Fixed-Point Comparison Uncertainties for Two Cell Geometries

To realize the ITS-90 according to its definition, among others, the melting and freezing temperatures of ideally pure metals are needed. Therefore, many national metrology institutes (NMIs) utilize a group of cells instead of one single cell as the national reference for each temperature. With direct fixed-point cell comparisons on a regular basis, it is feasible to account for the small differences between the individual fixed-point temperatures and to detect possible temperature drifts of the cells. At PTB (the German NMI), in recent years, these groups of national standard cells and the so-called transfer cells for calibrations have been complemented by newly developed slim fixed points. These cells typically contain 75% to 80% less fixed-point material compared with standard cells. Slim cells are used for homogeneity investigations of large batches of fixed-point material, for doping experiments to determine the influence of very small amounts of impurities on the fixed-point temperature with very small uncertainties, and for the investigation of contamination or purification effects after the manufacture of a fixed-point cell. These investigations have shown that the main limitation of slim cells is the quality of the phase boundary. The small dimensions of the cell do not allow the formation of a closed phase boundary (or even two of them). However, this can be compensated using a quasi-adiabatic realization procedure, and in this way, uncertainties comparable to those of standard fixed-point cells can be achieved. In this article, the design of the cells as well as typical measurement results and uncertainties for the direct comparison of fixed-point cells of both types, the standard size and slim design, are presented.