What might make the spectral response of an image sensor uneven?

Thread Starter

tim.zander

Joined Jul 5, 2022
4
I came across the Sony CMV4000 image sensor specification that has an amazingly uneven, jagged spectral response in it.

Why does the spectral response have so many local minimums and maximums? What might be causing this?

Most photodiode spectral responses are a lot more even. What might be causing such a difference?

1665819727870.png
 

Attachments

drjohsmith

Joined Dec 13, 2021
852
I came across the Sony CMV4000 image sensor specification that has an amazingly uneven, jagged spectral response in it.

Why does the spectral response have so many local minimums and maximums? What might be causing this?

Most photodiode spectral responses are a lot more even. What might be causing such a difference?

View attachment 278465
I dont know ,
so a guess.

The CCD is not colour sensitive,
so its a coating on the front that makes the dots wavelength sensitive
so I'd say its the coating thats used,
 

MrSalts

Joined Apr 2, 2020
2,767
I came across the Sony CMV4000 image sensor specification that has an amazingly uneven, jagged spectral response in it.

Why does the spectral response have so many local minimums and maximums? What might be causing this?

Most photodiode spectral responses are a lot more even. What might be causing such a difference?

View attachment 278465
@drjohnsmith is on the right track.
Coated optics or coated sensors produce "newton's rings" effect which is, in this case, essentially varying performance of a broad-spectrum anti reflective coating. Where the thickness of the various coating layers create better and worse AR effects as a function of wavelength. Anti reflective coatings is a "general" term here. Anylayer of the stack up where a refractive index change existed can cause this same effect of thin layers are part of the stack up. Generally layers under 10 microns will show some jagged edges as you see and be less definite but noticeable as layers get thicker.
 

Papabravo

Joined Feb 24, 2006
21,302
Image sensing is a Quantum Mechanical process. Incoming photons with sufficient energy will usually, but not always, raise electrons in atoms to higher energy levels. When the electrons return to a lower energy level, they emit photons whose frequency and wavelength are also quantized. The graph referenced in your original post shows QE (Quantum Efficiency) of the incoming photons to knock electrons from the stable crystalline structure out of place causing an event that the device can detect. Certain transitions are favored over others causing the bumpy and probabilistic nature of the process. A smooth continuous response would be evidence that quantum mechanics was not in play.
 

MrSalts

Joined Apr 2, 2020
2,767
Image sensing is a Quantum Mechanical process. Incoming photons with sufficient energy will usually, but not always, raise electrons in atoms to higher energy levels. When the electrons return to a lower energy level, they emit photons whose frequency and wavelength are also quantized. The graph referenced in your original post shows QE (Quantum Efficiency) of the incoming photons to knock electrons from the stable crystalline structure out of place causing an event that the device can detect. Certain transitions are favored over others causing the bumpy and probabilistic nature of the process. A smooth continuous response would be evidence that quantum mechanics was not in play.
This is very unlikely. Spectrophotometers have no problem scanning a sample with smooth response - even without a double beam to subtract a standard or leaving the standard blank to get absolute response - a smooth response is obtained without the Newton ring effect (jagged response).

The discrete molecular orbits and "allowed"/"forbidden" transitions that you are referring to are for molecules in solvated matrices or in the gas state. In the solid state (as in solid state semiconductor detectors), there is a fermi band that determines conduction or not. Any photon of sufficient energy that is absorbed can bump an electron into the fermi band to cause a signal as there are, essentially, a statistical "infinite" number states for absorbed electrons to occupy.

As I posted above, the jaggedness is a function of reflection losses caused by changes in refractive index and the fresnel equations that describe reflection vs transmission through a refractive index transition are wavelength dependent and depend on the thickness of each layer of the stack up of materials used to detect or focus the light.
 

DickCappels

Joined Aug 21, 2008
10,246
Most color sensing application, color TV cameras, color meters, etc. use filters to pass specific bands of wavelength, and the bands in all cases I have seen are much wider than the ripples shown in your illustration. As such, the small ripples should have have a negligable effect of color measurement.
 
Top