Optimised Methodology for the Determination of Total Sulphur and δ34S in Geological Materials by EA-IRMS

Sulphur concentration and isotopic composition (δ34SVCDT) are key tracers in geological, biogeochemical, and ecological studies, but achieving high precision by elemental analyser–isotope ratio mass spectrometry (EA-IRMS) remains challenging due to... Sulphur concentration and isotopic composition (δ34SVCDT) are key tracers in geological, biogeochemical, and ecological studies, but achieving high precision by elemental analyser–isotope ratio mass spectrometry (EA-IRMS) remains challenging due to SO2 adsorption and associated memory effects. Here, we present a systematic and integrated optimisation of EA-IRMS methodology for high-precision δ34S analysis of geological materials. A sulphur-only analytical mode was implemented to overcome limitations of conventional C/N/S analysis and to enable evaluation of key controlling factors, including SO2 signal height and stability, sulphur loading, memory effects, calibration strategy, combustion efficiency and matrix-matched calibration. Optimised conditions: 70–115 µg sulphur loading, ~5 V reference SO2 signal, ~50% dilution of sample-derived SO2 with He, inclusion of a blank between successive analyses, and two-point calibration yield a reproducible precision of ±0.3‰. Combustion temperatures >950 °C are required for optimal sulphur oxidation. Sulphur memory effects are more pronounced in geological samples than in synthetic standards and increase with progressive exhaustion of reactor fillings. Matrix-dependent variations in δ34SVCDT reproducibility (±0.24–0.47‰ sedimentary; ±0.17–0.31‰ magmatic; ±0.08–0.12‰ hydrothermal) highlight the necessity of matrix-matched calibration. Furthermore, δ34SVCDT values for five USGS magmatic and two GSJ hydrothermal reference materials are reported for the first time, providing new matrix-matched calibration benchmarks. This study establishes a robust analytical methodology for EA-IRMS that enhances the accuracy, precision, and inter-sample comparability of δ34S measurements in complex geological matrices.