Here is a link I found where someone else did this basic exercise. BTW they got some very nice, discrete peaks.
Obviously the alpha decay is not of interest here and is easily filtered out (physically). I cannot find a reference that indicates how often Am 241 gives off a gamma during the alpha decay? Some references seem to indicate that the gamma occurs for every event (1:1.)
The paper above reports detection of peaks related to Neptunium gammas as well. It will be interesting to see if and how you can distinguish the characteristic "appearance" of each metal with all the minor peaks and noise. On a basic level, assuming the hardware is adequate, you could just record each graph for a known material and they try to figure out what it most closely matches. This could even be done outside of the detector. Perhaps the detector sends the graph to a network resource and a program there does the matching. Maybe even web based? i.e I send my graph to your server and you tell me what I have. Just thinking out loud.
Robert
The paper shows a nice geometry for an XRF setup. I think it is pretty similar to at least one of the Theremino configurations and worth stealing, er, borrowing. The detector is a high-purity Germanium diode, cooled with LN2. Detectors like these have a large depletion region, created by "drifting" lithium into the semiconductor, and, once the lithium is where the manufacturer wants it the detector is cooled down to 77K. And for the rest of its useful life it MUST be kept at 77K. Not too practical for a hobbyist-level XRF.
At the lab where I used to work we had a lithium-drifted detector that was specified to have an energy resolution down in the low (single-digit) electron volt range. It was in a large Dewar that held about a gallon of LN2, and had a liquid level sensor in it that would alarm when the level got too low. We had to top it off every week. Rain or shine, Snowmageddon or not. That detector, more than anything else, was where the $$$ was in that system. Maybe I didn't add enough dollar signs but you get the drift (bad pun).
There are lots of "gun" style XRF tools offered for sale that just can't be using HPGe detectors so I know it must be possible. Unfortunately, the goal of separating all the peaks of metals that go into steel alloys, particularly manganese vs iron or iron vs chromium, is going to be challenging--IF all we are depending on is the resolution of our detector and S/W combination.
It might be possible to enhance the system resolution by using X-Ray energy filters. For an example, see
this paper. FYI, 1.5 Angstroms is 150nm, corresponding to about 8.6Kev. So the paper can't be applied as such. But....
Taking the X-ray filter notion a bit further, I made up a table that summarizes the x-ray lines and absorption edges for Vanadium through Nickel. I got the absorption information from
here, and the table is attached. The absorption edge is the energy at which the absorption abruptly decreases, so emission lines at an energy less than that will be attenuated more than those above it, by about an order of magnitude. So to distinguish manganese from iron we would need a filter whose ideal absorption edge lies between the emission lines of manganese and iron, 5.89Kev and 6.400Kev respectively. From the table, Chromium fits the bill. But if we want to tell the difference between iron and cobalt we might want to use a Manganese filter. And so on. Maybe I've just moved the challenge to finding thin metal foils of these metals. Some will be easier to find than others.
BTW, I don't think we will need to worry about x-ray fluorescence from the filter materials, as long as they are not exposed to the Americium gamma rays. The photon energy coming into the filter won't be much higher, if at all, than the filter's x-ray line so not a lot of extra energy available to excite fluorescence. I hope....
