On the long-term stability of reference gases for atmospheric O2/N2 and CO2 measurements

Measurements of changes in the atmospheric O₂/N₂ ratio have typically relied on compressed air derived from highpressure tanks as the reference material against which atmospheric changes are assessed. The validity of this procedure is examined here in the context of the history of 18 O₂/N₂ reference tanks compared over a 1₂-yr time-frame. By considering differences in tank sizes, material types, and by performing additional tests, the long-term stability of the delivered gas is evaluated with respect to surface reactions, leakage, regulator effects, and thermal diffusion and gravimetric fractionation. Results are also reported for the stability of CO₂ in these tanks. The results emphasize the importance of orienting tanks horizontally within a thermally insulated enclosure to reduce thermal and gravimetric fractionation of both O₂/N₂ and CO₂ concentrations, and they emphasize the importance of avoiding elastomeric Orings at the head-valve base. With the procedures documented here, the long-term drift in O₂/N₂ appears to be zero to within approximately ±0.4 per meg yr^-1, which projects to an uncertainty of ±0.16 Pg C yr^-1 (1ó) in O₂-based global carbon budgets.