While TEM is your best bet, you may also be able to use polarized filters if your silica particle have any birefringence. While Allan makes good points, I would suggest that you avoid a cryo TEM and use normal epoxy resin embedding procedures and an ultramicrotome. Cryo TEM is great for sectioning tissue samples but not so much for individual cells. We use TEM and polarized filters to image our nanoparticles and get up to 2nm resolution.