V33D-07:
Gaining Precision and Accuracy on Microprobe Trace Element Analysis with the Multipoint Background Method

Wednesday, 17 December 2014: 3:10 PM
Julien M Allaz1, Michael L Williams2, Michael J Jercinovic2 and John J Donovan3, (1)University of Colorado at Boulder, Boulder, CO, United States, (2)University of Massachusetts, Amherst, MA, United States, (3)University of Oregon, Center for Advance Materials Characterization in ORegon (CAMCOR), Eugene, OR, United States
Abstract:
Electron microprobe trace element analysis is a significant challenge, but can provide critical data when high spatial resolution is required. Due to the low peak intensity, the accuracy and precision of such analyses relies critically on background measurements, and on the accuracy of any pertinent peak interference corrections. A linear regression between two points selected at appropriate off-peak positions is a classical approach for background characterization in microprobe analysis. However, this approach disallows an accurate assessment of background curvature (usually exponential). Moreover, if present, background interferences can dramatically affect the results if underestimated or ignored. The acquisition of a quantitative WDS scan over the spectral region of interest is still a valuable option to determine the background intensity and curvature from a fitted regression of background portions of the scan, but this technique retains an element of subjectivity as the analyst has to select areas in the scan, which appear to represent background.

We present here a new method, "Multi-Point Background" (MPB), that allows acquiring up to 24 off-peak background measurements from wavelength positions around the peaks. This method aims to improve the accuracy, precision, and objectivity of trace element analysis. The overall efficiency is amended because no systematic WDS scan needs to be acquired in order to check for the presence of possible background interferences. Moreover, the method is less subjective because "true" backgrounds are selected by the statistical exclusion of erroneous background measurements, reducing the need for analyst intervention. This idea originated from efforts to refine EPMA monazite U-Th-Pb dating, where it was recognised that background errors (peak interference or background curvature) could result in errors of several tens of million years on the calculated age. Results obtained on a CAMECA SX-100 "UltraChron" using monazite standards and well-characterized unknowns yield excellent agreement with ages obtained by isotopic method. The MPB method aims thus to improve and facilitate trace element analysis in general, as we further demonstrate with blank tests and Ti-in-quartz measurements in reference materials and well-characterized quartz samples.