Approximating the ITS-90 temperature scale with industrial platinum resistance thermometers

Platinum resistance thermometers (PRTs) are widely used for accurate temperature measurements in industrial process control as well as in testing and calibration laboratories. Industrial-type PRTs (IPRTs) are available with platinum wires of different purity and can attain measurement accuracy at the level of few tens of millikelvin in a broad temperature range from -196 degrees C to 550 degrees C and above. For such IPRTs, the most-used interpolation model (resistance versus temperature) is based on the Callendar-Van Dusen (CVD) equation, which is also recognized in several industrial standards including IEC 60751 and the corresponding national standards. In recent years, several studies have shown that systematic differences exist between the ITS-90 temperature (T-90) and the temperature calculated by the CVD function. When the CVD equation is used to fit experimental data, the difference can be as large as several tens of millikelvin, even near a calibration point, i.e., of the same order of magnitude as the experimental uncertainty routinely achieved in laboratory calibrations. In order to overcome the above limitations, many interpolation models were proposed. The aim of this work is to assess the use of ITS-90 defining equations in precision laboratory calibrations of IPRTs in the temperature range from -196 degrees C to 420 degrees C. Twenty IPRTs with W(100) ranging from 1.384 to 1.392 were calibrated by comparison against a standard PRT, and the experimental data were processed using several interpolation schemes based on ITS-90 deviation functions with different degrees of freedom. The overall results showed that any ITS-90-based scheme performs better than the CVD equation, suggesting that it be applied to a broad spectrum of industrial and laboratory applications.