The beam halo rate (taken from g0rate3) did not seem to vary much with beam position on target, fig. 1, but it did vary with the beam raster size (and probably with the beam current), fig. 2.
Dropping the raster size from 4 mm to 3.5 mm, dropped the halo rate by ~45%, while the beam current was dropped only 17%. Dropping the raster from 3.5 mm to 3 mm, dropped the halo rate by a further ~50%, while the beam current was dropped by 14%.
Figure 1
Figure 2