Cavanna - some thoughts that may help. In my last position I ran a Mechanical testing lab. Our tensile and Rockwell testing equipment was calibrated twice a year by an outside vendor. We verified the equipment at the beginning of each turn prior to use. For the Rockwell testers we used certified test blocks for the scale we were testing on that were provided by an external vendor, and simply recorded results in a log. If we needed to switch Rockwell hardness scales during a turn, we'd reverify after making the appropriate changes to the tester. For the tensile test equipment we used internally developed standards and SPC charts. We would pull 2 tensile specimens and plot average and range for ultimate tensile strength, 0.2% offset yield strength, and % elongation over a 2" gauge length.
Our chemistry laboratory also used spc charts and standards to verify prior to the beginning of the turn, depending on the grades of steel being produced. (We produced electrical steels, PH grades of stainless steel, 300 series grades of stainless, and 400 series of stainless.) Some of the standards used were purchased and traceable to NIST (The US government body responsible for standards, measurements, etc.) Others had been developed through interlaboratory cooperative testing arrangements. (One lab would take it upon itself to produce a set of samples that were sent to other producing using labs, and they would do multiple rounds of analyses until a consensus agreement was reached regarding the level of various elements in the sample. Then all labs would use the new standard as part of their verification/calibration activities.) For carbon and sulfur, we typically used special analyzers developed by Leco that were more accurate than a spectrometer in the ranges we measured. Standards for verification of Leco instruments were purchased from Leco.
For your spectrometer, I'd probably be interested in seeing accuracy versus analytical ranges by element. For example: iron based alloys, Cr range of 0.40 % to 18.00 % with an accuracy of+/- 10% of reading over the entire range, or whatever might be appropriate if accuracy varies over the range of alloys you produce.
To help quantify lab competency, our chem lab took part on a regular basis in interlaboratory comparison testing that was managed by an external company, whose whole business was managing and reporting interlaboratory tests for many different types of labs. I used the same service for both tensile and B & C scale rockwell hardness testing. Hope the info helps some.