The oxygen isotope composition of San Carlos olivine on the VSMOW2-SLAP2 scale

Rationale: The precise determination of Δ'¹⁷O values in terrestrial material is becoming increasingly important to understand the mass-dependent fractionation processes that cause variations in oxygen isotope ratios. San Carlos olivine is widely used as the reference material for oxygen isotope measurements of terrestrial and extraterrestrial materials. We report new Δ'¹⁷O values for San Carlos olivine that were independently determined in two different laboratories (Geoscience Center [GZG], University of Göttingen) and Institute for Study of the Earth's Interior [ISEI], Okayama University, Misasa) in direct comparison with VSMOW2 and SLAP2 water standards. Methods: The δ¹⁷O and δ¹⁸O values of VSMOW2, SLAP2, GISP, and San Carlos olivine were determined relative to reference gas. In both laboratories, water and San Carlos olivine samples were prepared by BrF₅ fluorination. In both laboratories, the O₂ released from water and olivine samples was passed through the same purification system and measured using the same mass spectrometer relative to the same reference gas. Results: In both laboratories, the δ¹⁷O_VSMOW2 and δ¹⁸O_VSMOW2 scales were slightly compressed with respect to the recommended composition of VSMOW2 and SLAP2. The new Δ'¹⁷O_{0 .528} value (calculated from the VSMOW2-SLAP2 scaled δ values) of San Carlos olivine from GZG was −36 ± 9 ppm and, from ISEI, a value of −40 ± 7 ppm (1σ standard deviation) was determined. These values are ~50 ppm higher than previously reported from the same laboratories. Possible causes for the observed discrepancies are discussed. Conclusions: The results of this study in comparison with previous data from the same laboratories demonstrated that for high accuracy determination of Δ'¹⁷O values: (i) calibration of the reference gas relative to O₂ released from primary standards (VSMOW2, SLAP2) in the same laboratory is highly recommended, (ii) non-linearity of the mass spectrometer may not only affect δ¹⁷O and δ¹⁸O values but also Δ'¹⁷O values, and (iii) the VSMOW2-SLAP2 scaling should also be applied to analyses of rocks and minerals. Studies that are concerned with small differences in Δ'¹⁷O at similar δ¹⁸O values, however, are not affected by non-linearity of the mass spectrometer.