Doug,
If the objective of using the glass microspheres is to add buoyancy, then using the microspheres by themselves without a binder could be problematic over repeated cycles of compression and decompression due to breakage of the spheres. I would not use the microspheres without a binder.
As a point of reference, on my boat, I laid up a hydraulically clean fiberglass shell over my pressure hull. To add buoyancy and to structurally support the fiberglass shell, I cast syntactic foam in the annulus space between the pressure hull and fiberglass shell. After discussions with others that cast syntactic foam, I used a low viscosity two-part epoxy that has a slow kick. This made it easier to pour after mixing with the microsphres and since I was casting thick sections, minimized the chance of cracking due to the exothermic reaction. The volume of the syntactic foam I cast was 15.5 ft^3.
One other point, to get a low density (26 lb/ft^3), I had to use both microspheres and macrospheres. I experimented with different ratios of these three components until I got the density I was looking for. The ratio I ended up with was 19.3%, 55.0% and 25.7% by volume of microspheres, macrospheres and catalyzed resin, respectively.
The epoxy resin I used was Huntsman Araldite GY-9667 with Huntsman Jeffamine D-230 Polyoxypropylenediamine as the catalyst. I used 3M K1 microspheres and Cumings Corp macrospheres ( 1/4-3/8" diameter 12-15 lbm/ft^3 true density). The macrospheres were not off the self, they were fabricated to meet my 300 fsw design depth and have a crush depth of 1000 ft. The only down side to making this syntactic foam was the cost which ended up being $157 per cubic foot, ouch!