How well do we know VPDB? Variability of delta13C and delta18O in CO2 generated from NBS19-calcite

In order to generate a local daughter scale from the material defining the international delta13C and delta18O stable isotope ratio scales (NBS19-calcite),1,2 the carbon and oxygen must be liberated to the gas phase, usually as CO2, using acid digestion of the calcite with H3PO4. It is during this conversion step that systematic errors can occur, giving rise to commonly observed discrepancies in isotopic measurements between different stable isotope laboratories. Scale consistency is of particular importance for air-CO2 isotope records where very small differences in isotopic composition have to be reliably compared between different laboratories and quantified over long time periods.3 The information is vital for estimating carbon budgets on regional and global scales and for understanding their variability under the conditions of climate change. Starting from this requirement a number of CO2 preparations from NBS19 were made at Environment Canada (EC) and analyzed in our laboratories together with Narcis II, a set of well-characterized CO2 samples in sealed tubes available from the National Institute for Environmental Studies (NIES).4,5 Narcis II is very homogeneous in delta13C and delta18O with the isotopic composition close to NBS19-CO2. Among our laboratories the results for delta13C agreed to within +/-0.004 per thousand. The same level of agreement in delta13C was obtained when CO2 was generated from NBS19-calcite using different experimental procedures and conditions in the other two laboratories. For delta18O, the corresponding data were +/-0.011 per thousand when using NBS19-CO2 produced at EC, but discrepancies were enhanced by almost one order of magnitude when NBS19-CO2 was prepared by the other laboratories using slightly different reaction conditions (range=0.13 per thousand).In a second series of experiments, larger amounts of CO2 prepared from NBS19 at the Max-Planck-Institut für Biogeochemie (MPI-BGC) were analyzed together with Narcis II and then mixed into CO2-free air. The resulting artificial air samples then were measured by the same three laboratories for the stable isotopic composition of CO2 using locally established extraction and evaluation procedures. Comparison of the results with the prior CO2 values and between the laboratories revealed additional systematic differences owing to the local CO2 extraction processes and standardization procedures. For delta13C the results showed a narrow range of discrepancies of about 0.02 per thousand; for delta18O cumulative disagreements in the range of 0.1 per thousand were observed. From these results the following conclusions are inferred: NBS19-CO2 is a reliable primary anchor to the VPDB delta13C scale. Although prepared by different methods an accuracy of better than +/-0.003 per thousand has been reached. This applies to sample amounts of 5 mg calcite or more.NBS19-CO2 can be used as a general anchor to the VPDB delta18O scale only for accuracy requirements of +/-0.1 per thousand. For a higher scale resolution additional agreements regarding details of the acid digestion reaction will have to be worked out and agreed upon.Narcis II-CO2 comprises an ideal set of test samples for the VPDB scale. The delta13C value is +1.923+/-0.003 per thousand (combined uncertainty); delta18O is between -2.50 and -2.65 per thousand versus VPDB-CO2, with most of the variation in this figure depending on details of the NBS19-CO2 preparation used for the calibration. (Ampoule to ampoule homogeneity is better than 0.01 per thousand.)When mixing NBS19-CO2 into artificial air and using this to test performance between laboratories, the delta13C offsets are small with a remaining discrepancy of only 0.02 per thousand. For delta18O, systematic disagreements are considerably larger than those found for the pure CO2 comparison. Further experimental clarification is required.Artificial air samples such as NBS19-CO2 in air can be used as reliable anchors to create a unified stable isotope scale between different laboratories. An adjustment of local scales based on these air standards appears to be necessary for improving data comparability.