A system for high-quality CO2 isotope analyses of air samples collected by the CARIBIC Airbus A340-600

In 2007, JRC-IRMM began a series of atmospheric CO2 isotope measurements, with the focus on understanding instrumental effects, corrections as well as metrological aspects. The calibration approach at JRC-IRMM is based on use of a plain CO2 sample (working reference CO2) as a calibration carrier and CO2-air mixtures (in high-pressure cylinders) to determine the method-related correction under actual analytical conditions (another calibration carrier, in the same form as the samples). Although this approach differs from that in other laboratories, it does give a direct link to the primary reference NBS-19-CO2. It also helps to investigate the magnitude and nature for each of the instrumental corrections and allows for the quantification of the uncertainty introduced. Critical tests were focused on the instrumental corrections. It was confirmed that the use of non-symmetrical capillary crimping (an approach used here to deal with small samples) systematically modifies delta13C(CO2) and delta18O(CO2), with a clear dependence on the amount of extracted CO2. However, the calibration of CO2-air mixtures required the use of the symmetrical dual-inlet mode. As a proof of our approach, we found that delta13C(CO2) on extracts from mixtures agreed (within 0.010 per thousand) with values obtained from the 'mother' CO2 used for the mixtures. It was further found that very low levels of hydrocarbons in the pumping systems and the isotope ratio mass spectrometry (IRMS) instrument itself were critical. The m/z 46 values (consequently the calculated delta18O(CO2) values) are affected by several other effects with traces of air co-trapped with frozen CO2 being the most critical. A careful cryo-distillation of the extracted CO2 is recommended. After extensive testing, optimisation, and routine automated use, the system was found to give precise data on air samples that can be traced with confidence to the primary standards. The typical total combined uncertainty in delta13C(CO2) and delta18O(CO2) on the VPDB-CO2 scale, estimated on runs of CO2-air mixtures, is +/-0.040 per thousand and 0.060 per thousand (2-sigma values). Inter-comparison with MPI-BGC resulted in a scale discrepancy of a similar magnitude. Although the reason(s) for this discrepancy still need to be understood, this basically confirms the approach of using specifically prepared CO2-air mixtures as a calibration carrier, in order to achieve scale unification among laboratories. As important practical application and as a critical test, JRC-IRMM took part in the passenger aircraft-based global monitoring project CARIBIC (http://www.caribic-atmospheric.com). In this way, reliable CO2 isotope data for the tropopause region and the free troposphere were obtained. From June 2007 to January 2009, approximately 500 CARIBIC air samples have been analysed. Some flights demonstrated a compact correlation of both delta13C(CO2) and delta18O(CO2) with respect to CO2 concentration, demonstrating mixing of tropospheric and stratospheric air masses. These excellent correlations provide an independent, realistic data quality check.