Vol.2 No.3 2009

Research paper : A marked improvement in the reliability of the measurement of trace moisture in gases (H. Abe)−220−Synthesiology - English edition Vol.2 No.3 (2009) simple method to generate a standard gas. However, as found in this research, it should be noted that the reliability of this method in the trace concentration region may not be maintained without using a magnetic suspension balance when the evaporation rate is an extremely small value.6 Setting goal values for uncertaintyQuestion and comment (Kazuo Igarashi, Measurement Solution Research Center, AIST)In chapter 2, you mentioned that the reason for setting 11 % (corresponding to a frost point of 0.5 ºC) as the target value of the relative standard uncertainty near the frost point of -100 ºC is that the goal of NPL for the standard uncertainty at a frost point of -95 ºC was 0.5 ºC at the time. The research yielded the excellent result of maintaining the relative combined standard uncertainty at about 3 % by reducing the uncertainties of many components. However, I would like to know the suitability of the 11 % goal because the method of water vapor generation and the accompanying processes were different.Answer (Hisashi Abe)Because the significance of selecting the diffusion tube method would be lost if the uncertainty was much greater than that for the frost point method, to begin with we used the target value of the frost point method for the diffusion tube method. Furthermore, we expected that the residual moisture in the zero gas and adsorption/desorption moisture would be major uncertainties in the diffusion tube method, and for the high-performance purifiers used to remove the moisture for zero gas generation, the catalogs from many manufacturers state that the residual moisture is 1 nmol/mol or less. Assuming that the uncertainty caused by the adsorption/desorption of moisture would be at this level, the combined uncertainty of these components would be 1.4 nmol/mol ( 12 + 12 1.4 ), which corresponds to 10 % at 14 nmol/mol. Considering that other components of uncertainty would be included, we did not think at the time that we could achieve the 11 % figure easily. Note that NPL presently sets the lower limit of standard generation by the frost point method to a frost point of -90 ºC , and reports that the standard uncertainty at -90 ºC is 0.2 ºC (about 3.7 %). Therefore, the target value of 0.5 ºC for a standard uncertainty at -95 ºC initially set by NPL was perhaps slightly too high.7 Cause of temperature difference in the generation chamber and diffusion cellQuestion and comment (Kazuo Igarashi)The evaluation of uncertainty is performed in section 3.3, and you mention that it arises from the difference between the temperature of water inside the generation chamber and the diffusion cell. It seems that the reasons for the temperature difference may be attributable to the temperature control depending on the flow rate of the nitrogen gas and the low heat conductivity of the SUS stainless steel used as material of the diffusion cell. If you have considered these factors, please describe them, as it will be of use in detailed investigations.Answer (Hisashi Abe)I added the assumption of the factors that produced the temperature difference to the text. Investigating the reasons why the room temperature change affects the measurement results is an important issue when considering the validity of the evaluated uncertainty, or the uncertainty of the uncertainty (here, this means the uncertainty of the sensitivity coefficient determined by experiment). However, we have not yet performed experiments regarding this issue or to investigate the flow rate dependence as you indicated, so the clear reason is still unknown. However, for the uncertainty of the sensitivity coefficient, from the discussions in Ref. [23], I do not think that it is particularly large under the current experimental conditions. If the obtained sensitivity coefficient is correct, reducing the uncertainty caused by the change in room temperature is not difficult compared with reducing the uncertainties of other components at this stage, and therefore, we have not performed any additional investigation of the reasons for the cause of the temperature difference.8 Effect of purity of waterQuestion and comment (Kazuo Igarashi)In generating trace water vapor, you analyzed several factors such as the adsorption/desorption of moisture, residual moisture in the zero gas, and the gas flow meter, but you do not refer to the purity of water itself. Is this because you think that it is not a problem as long as the purity level does not affect the vapor pressure?Answer (Hisashi Abe)I often receive questions regarding the purity of water, but I did not discuss this in the text because of the space limitation of the journal, so I shall explain it here. In this research, we used highly purified water obtained from an ultrapure water production system. However, the purity of water itself is not a major issue. This is because the evaporation rate is measured as the mass change of the diffusion cell per unit time, and therefore, even if there is an impurity in the water, it will not be an issue as long as it remains in the diffusion cell. If a gas is dissolved in the water, it may become a factor of uncertainty because it evaporates with water, but such a gas is probably removed by the aforementioned ultrapure water production system, and I think that the effect can be ignored. Furthermore, the measurements are performed after placing the diffusion tube inside the generation chamber into which highly pure nitrogen is introduced for a sufficiently long time (normally 10 days or more). Therefore, even if there is a residual gas, it will be replaced by nitrogen. Finally, for the nitrogen dissolved in water, even if it is saturated at a pressure of 150 kPa and a temperature of 25 ºC inside the generation chamber, its amount-of-substance fraction is 0.002 % or less, which can be ignored in this research.


page 47