Temperature scales, such as the International Temperature Scale of 1990 (ITS-90), are established to ensure unique temperature measurements throughout the world with smallest possible uncertainties.
These include thermometric fixed points which are based on melting-, freezing- or triple points of pure substances. The comparison of fixed-point temperatures and the interpolation of the scale between fixed points in the temperature range between about -259 °C and 962 °C is carried out by platinum resistance thermometry.
The largest uncertainty contribution is the purity of the fixed-point materials. For the metallic fixed-point materials (Hg, Ga, In, Sn, Zn, Al, Ag, Cu) currently used at PTB the purity is between 99.9999 % and 99.99999 %. The resulting uncertainties in fixed-point temperatures are between 0.2 mK and 3 mK. In order to ensure the long-term stability of the scale realization the definition is based on the properties of ideally pure substances. This requires very demanding quantitative chemical trace analysis and the knowledge of phase diagrams at the pure-element side. Published phase diagrams are based on measurements by means of thermal analysis (DSC, DTA). The lower limit of determination for the mass fraction of impurities is about 0.1 % (10‑3 g/g) and for the determination of phase transition temperatures about 100 mK.
We investigate phase diagrams by means of doping experiments with modified fixed-point cells and quasiadiabatic calorimeters at impurity mass fractions down to 1 ppm (10‑6 g/g) with uncertainties in temperature measurement of about 100 µK.
