OK, maybe we can start to think about increasing the efficiency and power of minor and trace element EPMA by expanding the capability of the multipoint background.
Here is an example we deal with a lot in geochronology applications - The measurement of U in the presence of Th in monazite, xenotime, etc. is tricky business (lets assume we are not going to use Uranium La at high voltage). U Ma is severely overlapped by Th Mb, and is further compromised by the Ar K edge if P10 is used. U Mb is generally preferred due to less spectral interference and actually greater intensity than UMa in most circumstances (see attached figure).
Without getting into absorption edge issues in very high Th compositions, the remaining problem is the boundary fluorescence of potassium near interfaces with K-feldspar or micas. The first-order interference of K with UMb is minor, but in the trace element realm, it's a problem, causing "too many" apparent U counts and therefore a slightly younger age at the rim (hooray, we just discovered a new event!). So this is dealt with nicely as a mutual interference problem between UMb and KKa. So now it is interesting because the best way to characterize this complex background region is via the multipoint acquisition. Works fantastically well (see second attachment), but here is the problem, we measure all these background points for the U measurement, and characterize the background in this spectral region very accurately. If we want to do a K measurement the same way on the same spectrometer, we would have to re-collect this multipoint acquisition. So why not use that regressed background data (from the U measurement) for the K measurement on the same spectrometer also - just move the peak position from U to K? Why do an independent background acquisition for K - we already have the data! In fact, if I wanted too, I could use the ThMg line with this same background measurement, just at a different wavelength for the analytical line? This could be very powerfully applied in other minor element situations where measured lines are in the same spectral region (REEs?).
It seems odd to me when we measure the full suite of rare earths to measure 15 analytical lines that we have to do 30 background acquisitions in the classical two-point interpolation method (two backgrounds per element), when we could probably characterize broad ranges of background very accurately with less than half that actual number of background measurements with a multipoint approach.