Reference Materials in Automated Quantitative Mineralogy – experiences and approaches at the Freiberg Geometallurgy Laboratory

Quantitative mineralogy done by automated image acquisition, signal processing and analysis became the principle driving force behind geometallurgical research projects and industrial applications. The same analytical methodology is used in forensic examinations, as well as for petrological, mineralogical and archaeometrical studies.
Recent developments in instrumental techniques, new algorithms for image and signal processing and growing computing power form the basis for this enormous development. The level of automation, the reproducibility and 'superhuman' never-tiring endurance have made this methodology virtually indispensable.
As a consequence of this development, due to the growing economic impact and foreseeable individual consequences as the result of forensic studies, laboratories using automated quantitative mineralogical methods have to face more and more question relating to:

• the accuracy and trueness of measurement,
• measurement precision and measurement reproducibility, and
• metrological traceability.

The routine usage of reference materials (RM) is one of the cornerstones to meet such requirements.
The principle of measurement is the combined detection of back-scattered electron intensities and X-ray spectra, both depending basically on chemical properties of the material studied. Automated quantitative mineralogy is consequently no true (primary) phase analytical method (PAM) and has to be traced back to primary PAMs.
To identify the RMs needed the quantitative mineralogical measurement needs to be separated into distinct steps and tangible variables that influence the success of these steps. The main steps, excluding sampling, are:
1. sample preparation,
2. image acquisition,
3. X-ray spectra acquisition,
4. X-ray spectra processing,
5. image processing including stereological reconstruction of 3-D features,
6. calculation of derived data like mineral mode, particle size distribution, particle shapes,
degree of liberation or the resolution of intergrowth relationships.
In general, quality management for steps 2 and 3 is well established and described in standardized procedures and well covered with RMs. However, the lack of suitable RMs affects all the other steps significantly.
The largest contribution to the uncertainty budget is sample preparation. The preparation of “in-house” standards will improve the situation, but will not reach the effects of future RMs. RMs to assess the correct and reproducible processing of X-ray spectra and the stereological reconstruction are easier to define and fabricable, but very scarce.
The experience of the Freiberg Geometallurgy Laboratory with missing suitable RMs is illustrated and missing inter-laboratory comparability of results is identified as the most serious challenge. First approaches to a solution, focusing on the creation of “in-house” standards as a first step towards a broader approach involving other laboratories worldwide, are presented and critically evaluated.