Collector-metal behaviour in the recovery of platinum-group metals from catalytic converters

A commercial pyrometallurgical process for the extraction of platinum-group metals (PGM) from a feedstock slag was analysed with the use of a model based on computational fluid dynamics. The results of the modelling indicate that recovery depends on the behaviour of the collector phase. A possible method is proposed for estimation of the rate at which PGM particles in slag are absorbed into an iron collector droplet that falls through it. Nanoscale modelling techniques (for particle migration or capture) are combined with a diffusion-controlled mass-transfer model to determine the iron collector droplet size needed for >95% PGM recovery in a typical process bath (70 mm deep) in a realistic time-scale (<1 h). The results show that an iron droplet having a diameter in the range 0.1–0.3 mm gives good recovery (>90%) within a reasonable time. This finding is compatible with published experimental data. Pyrometallurgical processes similar to that investigated should be applicable to other types of waste that contain low levels of potentially valuable metals.

The recovery mechanism of platinum group metals from catalytic converters in spent automotive exhaust systems

The recovery of platinum group metals (PGMs) from catalytic converters of spent exhaust systems is considered in this paper. To be cost-effective, recovery processes must be well over 90% efficient and so the optimisation of their operation is vital. Effective optimisation requires a sound understanding of the operation and the underlying process mechanisms. This paper focuses on pyrometallurgical recovery operations used and typified by the Johnson–Matthey process. Analysis of this process reveals that it cannot be simply explained by the gravity model that is normally assumed. The analysis reveals that the affinity of PGM particles for the melted collector metal is a key factor in the behaviour of the process. A rational explanation of the key issues that govern the process behaviour is proposed and shown to be consistent with available operational data. The results generated would be applicable to other similar processes.