Other than a different periphery-area ratio probably no big
deal - but much of the art of (low noise, high sensitivity)
imager pixel design is in the periphery, bending junctions
away from surface traps and so on. You would be maybe
36 times as sensitive to -that- as a result of downscaling
the pixel X/Y dimension. But how much it matters will
depend a lot on the technology and the application
environment.
300um is not large, really. Check out Hamamatsu et al,
folks who sell single photodiodes. Pics may give you
some ideas about art.