Platinum Group Metals Recovery Using Secondary Raw Materials (PLATIRUS): Project Overview with a Focus on Processing Spent Autocatalyst

TECNALIA developed a two-stage leaching and reduction process using deep eutectic solvents (DES), comprising choline chloride and oxalic acid (Figure 2). The solvent is first used to leach the pgms from the solid feedstock at a temperature <90ºC, followed by a reduction of the pgms to their metallic form, by heating the solution to >100ºC. Centrifugation was employed at this laboratory scale based on equipment availability, separately both for the removal of the depleted autocatalysts substrate and the pgm product. Recovered DESs can be recycled for the next leaching batch.

Fig. 2

The solvents used are readily prepared from renewable, non-toxic and naturally occurring chemicals when compared to traditional hydrometallurgical processes that employ strong acids (15, 31). Another advantage is that it is a one-pot leaching and reduction process where no additional reducing agent is required. An increase in the pgm purity is achieved from the selective reduction process with a highly pure (90–95%) pgm solid generated, a significant improvement from the low content feedstock. Therefore, it is a simple process for the concentration of pgms, with reduced costs and wastes generated, compared with the cementation process (31–33) and with a very low energy consumption compared to existing pyrometallurgical processes (15). Further separation steps are needed to separate the three pgms from each other. The DES must be kept close to the leaching operation temperature to avoid handling issues arising from DES solidification.

An advanced leaching process using microwave (MW) technology has been developed by VITO using a laboratory MW digester (flexiWAVE, Milestone Srl, Italy). The feedstock is contacted with 6 M HCl and H₂O₂, as an oxidising agent, before being heated using MW radiation (2.45 GHz) to the reaction temperature of 150ºC. MW heating allows short leaching times, consisting of 15 min heating with 10 min dwell time at the reaction temperature (34). The resultant leachate, containing the pgms, is filtered, to remove the undissolved catalyst substrate, leaving an aqueous solution containing the pgms in a chloride matrix.

The MW heating promotes fast, homogeneous volumetric heating enhancing leaching reproducibility and efficiency, with a positive impact on the energy efficiency. Optimisation of the leaching process has resulted in a significant reduction in the HCl acid concentration (i.e. 6 M HCl) required compared to previous results reported in literature (i.e. 12 M HCl) (34). The necessity of the H₂O₂ addition is dependent on the feedstock (Table I, data provided in presence of 10 v/v% H₂O₂ (31%)). In some circumstances its addition has been shown to have a significant impact on the pgm leachability. Through modelling and analyses of the reactor head-space gas, it was shown that H₂O₂ addition increased the formation of Cl₂ and H₂ gas, which must be considered in the process safety assessment.

Solvometallurgy is an alternative branch of metallurgy that uses non-aqueous solutions (35). KU Leuven investigated the oxidative dissolution of pgms from the feedstock using two different solvometallurgical approaches. In the first, an innovative method for the selective leaching of palladium from the feedstock was developed, avoiding the co-extraction of platinum and rhodium whose recovery can be achieved from the palladium depleted feedstock by varying the process conditions. Dilute and concentrated solutions of FeCl₃ in acetonitrile are used for the selective dissolution of palladium and the dissolution of platinum and rhodium, respectively. This solvometallurgical approach provides a preconcentration of the pgms after which further purification is needed to separate platinum and rhodium. In the second approach, highly concentrated solutions of AlCl₃·6H₂O and Al(NO₃)₃·9H₂O were used to dissolve palladium selectively from spent autocatalysts; 95% palladium was leached in only 15 min at 80ºC (36).

Ionic liquids (IL) are perceived as ‘green solvents’, mainly due to their low volatility, which signifies a reduction in the degree of negative environmental impact and health hazards, low toxicity and non-flammability (37). In metal extraction applications, they have provided dramatically higher extraction efficiencies than commonly used solvents. They can act either as solvents in the presence of an extracting agent or as selective extractants, since via modifications in their anion or cation their selectivity towards a certain metal can be tuned (38, 39).

VUT explored the properties of ILs for both leaching and separation processes (for separation see Section 2.5). Leaching of the pgms from the feedstock was performed using hydrophilic and low cost choline-based ILs. The choline-based IL leaching process was selected as the optimum due to its high extraction efficiency and selectivity. The process operates at mild conditions in the presence of an oxidising agent: <100ºC for 4 h in a sealed vessel with continuous stirring (Figure 3).

Fig. 3

This process leaches the pgms, but further separation steps are needed to separate the three pgms from each other. Depending on the feedstock, the loaded IL can be re-used for subsequent leaching batches until the IL capacity is reached. Of note is the mild process conditions used.

The major drawback is the high IL viscosity, making implementation on an industrial scale challenging. Nevertheless, dilution of the IL with water is feasible without compromising the pgm extraction efficiencies. Furthermore, larger scale system design and operation must ensure the violent exothermic reaction between the feedstock and oxidising agent is managed safely.

Due to the increased interest in ILs, a new concept surfaced: IL immobilisation on solid support materials, which is the deposition of a fine IL layer on a solid surface. Use of supported IL phases (SILPs) is an ideal strategy in order to make use of the benefits of IL and simultaneously avoid the inherent complications of IL-based separations such as mass-transport limitations and excessive usage of the IL (40).

VUT exploited the properties of hydrophobic ILs in both liquid and solid extraction processes; this was employed in conjunction with the VUT ionometallurgical leaching process (Section 2.4). In the liquid-liquid extraction process, a solution comprising 50 wt% phosphonium-based IL in n-heptane was used to extract palladium, platinum and rhodium from the loaded pgm hydrophilic leachate (generated from ionometallurgical leaching) to the hydrophobic IL phase. Quantitative extraction is obtained after continuous stirring for 2 h at room temperature. The pgm-free hydrophilic IL phase can be recycled to perform the next batch of IL leaching. No significant performance loss of the hydrophilic ILs was observed between the different cycles; only five cycles have been tested to date. This is particularly attractive to minimise environmental impact and operating costs.

The alternative approach, solid extraction, relied on SILPs using hydrophobic phosphonium-based ILs (Figure 4). The pgms are adsorbed onto the solid enabling their one-step separation from the main impurities of aluminium, iron and cerium due to their low retention on the solid material.

Fig. 4

The stripping is a two-step process: (a) an acidified thiourea solution strips the impurities retained on the SILP; (b) a more concentrated acidified thiourea solution strips palladium and platinum (rhodium is retained on the solid).

The use of SILPs allows fast and simple separation of the pgms from other metals with reduced chemical reagent consumption compared to liquid-based separations. There is the possibility to re-use the solid material for further separations without any loss in its retention and separation performance, to date one recyclability experiment has been conducted. An advantage of SILPs compared to most commercially available resins is no pre-equilibration is required; this could have significant impact on cycle times and effluent generation if re-equilibrium is necessary, as part of the load-elution cycle. A drawback of the process is the retention of rhodium alongside chromium on the SILP and the removal requires a suitable stripping agent not yet identified.

A split-anion solvent extraction process has been developed by KU Leuven for the separation of an aqueous mixture of pgms into their individual elements (Figure 5) (41). Generally, the split-anion extraction relates to the solvent extraction process where different anions are present in the aqueous and organic phases and the distribution of the IL anions strongly favours the IL phase. The pgms are extracted from the chloride leaching solution using the iodide form of the quaternary ammonium IL Aliquat^® 336, [A336][I], dissolved in p-cymene. The iodide anions, which have a strong affinity for the organic phase, coordinate with pgms to form stable iodo-complexes that can be extracted to the ionic liquid phase (42, 43). The split-anion extraction allows not only efficient extraction of pgms without changing from a traditional chloride feed solution, but also the selective recovery of the extracted metal complexes from the loaded organic phases. The organic is scrubbed and stripped of its pgms, as detailed in Figure 5.

Fig. 5

The ionic liquid-based split-anion extraction process is simple, selective and effective for the sustainable separation of pgms, using only one ionic liquid [A336][I] as the extractant, which can be regenerated for consecutive extraction-stripping cycles. The high viscosity of [A336][I] is a drawback, which has shown, during its pilot scale application in mixer-settlers, to slow the mass transfer. Some measures have been identified to reduce the IL viscosity, such as the use of water-saturated ionic liquid or use of green diluents (i.e. p-cymene), but the measures can only partially resolve the problem. To the best of our knowledge this is the first time that a process based on IL for the separation of pgms is tested in continuous mode using real pregnant leach solutions as feed. Other separations using IL have been developed and show good performance but have been only tested with synthetic solutions and not in continuous mode (44, 45).

Sorbents are an established method to selectively recover palladium and other noble metals for acidic aqueous streams such as leachates (16, 46–48). Many materials have been developed by different groups and their performance tested in powder form. However, to be applied in a continuous way, the shape has to be optimised to avoid clogging and pressure build-up in a column or the material has to be modified for easy recovery afterwards with for example a magnetic core (49, 50). Therefore, VITO has developed solid sorbents by first selecting or forming a suitable solid backbone before grafting active organic scavenging groups onto the support using a green aqueous synthesis method. Two types of regenerable three-dimensional (3D) structured metal oxide supports (Figure 6) have been investigated: monodisperse microspheres and 3D printed monoliths. These types of supports have the advantage of reducing mass diffusion limitations, optimising packing density and decreasing pressure drop while allowing fast adsorption and desorption cycling times.

Fig. 6

The functionalised microspheres show good selectivity for palladium over other pgms and impurities giving a method to remove palladium from acidic solutions. The adsorbed palladium can be easily recovered by stripping with concentrated HCl acid resulting in a concentrated acidic solution of palladium. The sorbents have a palladium capacity of 0.33 mmol g^–1 from acidic solutions (pH 2). When stripped with 3 M HCl, a seven-fold increase in the palladium concentration between the feed (100 mg l^–1 palladium) and stripping solution was observed. No significant performance loss of the sorbents was observed between the different cycles; only seven cycles have been tested to date.

The developed material is a hybrid (ceramic-organic) adsorbent selective to palladium over the rest of the pgms with good hydrolytic stability that can be regenerated and reused. If a degradation of the organic scavenging groups and hence, decreased performance is observed over a higher number of cycles, the metal oxide support can be recovered and refunctionalised. The biggest drawback is the limited sorbent capacity, but more developments to improve this are ongoing by increasing the specific surface of the supports and optimising the grafting conditions.

SINTEF investigated extracting pgms from the feedstock using pyrometallurgy, employing copper as the pgm collector, followed by a molten salt electrolysis process (Figure 7).

Fig. 7

The pyrometallurgical process was investigated at two laboratory scales (10 g and 5 kg). The pgm recovery rates, in the alloy phase, were close to 100%, and the copper-collector recovery rates were in the range 82–100%, when using the optimised parameters: 10 wt% copper-collector, 10–15 wt% calcium oxide, 1600–1650ºC and 1–1.5 h holding time. This pyrometallurgy step pre-concentrated the pgms by ~10 times from feedstock to the generated metallic phase.

In the electrolysis step, the copper-pgm alloy is used as anode in an electrorefining cell with the eutectic LiCl-KCl as electrolyte at 450ºC. The experiments demonstrated the selective extraction of the metal phase (copper), which was recovered at the cathode with a current efficiency of ca. 70%. Under these conditions, the pgms (and other impurities) remain in the anode residue giving a solid with >99.9% purity. Further separation steps are needed to separate the three pgms from each other.

One significant advantage of pyrometallurgical over hydrometallurgical processes is the lower reagent use in relation to the feedstock pgm content (kilograms of reagent per kilogram of pgm) (51). As such, it provides attractive conditions for preconcentrating pgms from very dilute wastes, such as the PLATIRUS feedstocks; the optimised conditions of the SINTEF process showcase that. High recovery rates of the copper collector at the electrolysis cathode have been achieved and it can be recycled for the next pyrometallurgical step.

The electrolysis process allows the extraction of copper from the pgm-containing copper anode in a molten salt electrolyte with better selectivity and kinetics as well as lower energy consumption than in state-of-art copper-refining processes using aqueous solutions (52). Cu(I) species are stable in the molten salt electrolyte, thus the voltage (and energy) needed in the electrorefining process is lower than in an analogous aqueous solution process where Cu(II) are the solely stable species. In general, the kinetics of the electrode charge-transfer reaction in molten salts are considered faster due to the high operational temperature (53).

The challenges of operating the SINTEF pyrometallurgical step are the same as those found in industry. The separation of pgm microparticles from the molten slag phase is impacted by the slag viscosity and metal-to-slag interfacial tension and in turn affect the pgm extraction efficiencies. The energy consumption was ca. 5.5 kWh kg^–1 copper recovered in the pyrometallurgical step and ca. 7 kWh kg^–1 pgm recovered in the electrolysis step (equal to 0.3 kWh kg^–1 copper refined) under optimised experimental conditions. Energy consumption can be challenging to extrapolate from laboratory to industrial scale. Due to the energy intensiveness of pyrometallurgical processes, energy consumption and associated cost must be evaluated at larger scale.

Molten chloride mixtures can be used as a reaction media in the chlorination of oxide mixtures, ores or industrial byproducts. The dissolution reaction generates chloride compounds at much lower temperatures (ca. 450ºC) than those needed in solid-gas chlorination reactions (ca. 1000ºC). This is due to significant solvation effects of the dissolved metal cation with the chloride ions of the molten chloride media.

SINTEF investigated the selective recovery of the pgms using molten salts and chlorine gas as an oxidiser and chlorination agent, followed by an electrolysis process (Figure 8). LiCl-KCl eutectic mixture was chosen as the best candidate. The feedstock was fed into the reactor at 450ºC without any pretreatment or up-concentration steps, resulting in a single step pgm extraction process. Silicon, magnesium and aluminium were not dissolved thus remain as a solid sludge at the bottom of the reactor. The analysis of the residue shows that, after 3 h, the total chlorination and therefore dissolution of the pgms is close to 50%. The dissolved pgms are recovered in an electrolysis process as a metallic-pgm alloy at the cathode resulting in chlorine evolution at the anode. Only 80% of the dissolved pgms can be accounted for in the molten chloride as pgm-chlorocomplexes; it is believed unaccounted pgm mass corresponds to formed volatile pgm-chlorocomplexes that could be recovered from the off-gas system by condensation; estimates calculated from a mass balance suggested a feed volatilisation of ca. 10% palladium, 10% platinum and 20% rhodium.

Fig. 8

This chlorination process presents clear advantages in terms of rate, conversion and selectivity when compared with traditional gas-solid reaction systems. In the latter, the rate and conversion are limited by the contact of the gaseous chlorinating agent and the material to be chlorinated, and the reaction occurs non-selectively as, at the much higher temperature, chlorination of all other elements contained in the material occurs, not only pgms.

Though faster than gas-solid reactions, the kinetics are still slow when using chlorine gas and the use of other gaseous chlorination agents, such as HCl, should be tested. Further optimisation of the process is required to achieve competitive pgm recovery rates, including recovery of the volatile pgm-chlorocomplexes.

GDEx is defined as the reactive precipitation between metal precursors in solution and intermediates from the reduction of gases at a gas-diffusion electrode. When the gas is air or O₂, the O₂ reduction reaction leads to hydroxyl ions and hydrogen peroxide being formed in the pgm solution, which react, forming oxides or hydroxides (Figure 9). Alternatively, the process can run with other gases. The GDEx process is enabled by VITOCoRE^® multi-layered gas-diffusion electrodes (54).

Fig. 9

GDEx uses the cleanest possible reagent, the electron, and it is highly versatile, as it can be used to recover many different metals. The process uses an inexpensive reactor. As it operates in a flow-cell configuration, it is easily up-scalable by stacking multiple individual cells, without a reduction in performance. The process is highly reproducible, involves mild operation conditions (room temperature and atmospheric pressure). The process has a low energy consumption, for example ~2-6 kWh kg^–1 of materials recovered, when compared with electrowinning platinum from chloride media reported at 21 kWh kg^–1 platinum (55). Finally, it is efficient, i.e. 50% current efficiency for the formation of the reactive intermediates that fully react with the metal precursors to achieve the targeted recovery.

Notably, the recovery of dilute metals and simultaneous synthesis of nanostructures with GDEx is fast, with rates approaching ~3–15 kg per day, using a single, inexpensive, electrochemical reactor under flow regime.

The best results were achieved with dilute metal concentrations, and significant optimisation is required for solutions with metal concentrations above 10 g l^–1. Especially with high metal concentrations, it is expected that, after a period of operation, the electrode would become clogged and require an acid treatment to regenerate the electrode porosity. The recovery of the precipitated materials is impacted by the unoptimised downstream separation and drying processes. Some reagents, such as sulfur-based compounds, are known to interfere with the process under defined processing conditions, coprecipitating with the pgms.

A laboratory scale MW system (flexiWAVE) equipped with a spinning carousel holding 15 pressure-sealed Teflon-lined reactors was used to process ~1.3 kg autocatalyst (Figure 13).

Fig. 13

The leaching temperature is measured by a thermowell contained optic fibre, placed in one of the 15 reactors. Each of the 15 reactors was loaded with 5 g of the feedstock and 50 ml 6 M HCl solution. The MW-assisted reaction took place at 150ºC for 10 min, with a heating time to the set temperature of 15 min. Subsequently, the reactor was cooled, opened and the leachate was vacuum filtered. It is noteworthy that this leaching process did not require addition of H₂O₂ as an oxidation agent and thus the formation of hazardous head space gas mixtures (containing H₂ and Cl₂ gas) was avoided. During vacuum filtration, the leach residues were washed with 6 M HCl at room temperature. Both leachate and washing liquids were collected.

Overall, ~1.3 kg of material was leached by performing 18 MW-leaching runs. Three batches of material with different grain sizes and total masses were processed, hence three leachates (L1, L2, L3) and washing waters (W1, W2, W3) samples were obtained (Table III).

The leachate containing L1, L2 and L3 (Table III), as prepared by VITO, was processed using the solvent extraction flowsheet developed at KU Leuven (Figure 5). Due to the low pgm concentration of the leaching residue washings, these solutions were not processed. A continuous solvent extraction demonstration using multi-stage mixer-settlers was undertaken to process the leachate (Figure 14).

Fig. 14

First, Pd(II) and Pt(IV) were quantitatively extracted in two countercurrent stages with [A336][I] in p-cymene at O/A = 1/3, leaving Rh(III) in the raffinate. The impurities (mainly aluminium, barium, cerium, iron and tin) in the loaded organic phase were removed with NaCl solution. Then selective stripping of palladium was achieved by equilibrating the scrubbed organic phase with aqueous ammonia in NaCl solution in four-stage mixer-settlers at O/A = 3/1. Followed by the recovery of platinum from the palladium-free loaded organic phase was performed using acidic thiourea solution [CS(NH₂)₂/HCl] in four countercurrent stages (O/A = 2/1). After being washed with diluted HCl, the IL requires regeneration by contact with KI to replace the chloride for an iodide anion reforming [A336] [I].

The work at KU Leuven demonstrated that continuous solvent extraction is feasible for not only the preparation of [A336][I] but also the selective recovery of individual pgms, with recoveries of 81 ± 3% Pd(II), 62 ± 2% Pt(IV) and 54 ± 5% Rh(III), from the feedstock leachate. The mass loss was mainly as a result of the samples taken for analysis and the mixer-settlers start-up and finish steps. These losses are not representative of industrial scale operation in which mixer settlers would operate continuously.

The rhodium raffinate and platinum strip were further processed using conventional chemical transformations to convert the pgms to the nitrate form. The palladium strip was sent for further processing by the GDEx process.

GDEx attained a recovery of ~70 ± 1% of palladium from the KU Leuven strip sample (Figure 15). Palladium was recovered with a purity of 91–93%, with platinum, rhodium and aluminium as the major impurities, and additional minor impurities of barium, iron, magnesium, cobalt and copper. Only two batches were processed in comparison to the eighteen leaching batches performed thanks to the up concentration in the solvent extraction process. The selectivity of the recovery can be optimised by further investigation into the effect of the different GDEx operational variables, such as the influent concentration, hydraulic retention time, applied potential or current.

Fig. 15

The palladium sponge generated from the GDEx technology was further processed using conventional chemical transformations to convert it to palladium nitrate.