Most µXRF companies won’t have experience with putting them on an EPMA. My experience was with iXRF systems based out of Austin, Texas- the owner (Kenny Witherspoon, email me for his contact info) was very forthcoming with me regarding the lack of industry knowledge for operating a µXRF on an EPMA/SEM system and he was very eager to collaborate and try and develop new applications for their products.
I loved having the uXRF on the SEM in Cardiff. We had dual 150mm2 EDS detectors which were enabling us to resolve <10ppm concentrations for most elements Z>13 when using the µXRF. The only problem that we ran into was that there wasn’t much information in the literature regarding any experimental constraints on the actual penetration depths and interaction volumes. That being said, I found that these values could be modeled in the same way that we model electron-specimen interactions with relatively accurate results. I ran a series of tests before I left Cardiff on Ca in Olivine and Ti in quartz and the results were darn near spot on with various models that I had ran in CalcZAF.
The µXRF couples well with the EDS detectors on any instrument, but you need to make sure that you have room in your chamber for the µXRF to be adjusted for focusing the beam. It isn’t really all that well known how well the µXRF will couple with WDS detectors on an EPMA. We had a WDS on our SEM in Cardiff and my primary plan was to explore the possibilities of using the WDS and µXRF together with the E-beam for mixed analyses. I didn’t manage to get too far into that before I got the job offer to come to UT Arlington.
In regards to Jon's points:
"Mounted on a WDS port would mean that you would be looking at a tangental cross section of the uXRF beam (meaning an oval shape on your sample surface) rather than circular, and a Gaussian distribution of the photons across that." - This is true. We had our's mounted at a 32.5° takeoff angle which resulted in an oval-shaped spot roughly 10µm by 15 to 20 µm in size. We simply mapped out the location of the beam and overlayed the position and shape of the spot on our observation window.
"The Xray beam (coming in at the 'take off' angle to the sample surface rather than perpendicular to it) would travel not only through the grain you were looking at, but potentially also in to any neighbouring grains and possibly even in to the sample holder itself. The distance the beam travels would be dependent on the material that you were looking at and the energy of the primary Xray beam (ie whatever target you buy)" - Although this is true, this was not really a problem that we faced until we ran into a grain that was oriented perfectly to diffract our X-ray beam and cause a significant amount of secondary fluorescence. Occasionally we would get a little bit of an Al Ka signal fro the sample chamber, but most of the time we didn't see anything coming off of the sample holder or from surrounding grains aside from the expected secondary fluorescence.
"ZAF/PRZ would obviously be out of the window, but XRF quantification isn't exactly new so there must be an XRF-PRZ alternative(?)" - In the world of XRF they use a simple fundamental parameters (FP) method for quantification. It is relatively simple and is actually pretty good at standardless analysis. It is very easy to set up cal curves in conjunction with the FP method.
"Figuring out the geometric relationship between the point at which a characteristic X-ray was generated (remembering that the primary Xray beam would be travelling across as well as through the sample!) and the take off angle (and so extent of absorption by the sample) with respect to the position of the WDS crystal/detector would be... interesting!" - This was something that I was trying to work on before I left Cardiff, but I didn't get to make much headway on it.
Hopefully this helps. It has been a little while since I've really thought about any of this, but I can definitely dig up some of my data if anybody has any other questions.