An Intercomparison of High Temperature Platinum Resistance Thermometers and Standard Thermocouples

Abstract

To obtain some of the information necessary for redefining a practical temperature scale in terms of resistance thermometry up to the gold point, we have intercompared nine high temperature platinum resistance thermometers and eight standard thermocouples at 50 degree intervals over the range 630.74 to 1064.43 °C. Values of temperature were derived from "before and after" calibrations of the instruments at the freezing points of gold, silver, antimony, zinc, and tin, and at the triple point of water. – Analysis of the data showed that the standard deviation of temperatures measured with thermometers was about 4 mK, while with thermocouples it ranged from 25 to 30 mK. When systematic errors were accounted for, we found that values of temperature on IPTS-68 could be determined with a single thermometer, using an interpolation formula analogous to that used below 630.74 °C, with about one-third the uncertainty usually attributed to a single thermocouple determination (±200 mK). The data also revealed that there is a lack of "smoothness" in the scale at 630.74 °C – first derivatives with respect to t68 are discontinuous by about 0.1% at this temperature. – We conclude from the results of this work that if a practical temperature scale were redefined in terms of presently available high temperature platinum resistance thermometers, it could be realized with significantly less uncertainty (±12 mK for a single determination) than IPTS-68, as it is now realized with thermocouples (±200 mK for a single determination). We recommend, however, that the adoption of such a scale be deferred until indicated improvements in thermometers have been made, and until current researches in gas thermometry have provided new values of thermodynamic temperatures for calibrating fixed points.