John Francis wrote:
There are competing demands; to shrink the size of an individual sensor,
and to increase the precision of measurement (roughly corresponding to bits per pixel). We're not at the technologically imposed limits yet,
but getting beyond the next generation or two is going to require some
changes to either the materials used or the design of the sensor.
The big problem in shrinking the individual sensor (photosites) is not the material or the photomasking, etc. Its S/N ratio. There is just too much noise. While this is also a nuisance in digital logic circuits, its not as critical as a analog oriented application. Since digital logic only comes in levels 0 or 1, you can work around the S/N problems easier. On the other hand, trying to get clean, distinct measurements out of a photosite requires gathering 4096 levels of data (for 12 bits) is tough, a tiny bit of noise will make the measurement skip 10's and possibly 100's of levels. Cooling the sensor helps alot, but this brings up power issues. One technological improvement that would help would be 3D photolithography, this might enable the photosites to be larger while being denser, since the routing (wires) could be underneath. Routing currently takes up a huge amount of space on the surface, the use of microlenses above the sensor concentrate the light to the small sensor photosites. These are exciting times, there is no doubt that there are opportunities for small, nimble companies to create new technologies that can change the industry practically overnight.
rg
