When using ultrafiltration and centrifugation, you need to adjust the pore size of the membranes and the number of revs of the centrifuge, which is more difficult than studying the kinetics of particle settling. Choose what is easier for you to do in your lab.
if its okay that they can get wet, i would take a great measuring cylinder (2 or 3 L) add water and drop silicon in it. the coarse silicon is falling down and the nm scale one is floating. than you can decant it. just do it more than one time to be sure.