Reformulated 17O correction of mass spectrometric stable isotope measurements in carbon dioxide and a critical appraisal of historic 'absolute' carbon and oxygen isotope ratios

Mass-spectrometric stable isotope measurements of CO2 use molecular ion currents at mass-to-charge ratios m/z 44, 45 and 46 to derive the elemental isotope ratios n(13C)/n(12C) and n(18O)/n(16O), abbreviated 13C/12C and 18O/16O, relative to a reference. The ion currents have to be corrected for the contribution of 17O-bearing isotopologues, the so-called ‘17O correction’. The magnitude of this correction depends on the calibrated isotope ratios of the reference. Isotope ratio calibrations are difficult and are therefore a matter of debate. Here, I provide a comprehensive evaluation of the existing 13C/12C (13R), 17O/16O (17R) and 18O/16O (18R) calibrations of the reference material Vienna Standard Mean Ocean Water (VSMOW) and CO2 generated from the reference material Vienna Pee Dee Belemnite (VPDB) by reaction with 100% H3PO4 at 25 °C (VPDB-CO2). I find , 18RVSMOW/10-6 = 2005.20 ± 0.45, 13RVPDB-CO2/10-6= 11124 ± 45, and 18RVPDB-CO2/10-6=2088.37±0.90. I also rephrase the calculation scheme for the 17O correction completely in terms of relative isotope ratio differences (d values). This reveals that only ratios of isotope ratios (namely, 17R/13R and 13R17R/18R) are required for the 17O correction. These can be, and have been, measured on conventional stable isotope mass spectrometers. I then show that the remaining error for these ratios of isotope ratios can lead to significant uncertainty in the derived relative 13C/12C difference, but not for18O/16O. Even though inter-laboratory differences can be corrected for by a common ‘ratio assumption set’ and/or normalisation, the ultimate accuracy of the 17O correction is hereby limited. Errors of similar magnitude can be introduced by the assumed mass-dependent relationship between 17O/16O and 18O/16O isotope ratios. For highest accuracy in the 13C/12C ratio, independent triple oxygen isotope measurements are required. Finally, I propose an experiment that allows direct measurement of 13R17R/18R