A systematic evaluation of contemporary impurity correction methods in ITS-90 aluminium fixed point cells

Abstract

The fixed points of the International Temperature Scale of 1990 (ITS-90) are the basis of the calibration of standard platinum resistance thermometers (SPRTs). Impurities in the fixed point material at the level of parts per million can give rise to an elevation or depression of the fixed point temperature of order of millikelvins, which often represents the most significant contribution to the uncertainty of SPRT calibrations. A number of methods for correcting for the effect of impurities have been advocated, but it is becoming increasingly evident that no single method can be used in isolation. In this investigation, a suite of five aluminium fixed point cells (defined ITS-90 freezing temperature 660.323 °C) have been constructed, each cell using metal sourced from a different supplier. The five cells have very different levels and types of impurities. For each cell, chemical assays based on the glow discharge mass spectroscopy (GDMS) technique have been obtained from three separate laboratories. In addition a series of high quality, long duration freezing curves have been obtained for each cell, using three different high quality SPRTs, all measured under nominally identical conditions. The set of GDMS analyses and freezing curves were then used to compare the different proposed impurity correction methods. It was found that the most consistent corrections were obtained with a hybrid correction method based on the sum of individual estimates (SIE) and overall maximum estimate (OME), namely the SIE/Modified-OME method. Also highly consistent was the correction technique based on fitting a Scheil solidification model to the measured freezing curves, provided certain well defined constraints are applied. Importantly, the most consistent methods are those which do not depend significantly on the chemical assay.

Comparison between the liquidus temperatures of tin samples having different impurity compositions and correction of the impurity effect

“…As shown in table 1, for each sample, the impurity compositions obtained from different analysis laboratories showed poor consistency in terms of the detected elements and the concentrations, which may lead to significant discrepancies in the SIE corrected fixed-point temper atures. As similar inconsistencies have also been reported in several publications for different fixed-point substances [1,32,36], it is of paramount importance to assess the correctness and consistency of the chemical assay by comparing the measured difference in the fixed-point temperatures with the calculated difference in the SIE corrections before correcting for the impurity effect. Resistance ratios of the liquidus temperatures and the freezing temperatures, for which the SIE correction were applied, or the impurity effect was not corrected but its overall effect was given as the uncertainty factor (i.e.…”

“…the liquidus slope or the distribution coefficient). The difficulty of obtaining the former, by chemical assay, has been well documented and is still considered challenging due to the inconsistency in the results of the chemical assays [6,32,34,36]. On the other hand, the liquidus slopes and distribution coefficients have recently been well characterized and the situation continues to improve [1][2][3]35].…”

The equilibrium liquidus temperatures of rhenium–carbon, platinum–carbon and cobalt–carbon eutectic alloys

“…Thirdly, even when a number of independent laboratories agreed that a contaminant was present, the manufacturers assay may (though not always) give a much lower reading. In only one case in [8], though, was an impurity detected by two independent laboratories at levels substantially above the detection limits that was not also showing a high value in the third. It seems that false positives are common, but not false negatives.…”

“…Reliable measurement of low levels of contaminants is far from being routine. Taking the data from a recent study [8], a number of confounding problems are found. Firstly, particular laboratories consistently found particular elements across multiple samples which were not found by other laboratories.…”

Effect of oxygen on trace impurities in extremely pure metal ingots used for realising ITS-90 fixed point temperatures

“…Impurities in pure metal substances are a major source of uncertainty in realising the International Temperature Scale of 1990 (ITS-90) [3] fixed-point temperatures [2]. There are several techniques for accounting for the influence of impurities [4,5] and these have recently been compared using an aluminium fixed point [6]. A key technique for quantifying the effect of given impurities is the Sum of Individual Estimates (SIE) method [4,5,7,8].…”