The critical role of matrix effect correction in oxygen isotope analysis of clinopyroxene by SIMS

Raw stable isotope ratios measured by secondary ion mass spectrometry (SIMS) can deviate from their true values due to instrumental mass fractionation (IMF), a phenomenon known as the matrix effect. Accurate determination and prompt correction of IMF during SIMS measurements are essential for obtaining reliable stable isotope data. In minerals and glasses, relationships between IMF values and element concentrations, along with associated endmember components, are commonly used to account for matrix effects that can reach a few tens of permille. Although this approach is relatively universal, it can introduce significant bias in IMF determinations, especially when the chemical compositions of reference materials differ from those of the analyzed samples. The primary objective of this study is to evaluate the impact of diversities in clinopyroxene chemical composition in assessing matrix effects during oxygen isotope analysis. Our results demonstrate that clinopyroxene Mg-number is usually not a reliable proxy for determining matrix-related IMF. Instead, Ca content or the amount of wollastonite component provide reasonably accurate means of accounting for matrix effects. However, multivariable linear regression models can markedly improve matrix-dependent IMF prediction compared to single-variable approaches, particularly when the composition of unknown clinopyroxene samples deviates from that of the reference materials. We evaluate the strengths and limitations of the aforementioned IMF correction methods by examining oxygen isotopic compositions of coexisting olivine and clinopyroxene from the Midfell gabbro nodules, SW Iceland. • Although Ca or Wo contents in Cpx can be used as sole variables for IMF correction. • Multivariable linear regression models provide more accurate IMF predictions. • Clinopyroxene Mg# is not a reliable proxy for correcting O-isotope matrix effect. • Multivariable regression models were tested on natural Cpx samples.