Vol.2 No.3 2009

Research paper : A marked improvement in the reliability of the measurement of trace moisture in gases (H. Abe)−211−Synthesiology - English edition Vol.2 No.3 (2009) with each other within the uncertainty[5]. From this result, it was demonstrated that the magnetic suspension balance was extremely effective for reliably measuring an evaporation rate at the 10 µg/h level. Another finding from this result was that the evaporation in the generator could be explained on the basis of the diffusion phenomenon. This indicates that N in Eq. (2) can be sufficiently stabilized by controlling the temperature and pressure. This was also confirmed experimentally, and a stability within a relative standard uncertainty of 0.6 % was realized by precise temperature and pressure control[5]. It is interesting to note that the technology developed for the measurement of N led to the technology required for the stabilization of N.3.2.2 Measurements of adsorbed and desorbed moisture and residual moisture in zero gasMeasurements of Nb and xb were necessary to develop a trace moisture generator that fulfills the conditions xw >> xb and N >> Nb of Eq. (1) even for xw~ 14 nmol/mol (ppb) and N~ 780 nmol/h, and a high-performance trace moisture analyzer was necessary for this purpose. To address this issue, we initially considered using the vacuum ultraviolet fluorescent trace moisture analyzer (VUV)[13] developed at the National Research Laboratory of Metrology and the atmospheric pressure ionization mass spectrometer (APIMS)[3] that was introduced in 2000. VUV is an instrument that detects the fluorescence emitted from OH radicals in an electronic excited state produced by the photolysis of water molecules using vacuum ultraviolet absorption. APIMS is an instrument that detects a trace substance (in this case, trace moisture) in a gas by ionizing the sample gas using a corona discharge under atmospheric pressure, and then introducing the gas into a mass spectrometer by differential evacuation. Both VUV and APIMS are sufficiently sensitive in the trace moisture region, but the instrument itself is not scaled, and it is necessary to create a calibration curve using a standard. We considered creating the calibration curve by a standard addition method using trace moisture generated by the diffusion tube generator that was being developed. That is, assuming that Nb and xb of Eq. (1) maintain constant values, xw is varied by changing N and F, the indications of VUV or APIMS are recorded, and the calibration curve is created from the relationship of the differences between the N and F values and the indications. However, in the trace moisture range, because a long time was necessary for dry purging (the removal of residual moisture) after exchanging the diffusion cell (to change N), the baseline drift and sensitivity drift of VUV or APIMS during the experiment period could not be ignored because of their non-negligible magnitudes compared with the signal intensity. Moreover, it was unknown how much dry-purging time was necessary to ensure that Nb was the same as that before the exchange, and it was also unclear whether Nb could be considered to be approximately constant after dry purging. Therefore, as the research progressed, we found that it was difficult to create a reliable calibration curve using this method in the trace moisture region.In the trace moisture region, because we expected that the uncertainty due to adsorption/desorption moisture and residual moisture in zero gas would strongly dominate in the total uncertainty, we considered the measurement of Nb and xb to be particularly important in developing the generator. As the research progressed, we began to consider that it was problematic to use only VUV and APIMS.Around that time (about 2002), a new trace moisture analyzer based on the cavity ring-down laser spectroscopy (CRDS)[14]-[17] became commercially available. CRDS is a method where a laser pulse is coupled into an optical cavity composed of two high-reflectivity mirrors, and the detection sensitivity is increased through a long optical path length generated by bouncing the laser pulse back and forth between the mirrors. The sensitivity of this method of absorption spectroscopy has increased markedly in recent years. Because xw can be calculated directly from the absorption cross section Fig. 9 Near-infrared spectrum of water.All absorption lines in the figure are assigned to the vibration-rotation transitions of water molecule. The amount-of-substance fraction of water is determined using the 202 303 transition of the v1+v3 band that is the strongest in the figure (marked with asterisk).Fig. 8 Difference between indication of CRDS trace moisture analyzer and standard value (relative value).The dots represent [(indication – standard value) ÷ standard value] × 100.xw / (nmol·mol-1)Relative value of difference / %00102006004005-10-5 * 71757185718005Wavenumber / cm-1Residual×5Vibration-rotation transitions of water moleculexw / (nmol・mol-1)Standard value:72.1Indication:69.2Analyzed value:70.4Absorption coefficient (α) /(10-8×cm-1)


page 38