A primary calibration of platinum resistance thermometer (PRT) is performed in the temperature range between the triple point of oxygen (54.3584 K) and triple point of water (273.16 K) within the total uncertainty of ±1 mK based on the International Temperature Scale-1990 (ITS-90). For the primary calibration, a series of realizations of triple point of water, mercury (234.3156 K), argon (83.8058 K) and oxygen are practiced. Impurity effect on the realization of the triple point, β-γ phase transition of solid oxygen and triple point of nitrogen as a secondary fixed point are also performed to increase the accuracy of the calibration. An excellent agreement within ±0.1 mK is achieved in the inter - comparison of the KRISS - cell for the triple point of water to the cell verified in national standard laboratories of US (NIST), England (NPL) and Japan (ETL). Reproducibility of the measurement of the primary fixed points are within ±0.1 mK and the agreement of nitrogen triple point to calibrated temperature scale is within ±0.1 mK.
For the establishment of the accurate temperature scale, a number of investigations of the effect of content and amount of impurities even in ultra high quality specimens on the triple point have been performed. This study is a continuing effort, however in a dynamic point of view - to investigate the effect of redistribution of impurities on the reproducibility of the fixed point realization. Freezing and γ-β transition curves of oxygen are studied for trace impurity effects in commercial oxygens (Matheson Research Grade, 99.997% purity) and ultra high purity oxygens obtained from thermochemical decomposition of KMnO4. Both freezing and γ-β transition curves of oxygen are observed to show phase separations when freezing and melting processes were repeated with the rate of temperature change slower than 40 mK/min. The simple system of single phase, where impurities are distributed homogeneously, seems to be transformed to a complex system in the slow freezing or melting process where impurities are redistributed to form a heterogeneous phase of differentiated impurity concentrations. Temperature and bandwidth differences between the separated transition curves may be correlated with the impurity distributions in the oxygen to improve the data analysis for the oxygen fixed points. The separation is also observed in a melting curve of oxygen employing both continuous and pulse heating method.
The effect of the redistribution of impurity is utilized in the analysis of the content and amount of impurity in high purity oxygen. One may suggest to use the melting curve corresponding to the pure region after performing a simple procedure of inducing impurity redistribution into a pure and impure region, for a determination of the triple point of oxygen.