The physics is pretty simple.
Small pixels = small full-well capacities. As far as I am aware there are no 16bit small pixel devices as they simply don't have the well capacities to achieve high dynamic ranges. Saturation levels are also reached quite quickly.
Optical scattering within the CCD/CMOS substrate also becomes more obvious as pixel size decreases.
Apogee instruments cover much of this here
http://www.ccd.com/ccd111.html