Skip to main content

Quantifying method-specific biases in Measurements of Geomaterials: Insights from the GeoReM Database

Analytical methods for inorganic chemistry are found to be differentially effective for specific combinations of target elements, sample matrices and concentration ranges. Consequently, it is a significant challenge to ensure that the results provided in element concentrations are independent of the analytical method employed on the sample matrix. In order to address this issue, reference materials are produced and certified. However, even with the most rigorous production and certification procedures, it remains challenging to accurately determine the "true" concentration values due to the combined effects of measurement protocols and sample matrix. In particular, when dealing with "normal" (non-reference) samples, these effects may introduce a bias into the measurements. When compiling measurements from different analytical approaches and time periods into consistent data sets for further data analysis these method induced biases present a significant challenge. To explore and quantify the extent of methods/matrix-specific offsets between measurement batches, we exploited chemical analysis data from the database for reference materials, GeoReM. The selected data sets from GeoREM of the standards AGV-1&2, BCR-1&2, BHVO-1&2, BIR-1, BRP-1, JA-2, JB-1, OU-6 and W-1&2 initially comprised 83,158 measurements. Following homogenisation and filtering of the samples to ensure robust statistical analysis, 55,413 measurements remained. The statistical analysis (ANOVA) shows that the most frequently occurring offsets between methods are LA-ICP-MS to ICP-MS, followed by XRF to ICP-MS and finally XRF to LA-ICP-MS.

Details

Author
Solveig Pospiech1, Jens Walter2, Alexander Wellhäuser3
Institutionen
1HZDR, Germany; 2Georg-August-Universiät Göttingen; 3Alumnus der Georg-August-Universiät Göttingen
Veranstaltung
GeoSaxonia 2024
Datum
2024
DOI
10.48380/rvzv-5r93