This key comparison aims to assess the capabilities of the participants to determine the amount-of-substance fraction oxygen in nitrogen. The GAWG has classified this as a track B comparison, due to the unexpected 50 μmol/mol argon mole fraction content of the transfer standards, which effects the achievable performance of some measurement techniques such a GC-TCD. The separation of oxygen and argon is challenging, and not all systems in use are equally well designed for it. As this analytical challenge due to a substantial fraction of argon in the transfer standards became a reality, the Gas Analysis Working Group (GAWG) decided to qualify this key comparison as a regular key comparison and not as a core comparison, which may be used to support calibration and measurement capabilities (CMCs) for oxygen in nitrogen, or for oxygen in nitrogen mixtures containing argon only (see also the section on support to CMCs).
This project concerns the purity analysis of nitrogen as used in reference gas mixture preparation. This project was carried out without adding impurities to the gas used for this comparison, and is therefore more representative to evaluate the analysis of CO, CO2, CH4, O2, Ar and H2O impurities in high purity nitrogen. The analysis of the amount–of–substance fraction water was optional.
Two 50 litre high purity nitrogen cylinders were purchased from a well-qualified supplier of specialty gases. The listed components were expected to be present in the pure nitrogen at the target levels as a result of the purification of the nitrogen. From the start of this comparison it was clear that the comparison may not lead to reference values for the constituents analysed.
The results indicate that analyses of high purity gases are often limited by the limits of detection of analytical equipment used. The reports of the participating laboratories also indicate that there is no agreed method of determination of the uncertainty on a detection Limit value. The results provide useful information on the Performance of participants. For all analysed components there is reasonable agreement in results for LNE, VSL, Metas and NPL.
For BAM only the Argon result is in agreement.