Although some of the modern FEG-SEMs promise a resolution close to 1 nm, due to the creation of the secondary and back-scattered-electrons, you always have an interaction volume of radius 20-30 nm, while this is not a problem for imaging, for compositional analysis at the resolution you want, it is not practical.
Altough you can reach 2-3nm resolution for imaging in modern FEG-SEM, it is physically impossible to analyse elemental composition in 2-3nm range/resolution in SEM. The reason for this are the beam penetration depth and the interaction volume. Best case scenario the lateral resolution in SEM-EDX is around 100 nm.
For 2-3 nm lateral resolution, I would advise you to do TEM-EDX ou EELS. For 2-3 nm depth resolution, you can use a XPS.
You can use for SEM the same specimens (ultrathin) as for TEM and get about the same resolution for EDS. EDS resolution is determined mostly by geometry of specimen, not by microscope used (TEM or SEM)