Hello,
We are currently working on apatite 87Sr/86Sr measurements via LA-MC-ICP-MS using the Sr Isotopes Universal DRS (Mulder et al. 2023) and have some questions regarding the error propagation within this DRS.
1. Rb mass-bias correction & Propagated uncertainty
We osberved that “StdCorrRb_Sr87_86_2SE(prop) (propagated uncertainty after Rb correction)” is occasionally smaller than “StdCorr_Sr87_86_2SE(prop) (propagated uncertainty before Rb correction)” relative to their 87Sr/86Sr ratios.
For example, BCR-2G was used for Rb mass-bias factor (Rb Beta) calculation and Durango apatite as the primary reference material (RM) for final normalisation. When comparing the propagated uncertainties before and after Rb correction, we observed inconsistent behavior among six replicate analyses of a secondary RM in the same session:
For three replicates, "“StdCorrRb_Sr87_86_2SE(prop)" increased relative to its pre-correction one, whereas for the other three, it actually decreased.
After reviewing the DRS, we could notice that “StdCorrRb_Sr87_86_2SE(prop)” is calculated through the “data.propagateErrors” function as below:
if propErrors:
drs.message("Propagating errors...")
drs.progress(90)
groups = [s for s in data.selectionGroupList() if s.type != data.Baseline]
data.propagateErrors(groups, [data.timeSeries("StdCorrRb_Sr87_86")], data.timeSeries("StdCorr_Sr87_86"), Sr_rmName)
If I understand well from this, "StdCorr_Sr87_86_2SE(prop)" include internal 2SE and propagate the excess uncertainty derived from the Primary RM (Durango in this case).
Does this propagated error calculate solely the excess uncertainty based on the primary RM’s final corrected values after the Rb correction has been applied?
Or does this propagated error integrate the excess uncertainty derived from the Rb-correction RM (BCR-2G)?
Also, we would like to understand the statistical factors within the DRS that lead to this inconsistent response in propagated error across replicates in the same session.
2. Error propagation for CaPO correction
By evaluating the CaPO correction using various apatite RMs as CaPO-correction RMs, we observed that “CaPOCorr_Sr8786_2SE(int)” is provided, but a propagated version “2SE(prop)” is absent. Reviewing the DRS script, “CaPOCorr_Sr8786” is not included in the “propagateErrors” as shown above.
If we need to report the total uncertainty for CaPO-corrected values, should we manually propagate the model uncertainty of “CaPO_corrAmt_2SE(int)” in quadrature with the “CaPOCorr_Sr87_86_2SE(Prop)”, or is there a recommended way to modify the DRS to include this?
I'd be really grateful if anyone could help. Thank you in advance.