Hello Josh,
Here at the UO, we've been using a FIB to prepare samples for EPMA with great success. For us, the motivating reason was microcrystals of SiO2 intimately mixed with even smaller crystals of TiO2. Since we were interested in measuring Ti in the quartz lattice, secondary fluorescence was a huge issue - one *readily solved by using the FIB to either moat out regions around crystals or by extracting them as one would initially for TEM sample prep. Simply blast away the offensive phase. We did not thin the samples to the thickness desired for TEM, as that would not work for EPMA.
Are you wanting a workflow that looks like FIB --> EPMA --> TEM ? I think there would need to be a second FIB step between EPMA and TEM where you thinned the sample. Also, if your samples are susceptible to beam damage, thinning would be a way to get rid of the beam damage before going to the TEM ( I found this out the hard way by looking at a quartz crystal that had undergone beam-induced amorphization during EPMA - which actually looked kind of cool).
Overall, I think that there is a lot of potential to be had in the FIB-EPMA-TEM combo , though there is a lot of trial and error to be had before it reaches full potential.
Best,
Marisa