Furnace Design

Fig. 1 illustrates diagramatically the furnace arrangement adopted. A rhodium-platinum tube is heated to the desired temperature by the direct passage of a heavy alternating current fed in through the two pillars supporting the top water-cooled bridge piece, and leaving through the copper base plate which is maintained at earth potential. The tubular heater, approximately 13 inches long and 1 inch diameter, has a central slot ¼ inch wide, through which the specimen can be observed. Each end of this heater is cut into a number of ⅛ inch wide strips which are formed into a corrugated ‘fringe’ to facilitate clamping and permit thermal expansion and contraction without undue strain.

The test specimens are ⅛ inch diameter drawn rod. They hang down the axis of the heater and are directly loaded by weights as shown in the diagram. Thin strips of platinum tape are pressure welded to the specimen to define the 3 inch gauge length and to provide readily identifiable reference marks upon which the measuring microscope can be focused. As shown in Fig. 2, thermocouples are attached to the top, bottom and centre of the gauge length, and temperature uniformity is maintained by using the independently controlled booster heater. This adjustable heater is positioned 3 to 4 inches above the base plate of the furnace and by suitably controlling its power the temperature gradient along the gauge length of the specimen can be reduced to 5°C at 1500°C.

The complete heater assembly is normally surrounded by aluminous insulating bricks which are slotted in front so as not to interfere with microscopic observation. Fig. 3 shows the heater assembly with the front refractory slabs removed. In this photograph the slot in the heater has been opened up so that a better view of the test piece can be obtained. The low voltage current supply transformers for the main and booster heaters are fed from a constant voltage transformer which supplies three furnaces. The use of the travelling microscope for strain measurement is illustrated in Fig. 4. This measuring equipment, centrally pivoted to enable it to be used with three furnaces, permits elongation determinations to an accuracy of ±0.0001 inch.

Experimental Procedure and Results

The batch of 10 per cent rhodium-platinum rod used for this series of experiments contained the following impurities:

Of the total impurity content, 63 per cent is accounted for by platinum metals, and this material can be considered typical of that generally used for high temperature work.

The gauge marks and thermocouples were attached to the test pieces when these were in the work hardened condition. The complete assembly was then fitted into a cold furnace, slowly heated and steadied at the desired temperature for 16 hours before any load was applied. This treatment ensured that all tests were made on fully recrystallised material. Elongation measurements commenced immediately after the application of load and continued until the specimen had either broken or resisted fracture for at least 1,000 hours.

Some typical creep curves are shown in Fig. 5 and 6. The only test pieces that exhibited a steady creep rate were those stressed below 500 pounds per square inch at 1200°C or 250 pounds per square inch at 1500°C. At higher stresses the effects of primary and secondary creep could scarcely be detected and the curves all indicated a rapid increase of creep rate with time.

All the specimens tested at 1200°C failed in an intercrystalline manner as illustrated in Fig. 7. At 1400°C the specimens exhibited high ductility and necked down to a needle point with no sign of intercrystalline cracking. Some typical micro-sections are illustrated in Fig. 8. The experimental results are summarised in Table I, which shows that the transition from necking to intercrystalline failure occurred between stresses of 750 and 1,000 pounds per square inch at 1300°C. This transition was particularly well defined, and the section through the specimen tested at 750 pounds per square inch and illustrated in Fig. 9 (b) shows no signs of necking. At 1300° and 1400°C the elongations at fracture increased with increasing stresses, and those specimens forced to creep rapidly did not seem prone to intercrystalline failure.

Each experimental point on Fig. 10, which summarises the stress rupture data obtained during this series of tests, represents the mean of at least two, and possibly three determinations. The dotted line suggests the approximate boundary between the regions of transcrystalline and intercrystalline failure.

Discussion of Results

The elongation measurements obtained during this series of tests make it possible for the designer to specify a working stress which will keep distortion down to tolerable limits during the operational life of high temperature equipment. Table II gives the tensile stresses required to produce permanent elongations of 1, 5 and 10 per cent in times ranging up to 1,000 hours.

Creep elongations of the sort reported here are, of course, much higher than those usually considered in conventional engineering, where designers might be interested, for example, in keeping extensions less than 0.1 per cent for 10,000 hours or more. Rhodium-platinum alloys are, however, the only known materials of any ductility capable of resisting significant tensile stresses in oxidising conditions at temperatures above 1250°C. They are usually operated under fairly steady load conditions where a good deal of distortion is tolerable providing failure does not occur, and a high elongation before fracture is therefore a matter of primary concern.

The sudden transition to intercrystalline failure below 1300°C does not indicate that the 10 per cent rhodium alloy is unsuitable for use in this temperature range, where it has in fact been used industrially for many years with considerable success. Most of the reported failures have been due to contamination with some material which reacts with the alloy to form a liquid phase. When intercrystalline failure was observed during the present series of tests, the fracture elongation was never less than 24 per cent, and it is certain that any item of industrial rhodium-platinum high temperature equipment would be taken out of service before local deformations of this magnitude occurred.

The sudden change in the mode of failure is of theoretical rather than practical significance and it is interesting to speculate on the principles involved. Grain boundary failure is generally associated with the presence of harmful impurities which might, at high temperature, be either molten or considerably weaker than the parent material. This explanation does not account for the disappearance of the effect above 1300°C or indicate why the tendency towards intercrystalline failure decreases with increasing stress.

When the effects of impurities can be safely neglected grain boundary cavitation is usually related to the extent of grain boundary slide. The growth of cavities depends upon diffusion processes along the grain boundaries, and Blackburn and Brown (2) have shown that oxidation at the specimen surfaces produces grain boundary cracks rather than isolated cavities. A detailed mechanism for the failure of rhodium-platinum cannot be postulated on the evidence available, although it must obviously account for the increasing tendency towards grain boundary failure as the stress is reduced. The solution to this problem is probably dependent upon the mode of deformation inside the grains.

The boundary between the regions of intercrystalline failure and ductile necking has been tentatively indicated in Fig. 10. To substantiate the position of this line additional tests at higher stresses and lower temperatures will be required. The effect of trace impurities requires careful consideration and the possibility of oxygen diffusion along the rhodium-alloy grain boundaries cannot be neglected. A great deal of work will be required to permit the relative importance of factors such as these to be properly evaluated.