Platinum Metals Rev., 1979, 23, (1), 2
Ensuring the Most Advantageous Use of Platinum
Interactions to avoid in Industrial Environments
- By A. G. Knapton
- Group Research Centre, Johnson Matthey and Co Limited
The unique high temperature and corrosion resistant properties of the platinum group metals can best be utilised if the mechanisms which may result in the degrading of these properties are acknowledged and understood. This article reviews the more important of these mechanisms, thus enabling the conditions which promote them to be avoided, and the nobility of the platinum metals to be maintained in everyday practice.
Platinum belongs to the group of aptly called noble metals and possesses all of the properties that this name implies; corrosion resistance, freedom from oxidation and high melting point coupled with good catalytic activity make the metal an invaluable component of several important industrial technologies. Thus major outlets for platinum in bulk form are to be found in the oxidation of ammonia to nitric oxide for use in the manufacture of nitric acid, in glass fibre production (illustrated in Figure 1), in thermocouples and in the melting of optical glass. Lesser but important applications are furnace windings, electrodes and laboratory ware.
Now although everyone normally associates platinum with these desirable noble properties, a number of circumstances can arise in which this nobility may be undermined. It is the purpose of this paper to examine some of these circumstances and to outline the physical and chemical bases for the observed behaviour, illustrated with practical examples. An awareness of these factors may assist in delineating the limitations encountered in the use of platinum and thereby in avoiding frustration by attempting to use the metal beyond its capabilities. In the past many of the observed interactions between platinum and other materials have been poorly understood, but recent systematic studies, coupled with better thermodynamic understanding, have served to rationalise the behaviour of platinum in a variety of environments. It is now probably true to say that, provided the conditions of use are well defined, the service given by platinum equipment can be predicted with reasonable accuracy.
Many structural applications involve the use of platinum alloys with iridium or rhodium, and most of the considerations in this paper apply equally well to such materials.
All materials suffer from a loss in weight to a greater or lesser extent due to volatilisation on heating, and platinum is no exception. Indeed vaporisation of the metal is one of the basic limitations of its use at high temperatures and appreciable losses occur near the upper end of the operating range.
The vapour pressure (P) of platinum metal has been (1) expressed in the form:
and is shown plotted in Figure 2. Rhodium is an important alloying constituent in the platinum alloys used for many applications and data for this metal are also included. In the presence of air or oxygen the rate of platinum loss is considerably enhanced owing to the formation of platinum oxide which, as also shown in Figure 2, is very much more volatile than the metal at elevated temperatures. However, the oxidation phenomena involved are not simple. On heating to relatively low temperatures, of say 400 to 500°C, little visible change occurs on the metal but nevertheless there is good evidence that a very thin film of solid oxide forms on the surface, and this probably has the composition PtO2. This film thickens as the temperature is increased, but continued heating results in the oxide entering the vapour phase. There has been considerable discussion in the literature (2,3,4,5) on the precise mechanism by which this occurs. Chaston (2) suggests that the situation may be rationalised by considering that the oxide undergoes a phase change from solid to gas. From the free energy of formation data (1) on PtO2, (ΔG°=39270−0.93T), the positive ΔG° infers that the formation of the oxide is not energetically very favourable and as a consequence the partial vapour pressure of the oxide is low.
Also as a consequence of the negative entropy term, platinum migration is anomalous in that dissociation of the oxide to metal tends to occur on cooler rather than on hotter surfaces, unlike many other vapour transfer mechanisms. Tangible evidence of this type of transfer will be well known to users of platinum wound furnaces, where it results in the formation of the familiar bright crystalline spangles deposited in the lower temperature regions of the furnace tubes and insulating materials, typical examples being shown in Figure 3. The vapour transfer of platinum and rhodium has particular implications in the stability of thermocouples and other applications that will be described in more detail later. Quantitative data on the magnitude of the volatilisation loss of platinum is available from a number of sources. Little loss occurs on heating in air at temperatures up to 900°C, while values of 9.19 × 10−3 and 39 × 10−3 mg/cm2/h have been reported (6) for temperatures of 1400°C and 1700°C respectively. Significant losses are encountered from the 10 per cent rhodium-platinum gauzes used for ammonia oxidation in the production of nitric acid. The high gas flow rates and elevated temperatures employed in the process result in a slow but continuous volatilisation of platinum and rhodium from the gauze, the losses typically amounting to 50 to 300 mg of metal for each ton of nitric acid produced.
Various methods have been adopted in attempting to reduce volatilisation losses from platinum components. From the above it will be apparent that restricting the oxygen supply should be successful in inhibiting oxide formation. A graphic example of this is described by Powell (5). In continuous glass fibre production he noted that the platinum alloy bushings were normally enveloped in steam during use and underwent very little metal loss. However, when a similar bushing was exposed in air in the laboratory an overall weight loss of about 4.7 per cent had occurred over a period of seven weeks at 1450°C. Low oxygen pressures in the atmosphere are therefore obviously successful in this respect, but great care must be taken to avoid lowering the oxygen potential too far in view of the reactions which may occur with other materials under these conditions.
In general alloying does not have a marked effect in reducing platinum volatilisation unless large quantities of the added component are present. The life of platinum-wound furnaces is extended if the winding is completely embedded in refractory material such that air circulation round the metal is inhibited (5).
Possibly a further improvement would be obtained if the embedding medium was completely non-porous at the operating temperature. For example such a medium could consist of a two-phase refractory, one of the phases being liquid at elevated temperatures. However a system of this type has yet to be developed.
Platinum losses by volatilisation processes therefore cannot be ignored, and where accurate measurements are necessary these need to be taken into account. Thus crucible losses after high temperature usage may be appreciable and suitable compensation must be made for losses from the platinum suspensions if used in a thermobalance or similar equipment.
Interaction with Other Elements
Perhaps the next simplest type of effect to consider is the interaction of platinum with other individual elements. The first successful fabrication of platinum involved the addition of arsenic to the metal, the arsenic subsequently being removed by heating at high temperatures. However, in general, low melting point metals of this type cause very rapid grain boundary embrittlement of platinum ware. Among these we may list antimony, selenium, tellurium, lead, tin, zinc and aluminium. Most molten metals alloy with and dissolve platinum with relative ease, and the fusion of any metallic element in a platinum crucible is to be avoided. In the handling of laboratory ware, care should be taken to ensure that contamination does not occur. Crucibles should be handled with platinum tipped tongs and such actions as placing a hot crucible on a galvanised steel surface must be avoided. In a similar incident a mysterious weld cracking problem was solved when it was noted that the operator placed the hot platinum filler rod he was using on an aluminium-topped bench between welding passes.
Although not strictly a low melting point metal, silicon is also a troublesome element in platinum, and consideration of the constitutional relationships in platinum-silicon alloys shows the basis for this. The phase diagram for the system is given in Figure 4. Silicon will be seen to reduce drastically the melting point of platinum, to a minimum of 830°C at the eutectic composition between platinum and Pt5Si2. The practical implication of this is that even at low silicon contents a liquid phase will be present in the platinum at its operating temperature and lead to premature failure. Siliceous materials are found in a wide range of minerals such as sand and concrete, and may be readily converted to elemental silicon under reducing conditions. Thus platinum thermocouple failures have been traced (7) to excessive silicon contents in refractory sheaths or insulators, or an incidence of cracking in argon-arc welds shown to be due to the pick-up of oily siliceous particles on the platinum alloy sheet. Indeed, even pointing a platinum welding electrode on a carborundum (SiC) wheel has proved to be sufficient to cause embrittlement due to silicon pick-up (9). Scrupulous care is therefore necessary during the fabrication and subsequent handling of platinum to avoid contamination by siliceous material, and also by other common particulate workshop materials such as aluminium, solder and brass.
Among the category of lower melting point metals, a frequent offender in the glass fibre industry is silver. It is common practice to cool the glass fibres emerging from the platinum alloy bushings by silver cooling fins placed between the rows of nozzles. Of necessity these fins must be positioned in close proximity to the baseplate of the bushing and contamination may occur either by accidental transfer of silver during installation or from distortion of the plate during service. Rapid penetration of liquid silver occurs along the platinum alloy grain boundaries and results in failure of the bushing.
An example of this type of failure is illustrated in Figure 5, obtained from the electron probe microanalysis of a sample cut from a failed 25 per cent rhodium-platinum alloy bushing. The incipient grain boundary cracking shown in Figure 5(a) is clearly associated with the presence of silver in this region, Figure 5(b), discrete particles of silver also being detected at the surface of the alloy. The operating temperature of about 1400°C is well above the melting point of silver (960.8°C) and the presence of the liquid film means that grain boundary separation will occur at virtually zero tensile stress on the metal. It is not difficult to imagine that grain boundary attack of this type will be reduced if the contaminant is spread over a much larger grain boundary area, or in other words if the metal being attacked has a much finer grain size.
Considerable evidence has now been accumulated showing that the use of the ZGS dispersion strengthened forms of platinum (10) or 10 per cent rhodium-platinum (11) for bushing base plates can reduce appreciably the incidence of this problem. In addition to this resistance to attack, the good high temperature mechanical properties of these materials result in less base plate distortion during operation.
Besides reacting readily with molten metals, a number of common chemicals used for fusions in chemical work attack platinum ware to a greater or lesser extent. These will not be discussed in detail but some of the relevant data (12, 13, 14) are accumulated in the Table.
Reactions with Refractory Oxides
One facet of the behaviour of platinum that does not appear to be fully appreciated by many users is the extreme readiness with which the metal reacts with a very wide range of otherwise highly stable substances when the atmosphere has a low oxidising potential. These reactions often lead to considerable depression of the melting point of the metal by the formation of eutectic compositions, while they may affect thermoelectric properties. Platinum is nearly always used at high temperature in contact with some type of stable refractory oxide, and it is of importance to recognise the conditions in which such reactions can occur.
Consider a reaction of the type:
xPt + yMO → (PtxMy) + yO
The tendency for this reaction to proceed from left to right will depend on several factors, including the stability of the oxide MO, the stability of the alloy phase formed—which may be an intermetallic compound or a solution of M in platinum—and whether the oxygen formed is removed from the system. In more formal language, platinum will react with the oxide providing the free energy of formation (−ΔG) of the alloy phase is sufficiently large and the oxygen potential in the atmosphere is low.
The surprisingly high stabilities of certain platinum-containing intermetallic phases and the associated low activities of the base metals in solution in the platinum have been abundantly demonstrated by experimental observations over the past few years and are the subject of continuing thermodynamic investigations (15–20). These high stabilities mean that the reaction considered above will often proceed when the oxygen potential is not particularly low, and in conditions easily obtained from mildly reducing or vacuum environments. Although the deleterious effect of reducing atmospheres on platinum ware, thermocouples and the like had long been recognised, it was perhaps Bronger and Klemm (15) who first made deliberate use of the phenomenon for the preparation of a wide range of intermetallic compounds containing the platinum group metals. In these investigations the platinum metal was reacted with the oxide of the second metal at high temperatures in a hydrogen atmosphere, that is the oxygen potential in the atmosphere above the reactants was kept at a low value by the reaction to water vapour. By this means such stable oxides as the alkaline earths, Al2O3 or Sm2O3 were reduced with the formation of compounds such as Pt3Al or Pt3Sm. The objective of this work was primarily to prepare small samples for the identification of the crystal structure of the compounds, but additional thermodynamic information was obtained by Wengert and Spanoudis (18) from analysis of the experimental data reported.
Some of the more practical implications of the platinum-refractory oxide interactions were the subject of an investigation by Darling, Selman and Rushforth (21) in which reactions relative to instabilities in thermocouple materials were studied in some detail. These authors embedded self-heated platinum thermocouples in beds of refractory oxide powder in glass test cells, shown in Figure 6, which could be evacuated or provided with a controlled atmosphere. It was found that, when a low oxygen potential was maintained in the cells either by evacuating or by gettering an argon atmosphere, severe reactions were encountered between platinum and alumina, thoria or zirconia, often with the formation of liquid metallic phases at the test temperature. In the same conditions, however, magnesia (MgO) remained almost completely inert, very little reaction being detected after 450 hours at 1650°C. By comparison a similar experiment in alumina resulted in complete destruction of the couple after only a few hours. More recent investigations by Ott and Raub (22, 23) have included a wider range of oxides, reducing agents, and other platinum group metals. The presence of the metal was found to enhance the ability of carbon monoxide, hydrogen, carbon or organic vapours to reduce the more stable refractory oxides, reactions being observed at temperatures as low as 600°C. Once again the observations may be rationalised on the basis of the high affinity of the platinum metals for the metal of the refractory oxide.
Although for the purposes of this article the drawbacks of interactions of this type have been described, it should be emphasised that the stable intermetallic compounds containing the platinum metals are largely unexplored in their properties and could have many potential applications. A review of the possibilities together with a consideration of the theoretical basis for the stability of these compounds is given in a recent paper by McGill (24).
It should also be emphasised that providing the conditions are maintained strongly oxidising virtually no reaction between the platinum and the refractory can be detected. Thus platinum is regularly melted in zircon crucibles (ZrSiO4) with no detectable pick-up of either zirconium or silicon (25).
Some Further Examples
Much of the behaviour of platinum at high temperatures in practice can be ascribed to the mechanisms already outlined, either singly or in combination. A few further illustrations may serve to emphasise this conclusion.
When glass has been melted in platinum ware it normally contains a small number of platinum inclusions. These are usually innocuous, but when they occur in high power laser glass, fracture may result during operation. Of the various possible methods of entering the glass, the vapour transport of platinum oxide from the crucible walls followed by reduction in the melt has been demonstrated as the most important contaminating mechanism (26, 27). Melting is frequently carried out in a nitrogen atmosphere, but oxygen may arise from dissociation of the more volatile components of the melt. Contamination has been successfully reduced by the use of glass formulations containing the more stable alkali or alkaline earth ions, and by limiting the area of platinum exposed to the atmosphere above the melt. Other systems resistant to platinum contamination are based on fluorophosphates as described by Hornyak and Abendroth (28). The lower melting temperatures in such systems, and changes in other process parameters, were successful in reducing the platinum contamination to a low level.
The premature failure of platinum components used for such operations as glass melting or fluxing is almost invariably the result of some deleterious contaminating element being liberated in the melt by some inadvertent combination of conditions. Analysis of such failure is often frustrated by the fact that the source of the failure may be removed by subsequent oxidation or washing away by the melt. Nevertheless the electron probe microanalyser has proved to be an invaluable tool in investigating such failures, examples of its use in the glass industry being described by Jewell (29). Two similar failures examined recently in the Johnson Matthey Laboratories are illustrated in Figures 7 and 8. In the former, silicon and arsenic reduced to elemental form from an optical glass have caused local melting of the platinum component. Figure 8 shows grain boundary contamination, by lead, of a crucible used in the melting of a high lead content glass.
Metal Clad Thermocouples
A more complex interplay of factors may be recognised in the behaviour of metal clad thermocouples. When first introduced, such couples tended to be thermoelectrically unstable compared with their refractory sheathed counterparts. However substantial improvements have been made, mainly as a result of systematic investigation (30) of compositional changes in the thermocouple wires and thence a better appreciation of the metal transfer mechanisms involved. The importance of volatilisation processes was well illustrated in a variety of experiments in which the effect of a number of parameters affecting the transfer were examined. As previously mentioned both platinum and rhodium form volatile oxides, and the presence of these oxides within the sheath was shown to result in rhodium transfer to the platinum limb. Considerable improvements were obtained by removing the air from the sheath by evacuating before sealing, thereby inhibiting oxide formation, but even better stability was achieved by then introducing an inert gas to limit metal vapour transfer.
Another important factor was shown to be the quality of the insulant. Careful production procedures used in the manufacture of these couples now result in a dense, defect free, barrier which prevents vapour transport. Magnesia is used for this application because of the absence of reaction with platinum even under low oxygen potential conditions.
Even more stable thermocouple assemblies result by the use of a 6 per cent rhodium-platinum versus 30 per cent rhodium-platinum sheathed in 5 per cent rhodium-platinum alloy, illustrated in Figure 9. In such a combination minor losses from the positive limb (30 per cent rhodium) have little effect on thermal e.m.f. while composition changes in the negative limb are substantially avoided.
As a contrast to a thermocouple made from compatible materials, the tantalum-clad alumina-insulated assembly investigated by Kollie, Christie and Anderson (31) may be cited. Tantalum is known to be a good getter and as a result a low oxygen potential was maintained in the sheath. This allowed reduction of the alumina to occur with the formation of aluminium-platinum intermetallic compounds and consequently large and rapid decalibrations of the couples on heating above 1100°C. Similar effects were subsequently described in platinum-rhodium couples sheathed in Inconel (32).
It is hoped that this account has served to highlight some of the causes of difficulties associated with the use of platinum. Undoubtedly the most important single factor in promoting platinum failures is associated with reducing conditions. If platinum is in contact with easily reducible oxides, for example PbO or As2O3, even a temporary departure from an oxidising atmosphere can lead to low melting point phases and embrittlement. Such conditions may arise from luminous flames, carbonaceous contamination or non-stoichiometry in the compounds being melted. Even refractory oxides may react with platinum at high temperatures under reducing conditions, owing to the high stability of the platinum alloys formed. Attack by low melting point metals also occurs readily, and in certain applications the volatilisation of platinum oxide must be taken into account.
If all of these factors are considered when applications are designed then adverse reactions can be avoided, and the intrinsic noble advantage of platinum realised to the full.
- 1A. S. Darling, Internat. Met. Rev., 1973, 18, 91
- 2J. C. Chaston, Platinum Metals Rev., 1975, 19, (4), 135
- 3C. B. Alcock and G. W. Hooper, Proc. Roy. Soc., 1960 (A), 254, (1279), 551
- 4C. B. Alcock, Platinum Metals Rev., 1961, 5, (4), 134
- 5A. R. Powell, Platinum Metals Rev., 1958, 2, (3), 95
- 6C. J. Raub, Metall, 1978, 32, (8), 802
- 7H. E. Bennett, Platinum Metals Rev., 1961, 5, (4), 132
- 8M. Hansen and K. Anderko, “ Constitution of Binary Alloys ”, McGraw-Hill, New York, 1958
- 9V. V. Stepanov and V. A. Dimitriev, Svar. Priozvod., 1977, (12), 10
- 10G. L. Selman,, J. G. Day and A. A. Bourne, Platinum Metals Rev., 1974, 18, (2), 46
- 11G. L. Selman and A. A. Bourne, Platinum Metals Rev., 1976, 20, (3), 86
- 12D. Menzel, Prakt. Metallogr., 1976, 13, (5), 211
- 13J. M. Robertson,, J. P. M. Damen,, H. D. Jonker,, M. J. G. van Hout,, W. Kamminga and A. B. Voermans, Platinum Metals Rev., 1974, 18, (1), 15
- 14J. M. Robertson, Engelhard Ind., Tech. Bull., 1969, 10, (3), 77
- 15W. Bronger and W. Klemm, Z. Anorg. Allgem. Chem., 1962, 319, (1/2), 58
- 16W. Bronger, J. Less-Common Metals, 1967, 12, (1), 63
- 17H. Schulz,, K. Ritapal,, W. Bronger and W. Klemm, Z. Anorg. Allgem. Chem., 1968, 357, 299
- 18P. R. Wengert and L. Spanoudis, J. Am. Ceram. Soc., 1974, 57, (2), 94
- 19P. J. Meschter and W. L. Worrell, Met. Trans. A, 1976, 7, (2), 299
- 20P. J. Meschter and W. L. Worrell, Met. Trans., A, 1977, 8, (3), 503
- 21A. S. Darling,, G. L. Selman and R. Rush-forth, Platinum Metals Rev., 1970, 14, (2), 54 ; 1970, 14, (3), 95 ; 1970, 14, (4), 124 ; 1971, 15, (1), 13
- 22D. Ott and C. J. Raub, Platinum Metals Rev., 1976, 20, (3), 79
- 23D. Ott and C. J. Raub, Metall, 1978, 32, (1), 35 ; 1978, 32, (2), 140
- 24I. R. McGill, Platinum Metals Rev., 1977, 21, (3), 85
- 25J. Cochrane, Engelhard Ind., Tech. Bull., 1970, 11, (2), 58
- 26J. C. Chaston, Platinum Metals Rev., 1974, 14, (4), 122
- 27R. J. Ginther, J. Non-Cryst. Solids, 1971, 6, (4), 294
- 28E. J. Hornyak and R. P. Abendroth, Paper to American Ceramic Soc. Meeting, Detroit, 1978
- 29R. C. Jewell, Platinum Metals Rev., 1964, 8, (4), 122
- 30G. L. Selman and R. Rushforth, Platinum Metals Rev., 1971, 15, (3), 82
- 31T. G. Kollie,, W. H. Christie and R. L. Anderson, J. Less-Common Metals, 1978, 57, (1), 9
- 32W. H. Christie,, T. G. Kollie,, R. E. Eby and R. L. Anderson, J. Less-Common Metals, 1978, 59, (1), 17