MONAZITE DATING ON THE ELECTRON MICROPROBE: INSTRUMENT SYSTEMATICS

We have undertaken a systematic study of the instrument, spectral, and statistical characteristics of Pb (PbMß)analyses on the JEOL 733 Superprobe at RPI. The instrument is equipped with 4 PET crystals: Specs 2 and 3 have flow detectors (P10 gas); Specs 4 and 5 have sealed Xenon detectors. Spectrometer geometry is 140mm Rowland circle and backgrounds were collected at +3.45 and –2.4 mm on each. Materials analyzed include a monazite from the Trebilcock pegmatite quarry, Maine, and a synthetic, Pb-free CePO₄ crystal.

Systematic differences in the X-ray spectrum around PbMß exist between the flow and Xenon counters. Background around PbMß in the Xenon counters is relatively flat with no obviously interfering peaks. The flow counters show a peak at +1.55 mm (relative to PbMß) that is believed to be Cela2, which is not completely removed by the PHA. The difference between the flow and sealed counters is due to the different energy responses of the detector gases.

Precision on each spectrometer is consistent with counting statistics and Monte Carlo simulations (±20 ppm 1s s.e. with care). However, differences exist between the four spectrometers. The two Xenon counters give similar results on the standard (1318 ±74 and 1303 ±86 ppm; n=20; 1s s.d.). For the same spots, the flow counters give higher values (1698 ±176 and 1525 ±45 on Specs 2 and 3, respectively). Pb-free CePO₄ "blank" yields an apparent Pb concentration of 423±129, 169±40, 53±56, and 49±55 ppm for Specs 2, 3, 4, and 5, respectively. The observation of significant apparent Pb in a Pb-free CePO₄ likely results from interference due to the Cela2 peak observed with the flow counters (Specs 2 and 3). This "apparent Pb" represents 11-25% of the total Pb in the standard, and significantly reduces the accuracy of age determinations. For example, the difference in Pb counts between spectrometers can propagate to an age uncertainty of over 100 Ma in Acadian monazites.

Analyses of natural monazites using Specs 3, 4, and 5 show more scatter than the Trebilcock standard. Most disturbing is the occasional result that the identical Specs 4 and 5 yield average counts that differ by more than counting statistics. Periodic analysis of the secondary standard rules out the possibility of spectrometer drift, and the cause is unknown.