Point light source
I begin to realize that, because the dots' luminosity or opacity is faint as compared with stray light, opticians will probably use many dots at a time.
That is, they would have an eyepiece and CCD that give many pixels within the objective's diffraction limit, collect the diffraction pattern around many dots in the field, and have a software compute the image's autocorrelation to obtain a clear objective's diffraction pattern from many noisy patterns around each dot.
Or more precisely than "autocorrelation", something like an autoadaptive filter should run on the image, for instance a Kalman filter, to obtain first the positions of the nanoparticles. Then, the correlation between the raw image and the filtered one gives the diffraction pattern, accumulated over many particles. Or maybe the proper programmation of the autoadaptive filter gives the diffraction pattern directly.
For that purpose, they would prefer the dots on an irregular pattern, and I'd say, the distance between the dots should better be greater than one diffraction pattern diameter, which must be seriously bigger than 20nm.
Is a mist of hydrofluoric acid, with droplets 50nm wide, any conceivable? They might etch tiny holes in glass.
I hope a single tiny hole illuminated by the very concentrated light of a DVD burner laser diode and its lens does not need this software data processing.
Marc Schaefer, aka Enthalpy