Hi Michael,
I agree that if the composition you are scanning the beam over is homogeneous enough to make the non-linear matrix effects insignificant, then there would not be any point in averaging the intensities over a scanned area. Note however that matrix effects for oxygen are quite large in this matrix.
If I were tasked with this characterization I would probably acquire some rectangular x-ray maps across the interface and then fully quantify the pixels in CalcImage. Then perhaps use the "strip averaging" output method there.
If you want to utilize some special JEOL beam mode for "overscanning" in PFE, I would probably use the "Do Not Set Conditions During Acquisition" checkbox in Acquisition Options. But again, I would avoid these "defocus beam" or "overscanning" methods due to the reasons discussed in the topic linked to above.
john