I'm not sure it is possible to transfer from SiO2/Si, the graphene is very strongly adhered and would likely need to have the SiO2 layer etched from beneath the graphene to release it without damage. Most transfers to TEM grids are done from the PMMA supported graphene suspended on water. The Park and Muller groups at Cornell have a number of papers on TEM of CVD graphene.
This depends on the produced graphene and usually graphene is transported from other subsrates to SiO2 substrate. I think the adhesion of graphen to SiO2 is relativly high and It will be very difficut to transport it from SiO2.
Start with CVD graphene grown on Cu foil, instead of from SiO2 substrate, using standard PMMA mediated recipe. Handling of PMMA- graphene membrane is the main key for higher yield, which you will learn by practice.
If your graphene is highly adhesive to SiO2 (and okay with FIB for cutting the sample), I think you can look into TEM sample preparation using FIB/SEM (e.g: http://goo.gl/1Of76J).