Uncertainties in the SPRT Subranges of ITS-90: Topics for Further Research

Abstract

The CCT has completed the guide summarizing the uncertainties in the realization of the SPRT subranges of ITS-90 between the triple point of neon (24.5561 K) and the freezing point of silver (961.78 • C). This article identifies aspects of standard platinum resistance thermometry where either data or models are lacking and further research is required. In the calibration of SPRTs, the two main concerns are the need for data on liquidus slopes for the different impurities in the fixed points and improved understanding of the impact of the thermal environment of the fixed point on the realized temperature. In the use of SPRTs, the two largest sources of uncertainty are Types 1 and 3 non-uniqueness and oxidation. The causes of Type 3 non-uniqueness are not yet understood, especially at low temperatures, and there is a paucity of data for the high-temperature subranges. In respect of oxidation, there is a need for validation of the models developed in the 1980s, especially in light of the reduced partial pressure of oxygen used in modern SPRTs. A range of other effects including vacancy effects in SPRTs, isotopic effects in fixed points, and improved statistical methods are discussed.

Uncertainty Propagation for Platinum Resistance Thermometers Calibrated According to ITS-90

“…This result resembles the result of an example in CCT/08-19/rev [17] (Appendix C, Fig. C.3) for the same temperature range and with similar assumptions for the input quantity correlations (Eqs.…”

“…C.3) for the same temperature range and with similar assumptions for the input quantity correlations (Eqs. C.14 and C.23 in [17]). This may be interpreted as a justification of the interpolation method, since this method usually includes approximations in addition to the GUM [2] (e.g., W ≈ W r [17]), while the presented matrix formalism is fully GUM compliant for any temperature range and any conceivable correlations of the input quantities.…”

“…To cover all possibilities is way beyond the scope of this paper. A more thorough discussion of PRT calibration scenarios leading to different states of correlation is given in [9,17]. A striking advantage of the presented method of uncertainty propagation for PRTs is that the input quantity correlations solely affect the corresponding uncertainty matrix U in Eq.…”

“…, W FPn as input quantities, careful analysis of the calibration procedures is needed. First results to this are found in [9,17]. The uncertainty of R was set constant over the whole temperature range to focus on the influence of the correlations of the resistance ratios W FP on u(T ).…”

“…In [17], CCT WG3 recommends the use of the interpolation method after White and Saunders [11,12] rather than algebraic methods. But this interpolation method usually implies certain approximations in addition to the GUM [2].…”