color-channel imbalance, and when CRA compensation needs microlens or stack redesign.

Validation Example

  • At $\theta=0$, s and p responses should coincide for an isotropic planar stack.
  • Increasing angle should generally lengthen optical path and shift thin-film features; if no feature moves, check whether the stack is too simplified.

Core Equations

Layer refraction
$$\tilde{n}_0\sin\theta_0=\tilde{n}_j\sin\theta_j$$
  • \(\tilde{n}_0,\tilde{n}_j\): Complex refractive indices of incident medium and layer $j$
  • \(\theta_0\): External incidence angle or chief-ray angle in the planar-stack model
  • \(\theta_j\): Internal angle inside layer $j$

High-index silicon strongly reduces the internal ray angle, but upper polymer/filter layers can still see meaningful obliquity.

Polarized Fresnel amplitudes
$$r_s=\frac{n_i\cos\theta_i-n_t\cos\theta_t}{n_i\cos\theta_i+n_t\cos\theta_t}, \quad r_p=\frac{n_t\cos\theta_i-n_i\cos\theta_t}{n_t\cos\theta_i+n_i\cos\theta_t}$$
  • \(r_s,r_p\): Reflection amplitudes for s and p polarization
  • \(n_i,n_t\): Incident and transmitted side refractive indices
  • \(\theta_i,\theta_t\): Incident and transmitted angles at an interface

s and p responses can diverge strongly at high angles, especially near Brewster-like conditions.

TMM admittance at angle
$$\eta_j^{(s)}=\tilde{n}_j\cos\theta_j, \quad \eta_j^{(p)}=\frac{\tilde{n}_j}{\cos\theta_j}$$
  • \(\eta_j^{(s)},\eta_j^{(p)}\): Optical admittance of layer $j$ for s and p polarization
  • \(\tilde{n}_j\): Complex refractive index
  • \(\cos\theta_j\): Obliquity factor inside the layer

The same physical stack has different effective boundary conditions for the two polarization states.

Angular phase thickness
$$\delta_j(\theta_0)=\frac{2\pi}{\lambda}\tilde{n}_jd_j\cos\theta_j$$
  • \(\delta_j(\theta_0)\): Phase thickness induced by external incidence angle $\theta_0$
  • \(d_j\): Layer physical thickness
  • \(\lambda\): Vacuum wavelength

A phase shift with angle can move BARL minima and color-filter interference features across wavelength.

Relative angular response
$$AR_{p/s}(\theta,\lambda)=\frac{QE_{p/s}(\theta,\lambda)}{QE_{p/s}(0,\lambda)}, \quad AR_{\text{unpol}}=\frac{AR_s+AR_p}{2}$$
  • \(AR_{p/s}\): Normalized angular response for each polarization
  • \(QE_{p/s}\): Optical QE proxy for s or p polarization
  • \(AR_{\text{unpol}}\): Unpolarized angular-response estimate

Normalization removes absolute QE scale and highlights angular rolloff or angular gain.

Finite-cone average
$$\overline{AR}(\lambda)=\frac{\int_{\Omega}AR(\theta,\phi,\lambda)W(\theta,\phi)d\Omega}{\int_{\Omega}W(\theta,\phi)d\Omega}$$
  • \(\Omega\): Angular cone set by lens f-number and chief-ray direction
  • \(W(\theta,\phi)\): Angular weighting function of the optical system
  • \(\phi\): Azimuthal angle inside the cone

A real camera pixel receives a cone of rays, not a single plane wave, so edge-pixel behavior needs cone averaging.

Model Interpretation

  • Planar angular response is a stack property; full CRA response is a pixel property that also includes microlens focusing and lateral collection.
  • s/p separation is not optional at high angle because coating minima and Fresnel loss can split by polarization.
  • A good edge-pixel design usually needs stack tuning, microlens shift, and finite-cone validation together.

How To Read The Plot

  • If both s and p fall together, the loss is likely path-length or absorption driven; if they split, Fresnel/admittance effects are important.
  • A response above 1 at some angle can occur when interference minima shift in a favorable direction; it is not automatically an error.
  • Compare wavelength slices, not only broadband averages, because angular color shading is channel dependent.

CRA Design Implications

  • At the sensor edge, the chief ray may enter at high angle even when the lens cone has many nearby angles.
  • Microlens shift compensates the lateral focus displacement, while stack tuning controls planar-film throughput at that angle.
  • Use this angular page before expensive ray/FDTD studies to identify angle bands and wavelengths at risk.

Known Missing Physics

  • The model does not trace the focused spot to a finite photodiode or model microlens shift.
  • It ignores lateral diffraction, metal-grid shadowing, color-filter relief, and DTI-induced waveguide effects.
  • For finite aperture filters or Fabry-Perot stacks, integrate the full angular cone rather than relying on a single angle.

References