Leaching Platinum Group Metals from Simulated Spent Auto-Catalyst Material Using Ozone and Hydrochloric Acid

Abstract

This paper reports the development of a process for leaching Pt, Pd, and Rh from simulated spent auto-catalyst material using ozone and hydrochloric acid in order to produce a pregnant leach solution that could be fed to an industrial precious metal refinery. The effects of O₃ mass flow, initial acid concentration, and temperature were investigated using a Box–Behnken experimental design with three centre-point runs and a total leach time of 6 h. Set points of 3.34, 5.01, and 6.68 g/h; 1.0 M, 3.0 M, and 5.0 M; and 30, 60, and 90 °C were used for O₃ mass flow, hydrochloric acid concentration, and temperature, respectively. The optimal extractions for Pt, Pd, and Rh were 80%, 85%, and 42%, respectively, at 5.01 g/h O₃, 5.0 M HCl, and 90 °C. Statistical analyses indicated high dependencies of Pd and Rh on hydrochloric acid concentration and temperature, with only Pt displaying a significant dependence on O₃ mass flow.

1. Introduction

The catalytic converter was developed during the 1970s to treat harmful emissions from the internal combustion engine [1]. Modern three-way catalytic converters make use of Pt, Pd, and Rh to catalyse the conversion of CO, trace hydrocarbons, and NO_x gases into more environmentally benign gases [2,3]. The honeycomb substrate of an automotive catalyst consists of synthetic cordierite ((Mg,Fe)₂Al₄Si₅O₁₈), which is coated in a layer of 90% γ-Al₂O₃ as a washcoat. PGMs are impregnated onto this porous layer [4]. The PGM content of a catalytic converter varies by manufacturer, but a typical example manufactured by Heesung PMTech Ltd reported a PGM content of 750, 1000, and 250 mg/kg of Pt, Pd, and Rh, respectively [5]. This contrasts with a primary ore head grade that is typically of the order of 6 g/t (3PGE + Au), and clearly indicates the significant improvements in sustainability that recovery from used automotive catalysts would enable. Many recent developments in the field are reviewed elsewhere [6,7].

The first step when recycling auto-catalysts is mechanical pretreatment whereby the internal structure is removed from the stainless steel casing. Following this, the internal material is milled. Subsequently, either a pyrometallurgical or hydrometallurgical processing stage is conducted. Pyrometallurgy is the most popular method of PGM recovery from scrap or e-waste on an industrial scale. The black copper smelting process conducted by Umicore and the Boliden’s Rönnskär smelter with copper as the collector are examples of this [8,9,10].

Hydrometallurgical processing options include pretreatments, leaching, supercritical fluid extraction, microwave-assisted leaching, and bioleaching [5,11]. A drawback of these processes is the use of strong acids. This is especially true for alternatives to direct leaching, such as the dissolution of the washcoat to free the PGM particles from the carrier layer [12].

Chloride leaching is the most popular and frequently used complexation leaching system [13]. The mechanism of complexation of the PGMs with Cl⁻ during oxidative leaching is described by Equations (1)–(4):

$$Pt+4 C{l}^{-}\leftrightarrow {\left(PtC{l}_4\right)}^{2-}+2 e^{-}$$

(1)

$${\left(PtC{l}_4\right)}^{2-}+2 C{l}^{-}\leftrightarrow {\left(PtC{l}_6\right)}^{2-}+2 e^{-}$$

(2)

$$Pd+4 C{l}^{-}\leftrightarrow {\left(PdC{l}_4\right)}^{2-}+2 e^{-}$$

(3)

$$Rh+6 C{l}^{-}\leftrightarrow {\left(RhC{l}_6\right)}^{3-}+3 e^{-}$$

(4)

The redox potentials for these four reactions range from approximately −0.75 to −0.40 V relative to the SHE [14]. Furthermore, the introduction of an oxidant into such a system facilitates the increase in the potential to promote the formation of these chloride complexes. Popular oxidants include oxygen, ozone, hydrogen peroxide, ferric ion, manganese dioxide, and sodium nitrate [15]. Cl₂ gas is the most widely used oxidant that is combined with HCl, but it is highly corrosive. Alternatively, O₃ gas has a high oxidising potential and is considered safer and easier to use [16].

The oxidising potential of ozone is high (E⁰ = 2.076 V) and is only surpassed by fluorine and hydroxyl radicals (OH) [14]. However, hydroxyl radicals are generated through the spontaneous decomposition of O₃ over time [17].

Due to the unstable nature of O₃, the gas must be generated in situ and cannot be stored or transported. Although not as corrosive as Cl₂ gas, O₃ still requires safety measures and corrosion-resistant materials [16]. Additionally, a gas destruction method should be in place to treat excess O₃ present in the off-gas [18]. However, in medium- to large-scale processes, this off-gas can be reclaimed and used in O₂-consuming stages as a recycle stream [19].

Equation (5) highlights the half-reaction for O₃. Combining this with the reactions in Equations (1)–(4) gives the leaching reactions when using O₃ as an oxidant and HCl as a lixiviant for Pt, Pd, and Rh.

$$O_3+H^{+}+2e^{-}\leftrightarrow O_2+H_{2}O$$

(5)

Various investigations have been conducted into the use of O₃ in PGM leaching. The effect of H₂O₂ and O₃ on the extraction of PGMs from spent auto-catalyst material was investigated [20]. It was determined that Pt extraction displayed a significant dependency on adjustments in the O₃ flowrate. Extractions of 21.1%, 91.6%, and 29.3% Pt, Pd, and Rh, respectively, were achieved at 1 g/h O_3, while 47.8% Pt was achieved at 3 g/h O₃. Additionally, a study was conducted on the extraction of Pt, Pd, and Au from magnetite ore using O₃ [21]. At temperatures less than 30 °C, 3 h of leaching, 175 mg_ozone/L_sol.min (5.25 g/h if solution volume is assumed to be 0.5 L), and in 1 M NaCl, extractions of 90%, 70%, and 50%, respectively, were achieved for Pt, Pd, and Au. A study [18] on the leaching of Au and Pd using a dilute chloride medium and aqueous O₃ indicated that the extraction of Rh was dependent on elevated temperature, and high Pt extraction was only achieved at increased HCl concentrations ($\approx$6 M). Generally, increasing the O₃ concentration in solution was found to increase leaching efficiency. Finally, an electrochemical investigation was conducted to establish the impact of O₃ as an oxidant on Pt extraction [22]. In this study, 66% Pt extraction was achieved at 1 M HCl, a pH of 0.5, and 15 °C with a constant O₃ supply generated through the introduction of 1.92 LPM of O₂ to the generator.

During the use of auto-catalysts in automobiles, the temperatures of operation within the devices can reach up to and above 1150 °C. These temperatures can induce the formation of PGM oxides in the washcoat layer [23]. The formation of Rh₂O₃ is believed to be a particular issue resulting from these conditions. This oxide forms a protective layer on the surface of the auto-catalyst material, making it difficult to conduct Rh dissolution [12]. To combat this, the solid material can be subjected to a prereduction stage in which the oxides are converted to their pure metal form through roasting or chemical treatment [24,25].

In this paper, we investigated the extraction of PGMs using an HCl/O₃ leach of a simulated spent automotive catalyst consisting of PGM oxides and cordierite. The effects of O₃ mass flow, initial acid concentration, and temperature were investigated, and the investigation was optimised by using a Box–Behnken experimental design. An important consideration was that the leach solution feasibly be introduced into the precious metal refinery typical of a primary metal producer. As part of this process, the extraction of the impurities Si, Al, and Mg from the synthetic cordierite material was also investigated.

2. Materials and Methods

Concentrated formic and hydrochloric acid were used to make up the prereduction and lixiviant solutions, respectively. Technical grade (99.5%) O₂ was used to generate O₃ using the BIO-10 generator from Ozonetek cc, Paarl, South Africa. Anhydrous PtO₂ (99.95%), PdO (99.9%), and Rh₂O₃ (99.9%) powders from ThermoFisher Scientific (Waltham, MA, USA) were used to simulate the PGM oxide content of a spent catalytic converter. The 50% passing sizes of PtO₂, PdO, and Rh₂O₃ were determined by SEM analysis to be 35, 50, and 50 $\mathsf{\mu }$m, respectively. Synthetic cordierite powder (2MgO·2Al₂O₃·5SiO₂, -325 mesh) from American Elements (Los Angeles, CA, USA) was used to simulate the substrate of a spent auto-catalyst. Each experiment was conducted with 10.0 g samples containing 0.0087 g PtO₂, 0.0115 g PdO, and 0.0031 g Rh₂O_3, with cordierite making up the balance. Finally, solid potassium iodide (KI 99.5% from Sigma-Aldrich, St. Louis, MO, USA) was required for the make-up of the O₃ destruction solution.

Formic acid prereduction of the synthetic catalyst powder was conducted prior to each leach. The prereduction was conducted at 15 vol% HCOOH, a solid-to-liquid (S/L) ratio of 50 g/L, 400 rpm agitation, and 60 °C for 1 h. Three samples of the solution after the prereduction were analysed to determine the extraction of the PGMs during this stage. Between prereduction and the leach, the slurries were vacuum-filtered, and the solid residue was allowed to air-dry for 24 h.

Leaching was conducted in a 1 L glass vessel with a thermocouple and hotplate for temperature control. A magnetic stirrer bar was used for agitation at 500 rpm. A Box–Behnken experimental design (BBD) was used to determine the effects of O₃ mass flow (${\dot{m}}_{O_3}$), initial acid concentration (${\left[HCl\right]}_0$), and temperature ($T$). With a centre-point triplicate, this equated to 15 runs conducted at factor set-points of 3.34, 5.01, and 6.68 g/h O₃; 1.0, 3.0, and 5.0 M ${\left[HCl\right]}_0$; and 30, 60, and 90 °C. The calibrated mass flow of O₃ was controlled by varying the O₂ volumetric flowrate to the O₃ generator. The solid-to-liquid ratio was constant for all runs at 20 g/L. The total leach time was 360 min, and solution samples were taken at 10, 30, and 60 min, followed by samples on the hour for the remainder of each leach. All samples were analysed for Pt, Pd, and Rh using inductively coupled mass spectroscopy (ICP-MS), and a portion of samples were analysed for Si, Al, and Mg using inductively coupled optical emission spectroscopy (ICP-OES). Based on these concentrations and the mass of each component entering the prereduction, the extraction of PGMs during leaching was determined.

Statistica^® (Version 14.0.1, Tibco, Palo Alto, CA, USA) was used to evaluate the statistical significance and estimate the effects of the three factors which were investigated. Additionally, a second-order model was fitted for each PGM and used to determine whether a maximum stationary point exists within the experimental space.

3. Results and Discussion

3.1. PGM Extraction

The minimum, maximum, and centre-point average extractions for the PGMs are displayed in

Table 1. The minimum and maximum extractions for Pt, Pd, and Rh (in bold) after 360 min under their respective conditions and the averages at the centre-point settings.

Run Label:	PGM:	Ext: (%)	${\dot{\mathit{m}}}_{{\mathit{O}}_3}$: (g/h)	${\left[\mathit{H}\mathit{C}\mathit{l}\right]}_0$: (M)	$\mathit{T}$: (°C)
Minimums
9	Pt	2.04	5.01	1.0	30
	Pd	56.5
	Rh	2.24
Averages at centre point
13, 14, 15	Pt	64.3	5.01	3.0	60
	Pd	74.9
	Rh	9.51
Maximums
10	Pt	88.6	5.01	5.0	30
	Pd	85.0
	Rh	8.38
12	Pt	78.7	5.01	5.0	90
	Pd	85.0
	Rh	41.6

. Additionally, the values at each design point in the experimental space are displayed for Pt, Pd, and Rh in Figure 1.

The lowest extractions were all achieved with the middle O₃ mass flow, a low initial acid concentration, and low temperature conditions. The maximum extractions for all PGMs occurred with the middle O₃ mass flow and a high initial acid concentration; however, the low set-point for temperature gave the maximum extraction for Pt, while the high temperature set-point gave the maximum extraction for Rh. The maximum Pd extraction of 85.0% occurred at both high and low set-points for temperature.

Overall, this suggests a significant positive dependency for PGM extraction on increased initial acid concentration. Additionally, Rh displays a substantial increase in extraction with increasing temperature. These results align with the findings of [18]. The extraction of Pt showed the largest range, varying over approximately 86 percentage points, while the extraction of Pd had the smallest range within the experimental space at 30 percentage points. This can be seen in Figure 1a,b. The maximum Rh extraction did not exceed 42% throughout experimentation (see Figure 1c).

3.2. Extraction Kinetics

The extractions at each sample time were plotted to confirm whether complete extraction was achieved. Additionally, these plots provide an indication of the kinetics for the extractions of each PGM. Figure 2 shows the extractions for Run 10, which has the largest extractions of Pt and Pd. The extractions of Pt and Pd reached a steady state after approximately 60 min, while the extraction of Rh continued to increase over the full 360 min. This indicates incomplete extraction for Rh under these conditions, even after 360 min. The kinetics of Rh extraction are much slower those of Pt and Pd under these and all other conditions. However, extractions of Pt and Pd beyond 90% and 85%, respectively, were not achieved.

In Figure 3, the extractions for Run 12 are displayed. This run achieved the highest combined PGM extraction. Once again, maximum extractions of Pt and Pd were achieved within 60 min, and the extraction of Rh was still incomplete after 360 min. However, the extraction of Rh was significantly higher than that achieved in Run 10. Additionally, the rate of Rh leaching within the first 120 min was increased significantly at the higher temperature of Run 12, highlighting the sensitivity of Rh extraction kinetics to temperature.

Finally, the extraction kinetics for the run giving the lowest combined PGM extraction (Run 9) are shown in Figure 4. The extractions of Pt and Rh are low and indistinguishable, and the extraction of Pd does not reach a steady state after 360 min. This result further highlights a significant positive dependency for the extraction of Pt on a high initial acid concentration.

3.3. Satistical Analyses

The extractions were used to generate second-order models using Statistica^®. A significance ($\alpha$) value of 0.05 was set, and initially, all interactions were included. However, there were six interactions which were not estimable by Statistica^® as they are linear combinations of other effects in a Box–Behnken analysis. ANOVA tables and Pareto charts were generated for the results for each PGM. These tools helped rank the linear, quadratic, and interaction terms which were estimable based on their calculated significance and the size of their estimated effect. Additionally, they provided the R², adjusted R² (adj-R²), and lack-of-fit (LOF) significance values.

For each PGM, a backwards selection process was conducted, and insignificant interaction terms were excluded from the original second-order model. Subsequently, the final model was used to generate response surface and contour plots and to predict the optimum conditions. The fitted extraction models for Pt, Pd, and Rh can be found in Appendix A, together with statistical information on the lack-of-fit and adjusted R² values following the exclusion of insignificant parameters.

A fitted response surface plot and the corresponding contour plot at a constant initial acid concentration of 5.0 M for Pt are shown in Figure 5. The constant acid concentration of 5.0 M (the highest concentration tested) was chosen as maximum Pt extractions were achieved at this value. There is a maximum stationary point for Pt extraction of ~80% at approximately 4.5 g/h O₃ and a temperature of 65 °C.

Conducting the same process for Pd, a final second-order model was chosen, and the fitted response surface and contour plots using this model were generated at a 5.0 M initial acid concentration. These plots are displayed in Figure 6.

Optimum Pd extraction of 85% is indicated by the model at the centre-point ozone mass flow (5.01 g/h O₃) and maximum temperature (90 °C). Within the experimental space with 5.0 M HCl and at 5.01 g/h O₃, the extraction of Pd predicted by the model varies by only 4 percentage points, indicating that the extraction of Pd is relatively insensitive to temperature under these conditions. Therefore, using a high acid concentration is sufficient to enable Pd extractions > 80% in this system.

Finally, the fitted response surface for Rh extraction and the contour plots at a constant temperature of 90 °C (at which highest Rh extraction was obtained) are shown in Figure 7. No stationary point or inflection is visible in the experimental space, but the predicted numerical maximum extraction of approximately 40% is found at 3.34–6.68 g/h O₃, 5.0 M, and 90 °C.

The results above indicate that the reaction kinetics and overall extractions for all PGMs are significantly influenced by the initial acid concentration and temperature. The extractions of Pt and Pd were most influenced by the initial acid concentration, while the extraction of Rh was most impacted by the temperature. The strong dependence on the initial acid concentration was also noted by other researchers [5,23]. However, operation at very high acid concentrations is expensive and can cause significant substrate, washcoat, and overall impurity dissolution [5]. The addition of an oxidant such as ozone, as used in the current study, mitigates these effects by allowing for operation at lower acid concentrations. This is discussed in more detail elsewhere [26].

Figure 2 and Figure 3 show the rapid kinetics for leaching Pt and Pd under optimal conditions. However, both the leaching kinetics and the maximum extraction of Rh were poor. Similar conclusions have been drawn by other researchers, and it is hypothesised that Rh becomes passivated very quickly when exposed to an oxidative environment of certain leaching solutions [11,27]. The significant effect of temperature on the leaching of Rh also corroborates the findings of others [18,28]. These findings suggest the possibility for the selective removal of Pt and Pd extraction while limiting Rh extraction. This could allow for downstream processing of the filtered residue to target the dissolution of Rh more efficiently.

When integrating a PGM extraction process with the feed to a precious metal refinery, the introduction of Si, Al, and Mg from the possible dissolution of cordierite is an important consideration. From

Table 2. The minimum and maximum extractions for the impurities Si, Al, and Mg under their respective conditions and the averages at the centre-point settings.

Run No.:	Impurity:	Ext: (%)	${\dot{\mathit{m}}}_{{\mathit{O}}_3}$: (g/h)	${\left[\mathit{H}\mathit{C}\mathit{l}\right]}_0$: (M)	$\mathit{T}$: (°C)
Minimums:
10	Si	1.78	5.01	5.0	30
	Al	5.08
	Mg	5.42
9.	Si	2.24		1.0
	Al	1.40
	Mg	1.94
Averages at Centre-Point:
13 14 & 15	Si	2.65	5.01	3.0	60
	Al	16.6
	Mg	17.0
Maximums:
11	Si	7.82	5.01	1.0	90
	Al	39.7
	Mg	38.5
12	Si	1.93		5.0
	Al	66.6
	Mg	68.4

, the extraction of Si was found to be low under all operating conditions. However, both Al and Mg were significantly extracted under both centre-point conditions (Runs 13–15), and particularly under optimal conditions for PGM extraction (Run 12). The maximum extractions of Al and Mg (66.6 and 68.4%, respectively) represent a significant degree of impurity extraction. As the average mass of the cordierite per run is 9.98 g, with the PGMs contributing a trace proportion of the solid material, this highlights the substantial dissolution of the simulated substrate material under these conditions and would require further attention if the leaching process were to be integrated with a PMR.

4. Conclusions

Ozone-enhanced hydrochloric acid leaching experimentation to extract PGMs from a prereduced synthetic automotive catalyst material was conducted successfully using a Box–Behnken design. The maximum achieved PGM extractions were 88.6%, 85.0%, and 41.5% for Pt, Pd, and Rh, respectively. The run which showed the largest combined PGM extraction of 78.7%, 85.0%, and 41.5% for Pt, Pd, and Rh, respectively, was conducted at 5.01 g/h O₃, 5.0 M, and 90 °C. An analysis of the extraction kinetics indicated fast kinetics for Pt and Pd leaching, with complete extraction achieved between 60 and 180 min for runs with an initial acid concentration >3.0 M. However, the Rh extractions highlighted incomplete extraction even after 360 min of leaching.

Statistical analyses predicted the optimum conditions for Pt extraction at approximately 4.5 g/h O₃, 65 °C, and 5.0 M ${\left[HCl\right]}_0$. The optimal conditions for Pd and Rh were both estimated at 5.01 g/h, 5.0 M, and 90 °C. The most significant factor in PGM extraction was the initial HCl concentration followed by temperature. Rh extraction was especially dependent on elevated temperature. Pt was predicted to be significantly influenced by ozone mass flow. The optimum conditions for the maximum combined PGM extraction were found to be 5.0 g/h O₃, 5.0 M HCl, and 90 °C. These conditions gave extractions of 85% for Pd, approximately 42% for Rh, and 79% for Pt. Under optimal PGM extraction conditions, significant quantities of Al and Mg were co-extracted from the cordierite, which means that further attention must be given to this issue if the process were to be combined with a precious metal refinery.