1

Just read here that Samsung has produced a camera sensor with a pixel size of 0.8 micron ie 800nm. Is it possible, or useful, for the pixel size to go below the wavelength of the light being imaged? For example, would 350nm size pixels be of any use sensing (say) green light at 530nm? Or have we reached the limit?

Dirk Bruere
  • 13,425
  • 9
  • 53
  • 111
  • 2
    I'm voting to close this question as off-topic because its about the limits of technology and not design. – Warren Hill May 10 '19 at 08:53
  • @WarrenHill I **think** that in some sense it is related to antenna theory. – Harry Svensson May 10 '19 at 09:55
  • 2
    The dark-current noise is important parameter for camera pixels. And the sensitivity is linearly proportional to AREA. Thus a 0.8micron (square) pixel has 100X (7 Fstops) lower sensitivity than a 8 micron-square pixel. How does the dark-current noise change? lets see how others answer. – analogsystemsrf May 10 '19 at 14:01

2 Answers2

3

I don't know about the standard sensor technologies used (which capture charge from a photodetector, as far as I am aware). But to show that we can build smaller sensors:

Nantennas (I didn't come up with the term) do exist and have been demonstrated. As we can make antennas smaller than the wavelength (significantly so, in fact) I don't see any reason why it wouldn't be possible to make smaller pixel sensors. For example, a dipole will be about half a wavelength from one end to the other - this gives us about 250 um size of antenna for 500 um wavelength.

Also keep in mind that while the wavelength of visible light in free space might be say 500 nm, due to the refractive index of the materials used in a sensor, it will be shorter in the material itself, and as a result we can have smaller pixels.

But there is a catch. We should make the distinction of how big is the actual pixel and how close together can we place the pixels

We can make antennas that might be 1/2 wavelength, but that does not mean we can put then 1/2 wavelength apart and still get good results. The antenna is not like a bucket that can only capture what falls exactly on top of it, but it has an effective aperture (the size of the proverbial bucket) that can be bigger than the actual antenna itself. Overlapping antennas with large effective apertures will have them influence each other reducing performance, and on top of that, would perhaps not be useful because they are 'seeing' the same thing - similar to what you get when you take an image of 100x100 pixels and then tell Photoshop to up-scale it to 200x200 pixels - you just end up with more pixels showing the same data.

Joren Vaes
  • 12,376
  • 33
  • 64
2

There is nothing about light detection that requires the sensor area to have any particular relationship to the wavelength of the light.

The pixels are basically photodiodes, in which photons interact with individual electron-hole pairs, which are MUCH smaller than the wavelength associated with the photon.

Dave Tweed
  • 168,369
  • 17
  • 228
  • 393