At the request of the Consultative Committee for Thermometry (CCT), Working Group 4 (WG4) has critically reviewed all available measurements of the differences between thermodynamic and ITS-90 temperatures, (T − T 90 ), and documented the conversion of older data to the ITS-90. Particular attention has been given to the uncertainties. Based on this review, we provide consensus estimates of T − T 90 for selected measurements from 0.65 K to 1358 K. We provide two analytic functions for T − T 90 , one for use from 8 K to the triple point of water (T TPW ) and one for use above T TPW . The small discontinuity of the derivative dT 90 /dT at T TPW is discussed.J. Fischer (B) · L. Wolber PTB,We also identify temperature ranges where researchers are encouraged to undertake high-accuracy measurements of T − T 90 .

Within an international collaboration of the eight metrological institutes represented by the authors, the dependence of the triple-point temperature of equilibrium hydrogen on the deuterium content at low concentrations has been precisely determined so that the uncertainty in realizing the triple point as a temperature fixed point might be reduced by nearly one order of magnitude. To investigate the thermodynamic properties of the hydrogen-deuterium mixtures and to elucidate the factors that influence the melting temperature, 28 sealed fixed-point cells have been filled and measured, and some of these have been compared with an open-cell system. Hydrogen gas with a deuterium content ranging from 27.2 µmol D/mol H to 154.9 µmol D/mol H was studied using cells containing five different types of spin-conversion catalyst, with different catalyst-to-liquid volume ratios (a few per cent to more than 100%) and of different designs. The latter consideration is especially influential in determining the thermal behaviour of the cells and, thus, the temperature-measurement errors. The cells were measured at the eight participating institutes in accordance with a detailed protocol that facilitates a direct comparison of the results. Through analysis of the measurements, significant inter-institute deviations due to different measurement facilities and methods have been ruled out with respect to the determination of both the melting temperatures and the thermal parameters of the cells. The uncertainty estimates for the determination of the deuterium content have been verified by including isotopic analysis results from four different sources. The slope of the dependence of the triple-point temperature of equilibrium hydrogen isotopic mixtures on the deuterium content has been deduced from the melting temperatures of those sample portions not in direct contact with the catalysts. Evaluation of the data using different mathematical methods has yielded an average value of 5.4 2 µK per µmol D/mol H, with an upper bound of the standard uncertainty of 0.3 1 µK per µmol D/mol H. This is close to the literature value of 5.6 µK per µmol D/mol H that was obtained at higher deuterium concentrations.

Following the finalization of the work performed to establish the triplepoint temperature versus isotopic composition relationship for protium (Metrologia 42, 171 (2005)) adopted into the ITS-90 definition by the International Committee for Weights and Measures (CIPM) in 2005, and a preliminary exploration of the variability in the triple-point temperature of neon gas samples arising from differences in isotopic

As a basis for evaluating the results of an international star comparison of sealed fixed-point cells, dedicated investigations have been directed to the dependence of the melting temperature on different conditions concerning the preparation of the solid phase, i.e., fast and slow freezing, refreezing without supercooling, or annealing at a temperature of only a few mK below the melting temperature. Differences in the typical thermophysical behavior of the four fixed-point substances hydrogen, neon, oxygen, and argon have been found. In the case of hydrogen and oxygen, the quality of the crystal lattice has little influence on the melting temperature. This enables temperature widths of the melting curves of only a few tens of µK, if there are no additional influences. On the contrary, argon samples frozen after supercooling with different velocities of freezing typically melt within a range of 0.3 mK. The meltingcurve width can be reduced only by refreezing. A broader melting range of a few tenths of mK has been typically observed for neon cells. Unlike argon, an improvement of the crystal quality by a slow refreezing does not decrease the width of the melting-curve.

No abstract