The GLITTER software package was used for raw data reduction, applying initial calibration with NIST-SRM612 as the primary standard, using CaO g/100g for the internal count to mass fraction reference. Potential errors from surface contamination have been avoided by excluding the first seconds of data acquisition from integration. The signal trace over time for each element (m/z) measured was checked for its analytical and geological reasoning and excluded in case of abnormal pattern or spikes (e.g., vertical transitions into different mineral phases during ablation, mineral inclusions, sulfide micro nuggets in clinopyroxene). A second, matrix-matched calibration step was realized by correcting against mean offset factors of median values from at least three repetitive measurements on nano-particulate pressed powder tablets of reference materials USGS-BHVO-2NP, USGS-BIR-1NP, and GSJ-JGb-1NP. All reference materials were analyzed multiple times throughout each analytical session for bracketing and drift correction. Some sessions have been calibrated against the USGS glass microanalytical reference materials USGS-BHVO-2G and USGS-BIR-1G. Results of glass vs. pellet reference materials show no significant quality difference and lead to similar correction factors. Analytical measurement uncertainties were found to be in the range of 1% to 3% RMAD (1MAD) at mass fractions >10x LOD for most elements analyzed. For monitoring deviations from preferred or “true” values, the MPI-DING glasses GOR-128G, GOR-132G, and StHs6/80-G have been measured as unknowns. Additionally, JF-1P with extremely low mass fractions for most trace elements has been measured during most analytical sessions. Estimated deviations from preferred values of all analyzed reference materials are, for most elements, in the range of 5 to 10 RMSD% and within the uncertainties of the preferred or compiled values.
