ESTIMATION OF ANALYTICAL ERROR FOR MAJOR, MINOR AND TRACE ELEMENTS ANALYZED BY X-RAY FLUORESCENCE AT THE PETER HOOPER GEOANALYTICAL LABORATORY, WASHINGTON STATE UNIVERSITY

The preparation method used at the Peter Hooper GeoAnalytical Laboratory for the determination of 29 major and trace element concentrations by X-ray fluorescence (XRF) was originally developed ~50 years ago, and was initially used to distinguishing subtle chemical differences among lavas of the Columbia River Basalt Group. However, the increasing accuracy and precision of modern analytical instruments warrants reexamining the method and its associated precision. Sample preparation begins by reducing the samples to fine powders (~10 microns) using a hydraulic press, a steel jaw crusher, and shatterbox grinders with tungsten carbide, agate, or steel grinding bowls. Powders are mixed with Li2B4O7, using a 2:1 ratio (flux:sample), and fused at 1000°C in pure graphite crucibles. This is followed by a second grind and fusion to ensure sample homogeneity. The glass beads are finished by uniformly grinding the lower surface flat for analysis on a ThermoARL Advant’XP wavelength-dispersive XRF spectrometer.

For every batch of 10-23 samples, one sample is prepared in duplicate, providing an immediate check on analytical reproducibility. Comparing data from these duplicate samples can assess practical precision for the entire methodology, from hand sample through to analysis. Non-normalized data from 256 duplicate samples from a wide range of geologic materials, and spanning 3 instrumental re-calibration cycles, were used to calculate the average Relative Percent Difference (RPD) for each element, where RPD = (|Duplicate A – Duplicate B|) / ((Duplicate A + Duplicate B)/2)*100. Calculated RPD values are ~1% (Al, Si, K, Ca, Mn, Fe, Sr, and Zr), ~2% (Mg, P, Ti, Zn, and Ba), 3%-4% (V, Cu, Ga, Rb, and Y), 5%-7% (Sc, Cr, Ni, Nb, Ce, and Nd), and 8-10% (La, Pb, and Th). Average RPD generally increases with decreasing average concentration.