How precise and how accurate can magnesium isotope analysis be by the sample-standard bracketing method?

Abstract

Mg isotope compositions are widely analyzed using the sample-standard bracketing (SSB) method on MC-ICP-MS. However, it is not clear how accurate and how precise the Mg isotope data could be by using the SSB method on MC-ICP-MS. In this study, we conducted doping experiments to artificially produce a ∼0.03‰ δ26Mg offset to quantitatively evaluate the utility of 2SD, 2SE, and tSE (t is the Student’T) error in identifying subtle isotope variations. The 2SD represents the most conservative estimation for accuracy. For samples with a large number of repeated measurements, the dataset will follow a normal distribution, and accuracy at a 95% confidence level can be measured by ±2SE. For samples with a small number (<9) of repeated measurements, the dataset will follow a t-distribution, and the accuracy at a 95% confidence level can be measured by ± tSE. The results of a series of doping experiments show that the two standard deviations (2SD) of δ26Mg is 0.04‰, which is larger than the artificial 0.03‰ offset. The two-standard error (2SE) and the tSE are both smaller than 0.02‰ when measured by six times, which can both resolve the 0.03‰ offset. Given that Mg isotope analysis is routinely repeated about 3∼5 times in most laboratories, tSE is more robust than 2SE and 2SD in identifying subtle Mg isotope variation. Although we highlight the role of statistics in Mg isotope analysis, the readers should notice that chemical treatment and instrumentation settings are first-order factors that control data quality.