Primary measurement standards for trace moisture in multiple gases: extension of gas species to He and O2

Abstract

The National Metrology Institute of Japan developed a multi-gas trace-moisture generator which was designed for generating trace water vapor (trace moisture) in multiple gases. Using this system, we previously developed primary measurement standards for trace moisture in N2 (10 nmol mol−1–5 μmol mol−1) and Ar (10 nmol mol−1–1 μmol mol−1). In this study, we have extended the target gas species; trace-moisture standards in He and O2 in the rage of 10 nmol mol−1 − 1 μmol mol−1 were developed. The relative expanded uncertainty (k = 2) of the trace-moisture standard in He was 0.74 %–5.7 %, and that in O2 was 0.78 %–8.9 %. In addition, we have improved stability of the flow rate of the diluent gas by controlling the temperature around the flow measurement/control systems and re-evaluated the uncertainty of the trace-moisture standards in N2 and Ar. The updated relative expanded uncertainty (k = 2) of the trace-moisture standard in N2 and Ar is 0.41 %–2.6 % and 0.73 %–2.6 %, respectively. The uncertainty budgets of the standards in these four gases are presented. The trace-moisture standard in N2 developed using the multi-gas trace-moisture generator was confirmed to be consistent with the standards realized by other generators, whose international comparability were already ensured. A commercial moisture analyzer (MA) based on the cavity ring-down spectroscopy (CRDS) was connected to the multi-gas trace-moisture generator as a device under test, and its performance was evaluated by comparing the readings with the standard values. The MA showed lower accuracy and stability in the measurement of moisture in He when it was used in the default setting, whereas it performed well in N2. This performance difference is related to the pressure-broadening coefficient of water, which depends on the gas species. We estimated the pressure-broadening coefficients of water by N2, Ar, He and O2 and discuss the gas-type dependence of the performance of CRDS-based MAs.