Traceability of temperature measurements in Estonia

This paper presents an overview of the establishment of the national measurement standard for temperature at METROSERT – the national metrology institute in Estonia. The temperature scale and traceability arrangements are realized in the temperature range from – 40 °C to + 300 °C by means of a group of calibrated standard platinum resistance thermometers and water triple point cells. The linkage of the realized water triple point temperature to the international key comparison value with the expanded uncertainty of 187 μK (coverage factor k = 2) is demonstrated. The established laboratory is capable of providing calibration services to customers by comparison calibration with the expanded uncertainty (k = 2) values of 5 mK in the range from – 40 °C to + 200 °C and 12 mK in the range from + 200 °C to + 300 °C.


INTRODUCTION
* Social advantages and economic competitiveness in modern society are supported by the effective national measurement system which is a generally recognized instrument in providing reliable measurement results traceable to the units of SI [1]. The traceable measurement results are important in industry (e.g. fuel combustion technologies), consumer care (food production and inspection), emerging technologies (biotechnology and nanotechnology enterprises), etc. [2]. In general, most of the European countries have developed their national measurement systems with the history dating back to more than a hundred years. Nevertheless, new independent countries with evolving economy and needs for elaboration of the metrology infrastructure are still emerging. This is also the case for Estonia, where the development of the new national measurement system was initiated only in the early 1990s. The establishment of a new national measurement standard is a long-aimed process that must be supported by good planning for optimal use of resources. The development of the national standard for temperature in Estonia was * Corresponding author, riho.vendt@metrosert.ee started with the survey initiated by the Ministry of Economic Affairs in 2002 [3]. The survey gathered the needs for the calibration services of the local leading industrial enterprises and scientific institutions. As a result of this survey, the initial task for the establishment of the national standard for temperature in the range from -40 °C to + 300 °C with an approximate uncertainty level of 10 mK was raised. The present paper describes on the example of METROSERT -the national laboratory for measurement standards in Estonia -how a system of standards for temperature can be established. The traceability arrangements, linkage of the realized water triple point temperature to the international key comparison value, stability issues of standard platinum resistance thermometers (SPRTs), dissemination of the temperature scale, and uncertainty budget are presented in detail.

ESTABLISHMENT OF THE TEMPERATURE SCALE
The realization of the International Temperature Scale of 1990 (ITS-90) at METROSERT (Fig. 1) is based on a group of calibrated SPRTs (Table 1). Standard thermo-   Table 2). The properties of the bridge have been studied thoroughly in the range of the resistance ratios from 0.4 to 4.0. The deviations of the measured ratio values from the reference values were found to remain within ± 0.1 × 10 -6 [5]. Taking into account additional uncertainty sources in calibration, and the use of the bridge (e.g. linearity of the bridge, use of multichannel switch and environmental conditions), the total contribution to the uncertainty budget arising from the use of the thermometry bridge with the multiplexer is estimated to be less than 0.7 mK. The standard resistors are immersed into the temperature controlled oil bath Lauda Ecoline E200 in order to reduce the temperature effects on high-accuracy resistance ratios needed in temperature measurements. The temperature in the oil bath is set to the reference temperature at which the resistors have been calibrated, i.e. 23 °C with stability of ± 0.02 °C.

REALIZATION OF THE TRIPLE POINT OF WATER
Three different water triple point cells (Table 3) are used at METROSERT to represent the temperature of the triple point of water (TPW) according to the ITS-90 as the national reference. The realization of the TPW is obtained by following the standard procedures [6]. . Each of these laboratories had to determine and report the temperature difference (T x − T 679 ) between the local cell "X" and the circulating cell LNE-679 with the associated uncertainty U (T x -T 679 ) according to the prescribed procedures [9]. The results of METROSERT are presented in Table 4.   in Table 3  The results are presented in Table 5.

STABILITY OF SPRTs
The properties of SPRTs are always affected by several factors even when handled with care [14]. Therefore, the stability of the SPRTs is periodically checked by measurements of the electrical resistance at the temperature of the TPW. As the temperature scale is imported to METROSERT by calibration of the SPRTs at other laboratories, the resistance value at the TPW is checked before the SPRTs are shipped for calibration and after the SPRTs have returned from calibration. Comparison of these resistance values gives information about the shipping conditions and allows evaluation of the continuing validity of calibration results. A sample diagram of the stability monitoring of the SPRTs is depicted in Fig. 2. In general, the changes in the resistance values at the TPW are observed to remain within the limits of ± 0.1 mΩ during the calibration period of 2 years (i.e. 1 mK in terms of temperature for 25 Ω SPRT).

DISSEMINATION OF THE SCALE
The temperature scale is disseminated to the customers by comparison calibration. The calibration points for the platinum resistance thermometers are selected as the approximate fixed point temperature values according to the defined sub-ranges of ITS-90 with additional temperature values between the fixed point temperatures.  (Table 6). The particular parameters of the baths (temperature range, stability and uniformity) have been selected with the aim to achieve the overall expanded calibration uncertainty 0.01 °C (k = 2) in the temperature range from -40 °C to + 300 °C. The average number of 50 routine comparison calibra- --------* In equalization block.
tions per year since the official recognition is an indication of a live and well-operational laboratory. The established temperature scale is also a basis for scientific research activities and non-contact temperature measurements [15].

UNCERTAINTIES
The uncertainty evaluation is based on the Guide to the Expression of Uncertainty in Measurement [16]. The uncertainty budget for calibration of platinum resistance thermometers (PRTs) at the TPW is presented in Table 7. The uncertainty estimation for the realization of the TPW temperature 0.01 °C takes into account the bias caused by isotopic and other effects according to the results of the EUROMET project No. 714. The uncertainty budget for comparison calibration in a liquid bath at a single calibration temperature within the range from -40 °C to + 300 °C is presented in Table 8. It is assumed that the PRTs under calibration are featuring similar properties as SPRTs: good stability, no hysteresis, and purity of the metal as described in ITS-90 [6].

CONCLUSIONS
The international temperature scale ITS-90 has been established on the secondary level at the national standard laboratory for temperature in Estonia, METROSERT, in the temperature range from -40 °C to + 300 °C. The traceability of the scale is based on the group of SPRTs calibrated regularly by the National Metrology Institute of Finland. The temperature values realized by the TPW cells at METROSERT are in good agreement with the definition of kelvin and the reference value of the  Expanded uncertainty U (at confidence level of 95%, coverage factor k = 2) 5 12 --------* Instability of the standard thermometer is estimated by the regular checks at the TPW. In calibrations, the latest resistance value R (0.01 °C) for SPRTs is used.
international key comparison CCT-K7 within the estimated measurement uncertainties below 0.19 mK for all three cells. The temperature scale is disseminated to customers by comparison calibrations at selected temperatures with expanded uncertainty (k = 2) values of 5 mK in the range from -40 °C to + 200 °C and 12 mK in the range from + 200 °C to + 300 °C. The national standard laboratory for temperature has proved itself as a functional institution, providing customers routinely with about 50 comparison calibrations per year in Estonia.