QUANTIFYING REPLICATION THROUGH REPEATED ANALYSIS OF UVM-A, A LIQUID REFERENCE MATERIAL FOR COSMOGENIC 10Be AND 26Al STUDIES

In situ produced cosmogenic nuclide sample preparation and analysis is a multi-step process with numerous possible sources of error, few of which have been robustly quantified. Here, we use a liquid reference material (UVM-A) to explore the analytic precision associated with the isotopic ratios of 96 samples prepared in a single Be/Al purification laboratory by a large number of different users, measured at two Accelerator Mass Spectrometry (AMS) facilities over five years.

Freely available to the community, UVM-A contains ~250 ppm Be and ~1000 ppm Al in 1% nitric acid. Created at University of Vermont in 2018 by dissolving the CRONUS-A reference material, it is spiked with both the rare and the common isotopes such that the ¹⁰Be/⁹Be and ²⁶Al/²⁷Al ratios are set. Cosmogenic nuclide sample preparation laboratories can process UVM-A as a means of quality control. UVM-A is currently being used in 19 sample preparation laboratories in 5 countries; thus far, we have focused on studying the data from the University of Vermont preparation laboratory.

Based on this dataset, we suggest preliminary consensus values of 1.45 ± 0.06 x 10^-13 for ¹⁰Be/⁹Be (mean, one standard deviation, n = 96) and 4.47 ± 0.22 x 10^-13 for ²⁶Al/²⁷Al (mean, one standard deviation, n = 27). We find that ¹⁰Be/⁹Be ratios measured at the different AMS facilities (n = 26, Lawrence Livermore National Laboratory; and n = 70, Purdue Rare Isotope Measurement Laboratory) have statistically indistinguishable central tendencies, matching to within 0.1%. For both ¹⁰Be/⁹Be and ²⁶Al/²⁷Al, the relative standard deviations of the measured ratios are similar to the reported analytic uncertainties of sample measurement, suggesting that sample preparation introduced little additional scatter beyond the uncertainty of isotopic analysis.

This dataset demonstrates that robust sample preparation and measurement can generate cosmogenic isotope ratio data reproducible at the level of counting statistics. Future work will focus on comparisons between preparation laboratories and between a larger number of AMS facilities.