Platinum Metals Rev., 1972, 16, (1), 2
Platinum Metal Thermocouples
New International Reference Tables
As a result of a remarkable piece of international collaboration between three national standards laboratories and seven United States and United Kingdom manufacturers, new reference tables have now been completed for platinum : 10 per cent rhodium-platinum and platinum: 13 per cent rhodium-platinum thermocouples. These new tables take into account the changes in the temperature scale resulting from the introduction of the International Practical Temperature Scale of 1968 (IPTS-68) and also provide reference tables which will be common to both U.S. and U.K. manufacturers and users of thermocouples.
The changes in the temperature scale (1) when IPTS-68 was introduced highlighted the problem hitherto of the two conflicting reference tables for rhodium-platinum thermocouples. One table was based on work done in 1933 at the National Bureau of Standards (NBS) by Roeser and Caldwell (2) and was published as NBS 561, and the other was based on work done at the National Physical Laboratory (NPL) in 1950 by C. R. Barber (3) and was published in the United Kingdom as BS 1826. These tables differed from one another as a result of differences in both the realisation of the temperature scales in the original calibrations and in the compositions of the platinum and rhodium-platinum wire. Since the original measurements were made, particularly those in 1933, there have been substantial improvements in the purification of both platinum and rhodium and thus, in order to continue to meet the old tables, changes were made in the composition of the alloy arms of the thermocouples. The result was that differences between thermocouples made to meet NBS 561 and those made to meet BS 1826 have become quite substantial. It has been clear for some time that it was uneconomic for manufacturers to have to make material of nominally the same composition to these two speicifications.
In September 1967 an informal meeting took place at the NBS between representatives of the NBS, National Research Council of Canada (NRC) and the NPL. At this meeting it was agreed that the introduction of the IPTS-68 would provide an excellent opportunity to unify the reference tables for platinum thermocouples. It was agreed that a joint approach should be made by the three national laboratories to all the U.S. and U.K. manufacturers with a view to carrying out a programme of research leading to new reference tables. This was welcomed by the manufacturers, many of whom had been aware of and had encouraged the proposals discussed at this meeting, and it was agreed that the terms of reference for the project would be as follows:
Each of four American and three British manufacturers would contribute 24 metres of pure platinum wire and 12 metres each of 10 per cent rhodium-platinum and 13 per cent rhodium-platinum wire.
Each of the three types of wire would have nominal diameter 0.5 mm and be supplied in a continuous length.
The pure platinum wires should have a steam point to ice point resistance ratio not less than 1.3924.
The two alloy wires would contain as closely as possible 10 per cent rhodium and 13 per cent rhodium respectively, rather than have the rhodium contents adjusted to match particular specified e.m.f.s of the gold point. It was generally acknowledged that this would lead to slightly higher values of e.m.f. for given temperatures than in existing tables.
It was decided also to split the experimental work among the three national laboratories in the following way:
The materials would be collected and the thermocouples assembled by NBS, then half of the completed thermocouples would be sent to NRC.
NBS and NRC would perform primary calibrations on some and comparison calibrations on all of their thermocouples from 0°C to the gold point.
A selected number of thermocouples from each of the NBS and NRC groups would then receive primary calibrations at NPL from the gold point to the platinum point against a photoelectric optical pyrometer using a suitable black-body cavity. Enough calibrations would be done at the gold point to ensure agreement with the NBS and NRC calibrations.
The thermocouples retained by NBS and NRC would be intercompared from the gold point to the platinum point and also be compared with those from NPL upon their return.
NBS and NRC would ensure agreement with the NPL by a limited number of high temperature calibrations obtained by measuring palladium and platinum points by the wire method.
The Reference Tables
This work has now been completed and the new reference tables were presented at the 5th Symposium on “Temperature” held in Washington in June 1971.* These new reference tables, unlike the old ones, have been produced by means of agreed sets of polynomial functions fitted to the results of the experimental work. These functions are listed in Tables I and II and skeleton reference tables derived from them appear in Tables III and IV. The differences between the new tables and the old are shown in Figs. 2 and 3, which also clearly indicate the differences between the NBS 561 and BS 1826 tables.
|−50°C to 630.74°C|
|630.74°C to 1064.43°C|
|1064.43°C to 1665°C|
|1665°C to 1767.6°C|
|−50°C to 630.74°C|
|630.74°C to 1064.43°C|
|1064.43°C to 1665°C|
|1665°C to 1767.6°C|
The experimental work carried out at NPL that provided the data for the new reference tables above the gold point was undertaken using the NPL photoelectric pyrometer (Figure 1). This was used to measure the temperature of a black-body cavity which could accommodate up to four thermocouples at a time. The cavity used from the gold point up to 1748°C was made from solid platinum and was loaned to NPL for this work by Johnson Matthey. It is illustrated in Figs. 4 and 5. Using a furnace wound with pure rhodium ribbon and 40 per cent rhodium-platinum wire internal end heaters, a temperature uniformity, at about 1500°C, of within 0.3 deg C was achieved over the whole length of the block.
It was found that the reproducibility of platinum : 13 per cent rhodium-platinum thermocouples was significantly better than that of platinum : 10 per cent rhodium-platinum thermocouples over the whole temperature range. For example, the mean gold point e.m.f. determined by NPL for eight platinum : 13 per cent rhodium-platinum thermocouples was 0.4 microvolts above the NBS and NRC mean ingot value, while that of the platinum:10 per cent rhodium-platinum thermocouples was 2 microvolts higher. A difference in behaviour of this sort can be reasonably accounted for by the fall in slope, between 10 per cent rhodium and 13 per cent rhodium, of the e.m.f./composition curve for rhodium-platinum alloys. It would seem reasonable therefore to hope that in due course the platinum:13 per cent rhodium-platinum thermocouple would supersede the platinum:10 per cent rhodium-platinum thermocouple in general use, particularly if the IPTS-68 between 630.74°C and 1064.43°C is eventually defined in terms of the platinum resistance thermometer rather than the platinum: 10 per cent rhodium-platinum thermocouple.
To cover the range between 1748°C and the melting point of platinum further measurements were made using a black-body cavity made from alumina. It was found with this cavity that there is a significant drop in the thermoelectric power both of platinum:10 per cent rhodium-platinum and platinum:13 per cent rhodium-platinum thermocouples above 1700°C. Figures 6 and 7 show the results of these high temperature measurements of thermoelectric power. The change in slope of the thermoelectric power/temperature curve above about 1100°C can be accounted for qualitatively by the effects of the increasing concentration of lattice vacancies at high temperatures. The drop above 1700°C, however, seems too steep to be accounted for solely by lattice defects; there must be another factor which is becoming important. One such factor could be the conductivity of the alumina refractory which is increasing at a significant rate at these temperatures.
The Freezing Point of Platinum
It became apparent during the course of this work that a temperature of 1772°C (IPTS-68) for the freezing point of platinum would not be consistent with the results of measurements made in the two black-bodies from the gold point upwards. The e.m.f./temperature curve thus obtained showed that the temperature at which the platinum arm of the thermocouple melted was some 4 deg C below 1772°C. A similar result was obtained from platinum wire-point measurements made at NRC. That the freezing point of platinum was lower than the previously accepted value was subsequently confirmed at NPL by measurements made with the photoelectric pyrometer using substantial ingots of pure platinum (4). Three series of measurements were made, two ingots being supplied by Engelhard (U.K.) and one by Johnson Matthey. There was no significant difference found between the results from the three ingots, nor between the melts and the freezes. The final value for the freezing point of platinum was found to be 1767.6±0.3°C (IPTS-68).
The authors are pleased to acknowledge the generous assistance given by Johnson Matthey & Co Ltd, throughout this work by the supply of the platinum and rhodium-platinum wire, the construction and loan of the platinum black-body, the machining and loan of one of the ingots of platinum used for the melting point work, and for spectrographic analysis of pieces of the platinum before and after melting.
Much of the impetus behind this work, together with invaluable advice and encouragement during its execution, came from the late C. R. Barber of NPL.
- 1C. R. Barber, Nature, 1969, 222, 929
- 2W. F. Roeser and H. T. Wensel, J. Res. Nat. Bur. Stds., 1933, 10, 275
- 3C. R. Barber, Proc. Phys. Soc., 1950, B63, 492
- 4T. J. Quinn and T. R. D. Chandler, Metrologia, 1971, 7, 132