Observation of the Surfaces

The results of exposing rhodium-platinum gauzes to 6, 8, 10 and 12 per cent ammonia-air mixtures in the miniature reactor for fifteen hours are shown in Figure 2. The reconstruction process occurs only over a limited range of gas mixtures. It is particularly interesting to note that at 6 per cent ammonia, the catalytic reaction virtually stopped after only four hours, leaving a herringbone-type of structure on the surface. After operation in 12 per cent ammonia, the gauze surface was smooth with angular holes and cavities. Only at intermediate ammonia concentrations did extensive surface reconstruction occur. All the effects occurred at slightly lower ammonia levels than in a commercial plant, apparently due largely to differences in heat dissipation conditions in the miniature reactor.

The nucleation of reconstruction was studied using a shorter exposure time of two hours. Figure 3 shows the role of recrystallisation and of grain boundaries. The as-received gauze shows surface markings due to wire drawing. However, during pretreatment with a hydrogen torch the gauze recrystallises from its drawn texture, and grain boundary grooves become clearly visible. Following exposure for two hours in the reactor in 10 per cent ammonia, deep cavities develop along grain boundaries, while similar exposure in 13.5 per cent ammonia results in a smooth more rounded surface.

The progressive development with time of the cage-like growths, including the encroachment into grain interiors, can be seen in Figure 4, which shows the effects of exposure for 0.5, 1, 2 and 4 hours in an air 8 per cent ammonia mixture. After only half an hour the previously smooth recrystallised surface shown in Figure 3 (b) has started to reconstruct, Figure 4 (a), this seeming to occur most extensively on some preferred-orientation grains. Figure 4 (b) shows the build up of the growths on these crystals. It is interesting to note the part played in propagating the reconstruction into the grain interiors by the intersection of thermal facets and grain boundary grooves. Figures 4 (c) and 4 (d) show clearly that the reconstruction spreads out from these intersection points. The surface is finally completely covered by the cage-like growths as in Figure 2 (b).

The larger pilot plant was used to compare pure platinum and rhodium-platinum alloy catalyst gauzes. This allowed accurate observation of the two types of catalyst materials under identical operating conditions. The pilot plant gave the same type of reconstruction for a given gas mix as a commercial plant, with a rate of reconstruction somewhat faster than that of a commercial plant but slower than for the miniature laboratory reactor. The reconstruction in the case of pure platinum gauzes was significantly faster than that for identically treated alloy gauzes. For example, Figures 5 (a) and 5 (b) show the microstructures observed on the two types of gauze after a two hour exposure to a 10 per cent ammonia mixture. It is possible that the presence of the rhodium in the surface layer of the catalyst restricts the rate of platinum evaporation, so reducing vapour transport processes. This is considered below.

Discussion

The high vapour pressure of PtO₂ compared to that of platinum (10), along with the observations of weight losses from operational gauzes—and hence the need for catchment gauzes (11)—indicate that there is substantial PtO₂ volatilisation in the ammonia oxidation process. A vapour phase mechanism for reconstruction is therefore likely, as first proposed by Schmidt and colleagues (12, 13). We have developed this model further, to explain the microstructural features of reconstruction and their dependence on conditions such as gas composition and temperature (14). Our proposed mechanism is illustrated diagrammatically in Figure 6. Consider the boundary between two grains, where a groove will develop on heating simply due to surface tension effects. The feed gas contains excess oxygen and the catalytic reaction is very efficient, so most of the ammonia will be oxidised on initial contact with the outer surface of the gauze. The gas reaching the bottom of the grain boundary groove will therefore be more strongly oxidising than the gas at the surface. Platinum oxide will form here preferentially, and will diffuse outwards, to where it meets a more reducing atmosphere, and so platinum will be deposited on the outer surface. Thus the grooves will deepen, and the surface will build up, resulting in the hollow cage-like growths seen on reconstructed gauzes, the process being driven by gradients in oxygen potential. The intersections between thermal facets on the grain surfaces and the grain boundary grooves form particularly potent sites for the nucleation and development of the surface growths, since these are the most deeply recessed points on the gauze surfaces.

To understand why the reconstruction occurs only over a limited range of temperature and gas mixtures we have to consider the main physical processes taking place on the gauzes:

The extent of each process is dependent on the reaction conditions, the vapour transport process dominating over only a limited range of these conditions. If the temperature is too low (low ammonia content in the gas stream), there is insufficient PtO₂ vapour pressure so the dominant process is the selective oxidation of rhodium to form Rh₂O₃. Under these conditions, the catalyst is rapidly deactivated. If the temperature is too high (high ammonia content), then surface and bulk diffusion are greatly enhanced, leading to a smoothing out of the surface features and a consequent large reduction of surface area. The lower rate of reconstruction observed in the case of rhodium-platinum gauzes suggests that the partial vapour pressure of PtO₂ over the alloy is lower than over pure platinum. This would help to explain the reduced weight losses observed commercially when alloy gauzes are used.

It is interesting to consider how the ideas developed above might be applied to the case of hydrogen cyanide production. In this case the gauze temperatures are 200 to 300°C higher than in the ammonia oxidation reaction, but the equilibrium partial pressure of oxygen in the gas mixture is many orders of magnitude lower. The weight losses from the gauzes are much smaller in comparison. The reconstructions extend much deeper into the gauze, and gauze spaces are eventually blocked during use. Two mechanisms of reconstruction seem possible in this case. First, as the vapour pressure of PtO₂ is orders of magnitude higher than that for platinum (10), even a very small partial pressure of oxygen could produce localised oxide vapour transport. Any volatile oxide formed would be decomposed after travelling only a very limited distance, so metal loss would only be due to direct metal volatilisation. Secondly, there is the possibility of grain boundary and surface diffusion controlling the movement of material, possibly involving the migration of a Pt-O complex. As in the case of the ammonia oxidation process the driving force for reconstruction must be the localised variations in oxygen potential.

We now have a clearer understanding of the structural changes occurring on rhodium-platinum gauzes, and the mechanisms involved. The next stage is to look at the surface composition changes involved in more detail. This will be the subject of a later paper.