High purity tin samples were doped with Fe, Sb, Pb, In and Al. These impurities shifted the run-off point of the melting curve of tin by -0.75mK (ppm wt)(-1), + 0.23mK(ppm wt)(-1), - 0.16mK(ppm wt)(-1) , -0.21mK(ppm wt)(-1) and - 2.3mK(ppm wt)(-1), respectively (ppm wt denotes the mass fraction in parts per million). However, 1654 ppm wt of Fe and 12.2 ppm wt of Al concentrations appear to supersaturate tin. The extra amount of these impurities then gradually precipitated as the doped samples were subjected to repeat melting.In the 10 ppm wt Fe to about 970 ppm wt Fe concentration range tin has a solid - liquid phase transformation of very narrow melting range, 6.9mK below the tin-point.

Experiments were carried out in order to determine why a given eutectic sample approaches its state of equilibrium either rapidly or slowly during melting-curve determinations. The results of these experiments suggest that, when eutectic liquids are frozen, nonequilibrium solids normally form. The experimental results also suggest that the melting temperatures increase and the reproducibility of the melting curves improves considerably if the nonequilibrium solid is converted into the equilibrium eutectic solid lattice. As specific examples, Al–Cu and Al–Ag eutectic samples were studied.

Equilibrium melting curves of nominally 99.9999% pure Al-Cu eutectic samples were realized. The end of the linear portion of the melting curves, the run-off point, was found to be (548.1608 +/- 0.0007). C. The uncertainty is one standard deviation of the twelve measured values. Some of the samples were then doped with Si, Fe and Ag. These impurities shifted their run-off points by -0.68 mK ppm(-1), -0.29 mK ppm(-1) and -0.17 mK ppm(-1), respectively.

The shift of the melting curves of Zn samples was measured as a function of impurity concentration by explicit doping with some of the impurities that are usually present in high purity Zn samples. The following shifts were obtained in mK/ppm wt ( mass fraction in parts per million) for Fe, Pb, Cd, In, Cu, Ag and Au, respectively: -0.51, -0.27, -0.33, -0.39, +0.23, +0.26 and +0.078.

A series of experiments were carried out in order to observe the melting and the freezing behaviour of Al-Si eutectic samples under static (adiabatic) and dynamic (steady heating or cooling) conditions. Afterwards the samples were analysed by glow discharge mass spectrometry. Subsequently they were doped with the detected major impurities of Ag, Cu and Fe to find out by how much these impurities shift the melting curves. The following results were obtained: the equilibrium melting temperature at the run-off point was found to be (758.792 +/- 0.003)degrees C and the above impurities shifted this value by 0.12 mK ppm(-1), 0.20 mK ppm(-1) and 0.22 mK ppm(-1) by weight, respectively.

Eleven platinum resistance thermometers (PRT) were compared at the Al - Cu eutectic point (548.2 degrees C) in order to measure the non-uniqueness of ITS-90 roughly half-way between the Zn and the Al points. It turned out that the cubic deviation function describes this set of PRTs within +/- 0.6 mK in this temperature interval.In addition, the measured data were used to determine the non-uniqueness and the sub-range inconsistency, using cubic and quadratic deviation functions, at the In point (156.6 degrees C). The non-uniqueness was found to be +/- 0.26 mK, and the sub-range inconsistency +/- 0.12 mK.

Twelve platinum resistance thermometers (PRTs) were repeatedly calibrated between the triple point of water and the silver point. During each calibration their resistance was also determined at the Cu-Ag eutectic point. The temperatures indicated by the PRTs were within +/-1 mK, except for one that differed by 4 mK from the average, 779.6360degreesC, of the other eleven.Four of these PRTs had a lower W (Ag) value than acceptable by the ITS-90, including the one that differed by 4 mK. Thus, within this set of PRTs all the acceptable' ones and three out of four 'unacceptable' ones indicated a unique temperature within I mK halfway between the aluminium and the silver points.

Experiments, described below, revealed certain shortcomings of the current definition of the Al point: pressure in sealed type Al cells made of quartz was observed prior to their sealing and was found to steadily decrease. Consequently, the realized Al points using sealed quartz cells are in a state of drift. In addition, the initial Al point values yielded by open cells are higher than the subsequently realized ones: five different gases were used to apply the initial 1 atm pressure over the Al samples (air, N2, Ar, He, CO2). Each of them appeared to contaminate slightly the Al samples, lowering the initial value of the Al point by 1 or 2 mK. However, sealed quartz triple point cells remained stable at their initial value indefinitely. Thus, the triple point of Al is a more reliable fixed point than the defining Al freezing point itself.

Aluminium samples were doped with an assortment of impurities and the shifts of their equilibrium melting curves were determined under static adiabatic experimental conditions. The shift of run-off temperatures (i.e. when the sample temperature breaks off from the melting `plateau' and rapidly begins to rise) of originally 99.9999% pure Al samples caused by impurities was determined to be as follows (in millikelvin per part per million (ppm) by weight): Ag, −0.12; Zn, −0.16; Cu, −0.29; Fe, −0.35; In, −0.21; Quartz, −0.33; Si, −0.66; Ti, +3.30; Mn, −0.063; Cd, −0.10; Sb, −0.19; Ca ?; Ni, −0.43.

PRTs were soaked at different temperatures, and the resulting changes caused by oxidation and the subsequent dissociation of the formed oxide were observed. Soaking below 400 degreesC initiated oxidation, in some platinum resistance thermometers (PRTs), and then their resistance steadily increased with increasing temperatures up to about 525 degreesC. When soaking near 525 degreesC, the resistance of the PRT remained stable, suggesting that the thermal energy "kT" equals the energy that binds oxygen to platinum, at the existing oxygen pressure. At temperatures higher than 525 degreesC the formed oxide dissociated. The higher the temperature, the higher the thermal energy and thus the higher the rate of dissociation. One hour of soaking near the Al point dissociated the oxide that took hundreds of hours to accumulate. Three out of seven PRTs tested however would not oxidize while soaking for several months. Further soaking, however, did initiate oxidation in two of the three PRTs.

A set of fixed points maintained by national metrology institutes (NMIs) for calibrating platinum resistance thermometers (PRTs) in Canada (National Research Council, NRC), Argentina (Instituto Nacional de Tecnologia Industrial, INTI) and Brazil (Instituto Nacional de Metrologia, Normalizaçãoe Qualidade Industrial, INMETRO) were compared using two methods. A set of portable cells was used in one method, and in the other a calibrated PRT. The fixed-point differences were as follows. For the INTI the differences ΔT = T(INTI) − T(NRC) found at the Ag, Al, Zn, Sn, In, Ga and Hg points were 3.5 mK, 2.8 mK, 1.8 mK, −0.5 mK, 0.2 mK, −0.2 mK and −0.5 mK, respectively. For the INMETRO the differences ΔT = T(INMETRO) − T(NRC) at the above fixed points are 0.7 mK, 3.3 mK, −1.3 mK, −1.9 mK, −0.3 mK, −0.1 mK, −0.4 mK, respectively, and 0.8 mK at the Ar point.

Six silver samples were doped with Pd, Fe, Au, In, Sn, Pb, Zn, Sb, Cu, Ni and Al. The resulting shifts in the equilibrium melting curves were then determined. All impurities caused a shift in the liquidus point of silver except aluminium, for which no conclusive results could be obtained. The shifts, in millikelvins, caused by one part per million by weight of each of the above impurities (with the exception of the last mentioned, Al), were determined to be +0.97, +0.13, +0.09, −0.49, −0.60, −0.48, −1.10, −0.73, −0.69, and −0.74, respectively. Four of the initial samples were recently acquired and of 99.9999 % purity. Their melting points were realized within a scatter band of ±0.55 mK. Using the above measured shifts and the amount of impurities detected in these samples by glow discharge mass spectrometry (GDMS), it is estimated that the liquidus points of these samples differ from that of ideally pure silver by no more than 1.0 mK within an expanded experimental standard deviation of the mean of 0.2 mK, using a coverage factor of 2.

A high-temperature calorimeter has been developed for studying the equilibrium melting behaviour of substances used for realizing the thermometric fixed points. Conventional cells of either 5 cm or 2.5 cm diameter can be accommodated. Using the calorimeter, the equilibrium melting curves of three high-purity silver samples were measured. The samples have narrow melting ranges: two of 1.5 mK and one of about 0.5 mK from 10 % liquid sample fraction to the last measured point near 70 % liquid, yet their liquidus points differ by nearly 5 mK. It was found that 1/F plots cannot be used to determine the deviation of the melting point of real silver samples from that of ideally pure silver.

The reproducibility of some thermometric fixed points and the accuracy of four platinum resistance thermometers (PRTs) were studied. It was found that the fixed points of aluminium (Al), zinc (Zn), tin (Sn), indium (In) and gallium (Ga) were realized reproducibly within ±0.17 mK; ±0.11 mK; ±0.10 mK; ±0.13 mK and ±0.12 mK, respectively. Because the actual impurities and their concentration in our samples (of 99.9999% or 99.999 99% purity) are unknown, the systematic uncertainty due to impurities cannot be estimated. However, any of the samples of Ga, In, Sn, Zn and Al is consistent with the rest within ±0.2 mK, using a cubic or quadratic deviation function, in the temperature range 0 °C to 660 °C. This indicates that the effect of impurities is negligible. Four PRTs were selected at random. They were calibrated repeatedly, first up to the Zn point and then up to the Al point. The resistance of each PRT drifted. From time to time, for each PRT, a seemingly well-established resistance drift suddenly and unpredictably changed to a different rate of drift. Occasionally, the resistance of the PRTs shifted. Such unpredictable changes obviously limit the accuracy of temperature measurements using PRTs no matter what the accuracy of their calibrations. In the case of our four PRTs, the uncertainty of temperature measurements near 660 °C ranged from about ±1 mK to about ±2.5 mK even though they were all calibrated at all fixed points well within ±0.25 mK uncertainty. Possible explanations are offered for the apparently permanent drifts and the erratic shifts in the resistance of the PRTs. Some comments are made concerning the ambiguity of "immersion tests" in general. The furnaces of the National Research Council of Canada used in this work are high-temperature adiabatic calorimeters.

Causes of uncertainties in temperature measurements are analysed. Some uncertainties are intrinsic to the temperature scale itself (built-in uncertainties). The rest are due to an assortment of experimental uncertainties. A simplified calibration procedure is proposed in the temperature range 0 °C ≤ T ≤ 660 °C that reduces the time (and thus the cost) of calibration by about 30% without a significant change in the accuracy of the scale imprinted on the calibrated platinum resistance thermometers.

The triple point and melting point of mercury were studied by continuous melting and adiabatic calorimetry techniques. Two geometrical configurations of the sample are compared and the effect of impurities considered.

Drift of emf in bare-wire, base metal thermocouples is linked to the precipitation of alloying components that supersaturate the host (Ni) lattice. Using a simple phenomenological description, one can visualize the process of precipitation from supersaturated solid solutions. Some possible reasons for the changes in alloy composition in metal sheathed thermocouples are suggested.

The realizations of the fixed points of the International Temperature Scale of 1990 (ITS-90) maintained in Canada and Mexico were compared between -38 °C and 420 °C, using a portable fixed-point realization apparatus. In this range, the routinely realized defining fixed points of the ITS-90 in these countries were found to coincide to within 0,5 mK.

An analysis of calibration data suggests that Wref(In) and Wref(Ga), the resistance ratios at the freezing point of indium and the melting point of gallium, may have been assigned temperatures a fraction of one millikelvin higher and lower, respectively, on the ITS-90, than the values indicated by some high-purity samples. Illustrations are given of how the quality of calibration services can be assessed using issued calibration data.

We present the design of a cryostat that houses trichlorofluoromethane triple-point cells. The apparatus was used to measure the triple point adiabatically and then compare results with the values obtained using the continuous melting technique.