Limitations of chemical dating of monazite

Abstract

Instrumental and spectral characteristics germane to chemical dating of monazite have been tested using the Cameca SX-100 at Rensselaer Polytechnic Institute. Statistical analysis demonstrates that, for trace element analysis, equal counting time on peak and background is required for optimal statistical precision, thus rendering impractical the procedure of fitting the entire spectrum to obtain background values. Energy shifts require shifting the detector voltage window between peak and background positions, and it is concluded that the differential auto PHA mode works optimally for this. Analyses of Pb-free phosphates, silicates, and oxides are used to measure spectral interferences with the PbM α peak and background positions. Backgrounds were modeled using both linear and exponential fits. It was found that the difference in background counts using the two fits varies with each of the five spectrometers examined, and that the high-pressure (3 bar) detectors show larger differences in exponential vs. linear peak-minus-background (P-B) values than the low-pressure (1 bar) detectors. In addition, every spectrometer requires a unique correction for every major element in monazite. An analytical protocol is presented that incorporates these results. This protocol was applied to several monazite standards to determine inter-spectrometer variability, and spectrometer reproducibility from session to session. It was found that the difference in composition (and age) between spectrometers on identical spots exceeds the 2 sigma standard error of the mean of composition (or age) on either spectrometer. This means that (a) additional sources of error beyond the counting statistics exist between spectrometers; (b) the precision of microprobe ages cannot be continuously improved by additional counting; and (c) the minimum realistic precision is on the order of ± 2–3% for monazites with around 1500–2000 ppm total Pb, or an additional absolute uncertainty of 20–50 ppm Pb.