To find out about current spectrometer sampling speeds, research the term: "spectra per second"
There is alot of precis informations i.e.:
"The standard interface supplied with all nTM detectors is
12 bit USB2.0. With this interface and our SpectraArray SL software,
up to 30 spectra per second can be captured.
For even faster acquisition, a PCI bus interface is available as an
option. This allows up to 160 spectra/second to be acquired. "
50 to 80 micrograms of mineral is sufficient for a typical analysis, although small samples on the order of 12-20 micrograms (single foraminifers) can also be analyzed
So it seems that a major challenge is the data bandwidth between the spectrometer and the SSD/HDD.
Spectra per second returned by the machine are not going to be the limiting factor, because they can go up to 10 000 spectra per second. A graph with 250 floating point humbers is 4bytes times 250, so about 1kb per graph, research is necessary to understand why USB can only handle 12 samples per second, it should be in the 1000ds.
In spectroscopy, the key is to constantly calibrate the sample against an empty control sample. If the sample is drilled, it can be ground into a find dust and then it should be possible to process it in a homogenous way similar to liquid or gas spectroscopy.
All the different spectroscopy methods can be considered, they are all very fast, and have different detection strengths. IR and Xray have different advantages, IR is easier to deal with. The selection of appropriate IR or other wavelengths to use for a specific type of study is an advanced topic.