Determination of rubidium and strontium in silicate rocks by energy dispersive and wavelength dispersive X-ray fluorescence analysis: a comparative evaluation of precision

Abstract

Levels of precision routinely attainable in the determination of Rb and Sr at trace abundances by energy dispersive X-ray fluorescence (EDXRF) analysis are reported and compared with results obtained by wavelength dispersive X-ray fluorescence (WDXRF) analysis. Surprisingly, the precision of EDXRF determinations is similar to that for WDXRF analysis at concentrations less than about 20 ppm. The favourable characteristics of EDXRF analysis at these low concentrations are attributed to the use of a primary beam filter which reduces backgrounds in the spectral region of these elements. At higher concentrations, the precision of WDXRF results becomes progressively better than EDXRF owing to the superior sensitivity of the technique. Above 50 ppm, the relative standard deviation of WDXRF results is less than 1%. The EDXRF technique achieves comparable levels of precision only above 100 ppm. The parallel acquisition of Rb and Sr data by the ED method appears to provide no advantage in minimizing errors in the Rb:Sr ratio for which the optimum precision is required by isotope geochemists for geochronology. The precision of WDXRF measurements can be further improved by longer or multiple counting, but this procedure would be inefficient in EDXRF analysis where routine count times are already long (about 1200 s). It is noted that representative sub-sampling may not always be achieved in rock samples routinely prepared for geochemical analysis. If the highest precision is required then it is preferable to count replicate samples rather than repeatedly counting the same sample. Any form of multiple counting would not be an efficient proposition by EDXRF.