Effects of SiO₂ and CaO on Distributions of Platinum Group Metals Between Cu–CuO_0.5 and Pb–PbO-Based Slags at 1523 K

Abstract

Platinum group metals (PGMs) are recovered using molten Cu or Pb as collector metals during pyrometallurgical recycling processes. The collector metal containing PGMs is oxidized to produce a PGM-enriched alloy and slag, primarily containing Cu or Pb oxide. Additionally, the slag comprises a considerable amount of PGMs. Therefore, it is returned to the upstream processes to be used as a secondary raw material, and some of PGMs remain in the recycling processes. Herein, the distributions of Rh, Pd, and Pt between molten metals (Cu, Pb) and corresponding metal oxide-based (CuO_0.5, PbO) slags were investigated at 1523 K. SiO₂ or CaO was added to the slag to a maximum concentration of 20 mass pct, thereby reducing the oxygen partial pressure in the system. Furthermore, the addition of an optimum amount of SiO₂ or CaO into various combinations of slags and collector metals reduced the concentrations of PGMs in the slags to less than approximately 1/10 of those without adding SiO₂ or CaO. Therefore, the addition of SiO₂ or CaO during the oxidation smelting process effectively reduced the dissolution of PGMs in the slag, resulting in a less amount of PGMs circulating in the recycling processes.

Introduction

Platinum group metals (PGMs) are classified as precious metals. The natural distribution of PGMs is highly concentrated in South Africa and Russia, accounting for more than 80 pct of the total production from ores.^[1] Moreover, PGMs are irreplaceable for several applications owing to their unique physicochemical properties. PGM recycling technologies are expected to gain considerable significance for facilitating sustainable management and the efficient use of natural resources in future.^[2,3] Automotive catalytic converters account for over 60 pct of the global demand for PGMs; therefore, the recovery of PGMs from catalytic converters is receiving significant attention worldwide. During the pyrometallurgical processes for recovering PGMs from automotive catalytic converters, the PGMs are concentrated in molten Cu or Pb, and the supporting ceramics are separated into the molten slag.^[3,4,5] Subsequently, the molten Cu or Pb, containing PGMs, is oxidized to produce a PGM-enriched alloy and slag, which comprises primarily Cu or Pb oxide. In addition, these CuO_0.5- and PbO-based slags comprise a considerable amount of PGMs. Therefore, they are returned to the upstream processes to be used as a secondary raw material, and some of PGMs remain in the recycling processes. The prices of PGMs frequently fluctuate; therefore, they must be recovered rapidly, and the amount of PGMs circulating in the process must be reduced. Therefore, understanding the dissolution behavior of PGMs in the molten slag is essential.

The distribution behavior of PGMs between the molten metal and the slag has been reported earlier. Nakamura et al.^[6,7] measured the solubility of Pt in various slags and suggested that Pt is stable as a platinum cation (Pt²⁺) and platinate ion (PtO₂^2–) in acidic and basic slags, respectively. Wiraseranee et al.^[8,9,10] and Morita et al.^[11] determined the solubility and dissolution states of PGMs in Na₂O–SiO₂- and CaO–SiO₂-based slags. Baba et al.,^[12] Yamaguchi,^[13] Shuva et al.,^[14] Avarmaa et al.,^[15,16] Klemettinen et al.,^[17] and Chen et al.^[18,19] investigated the distribution of precious metals in a FeO_x-based slag system, which is primarily used in the smelting of non-ferrous metals. Borisov et al.,^[20] Yamaguchi,^[21] and Nisijima et al.^[22] investigated the distribution of precious metals in an Al₂O₃–CaO–SiO₂–MgO slag system. In most of these studies, the distributions of precious metals to the slag are demonstrated an increase in the dissolution of PGMs in slags with increasing oxygen partial pressure and concentration of Cu in slags. We investigated the distribution of Pt between the molten Cu and the SiO₂–CaO–CrO_x slag system in our previous study^[23] and observed an increase in the dissolution of Pt in the slag with increasing concentration of Cr in the slag. Swinbourne et al.^[24] investigated the dissolution of Au in PbO–SiO₂ slag and reported that the Au concentration increased with increasing PbO concentration in the slag. However, to the best of our knowledge, the distribution of PGMs in CuO_0.5- or PbO-based slags has not been reported yet. Understanding the distribution behavior of PGMs in slags is important for the effective recycling of PGMs using the oxidation smelting process.

In this study, the distributions of Rh, Pd, and Pt between molten metals (Cu and Pb) and their corresponding metal oxide-based (CuO_0.5 and PbO) slags were investigated at 1523 K by adding SiO₂ or CaO to the slags to a maximum concentration of 20 mass pct.

Experimental Procedure

A total of 6 g of CuO_0.5 (99 mass pct purity), SiO₂ (99.9 mass pct purity), and CaO (99.9 mass pct purity) powder as a slag and 4 g of the homogenized 91 mass pct Cu and 3 mass pct of Rh, Pd, and Pt alloy were put into a MgO crucible with an inner diameter, a height, and a wall thickness of 18, 40, and 2.5 mm, respectively. In addition, a total of 9 g of PbO (99.9 mass pct purity), SiO₂, and CaO powder as slag and 6 g of the homogenized 97.06 mass pct Pb and 0.9814 mass pct of Rh, Pd, and Pt alloy were placed in a MgO crucible of the same dimensions as mentioned above. Different total masses were used to ensure the volume required for the measurement of the oxygen partial pressure. Furthermore, different alloy compositions were used to ensure equal mole fractions of PGMs in the Cu and Pb-based systems. Subsequently, the sample was injected into a mullite reaction tube with an Ar (99.99 vol pct purity) gas flow rate of 200 mL/min. After holding the sample at 1523 K (± 3 K) in a furnace using SiC-heating elements for 4 h, the oxygen partial pressure was indirectly measured via an oxygen-concentration cell using an MgO-stabilized ZrO₂ solid electrolyte, with the following formula [1]:

$$ + {\text{Pt}}/{\text{Ni}} \cdot {\text{NiO}}/{\text{ZrO}}_{2} \cdot {\text{MgO}}/{\text{sample}}/{\text{Re}}/{\text{Pt}} -. $$

(1)

Both electrodes were immersed in the slag and alloy phase. The change in the potential difference between 1 min was confirmed to be less than 0.005 V (±0.03 in terms of ${\text{log}}\,p_{{\text{O}}_{2}}$). Even when the Ar gas supply was temporarily stopped, the same potential difference was measured as when the gas was supplied. Therefore, it is considered that this measurement was not affected by the impurity oxygen in the Ar gas. The sample was held for another 0.5 h and quenched with water. A schematic of the experimental setup is shown in Figure 1. Subsequently, the slag and alloy phases in the crucible were separated and dissolved in an acid. The concentrations of the slag and alloy phases were quantified using inductively coupled plasma optical emission spectrometry (ICP-OES) and inductively coupled plasma mass spectrometry (ICP-MS). Tables I and II list the initial slag compositions and experimental results for the Cu–CuO_0.5 and Pb–PbO systems. The maximum amount of SiO₂ or CaO added was determined based on the binary state diagram of the respective slag system.^{[25,26,27,28]} The respective binary phase diagrams for slags are shown in Figure 2. Although MgO dissolves in the slag after equilibration, preliminary experiments confirmed that it is a homogeneous liquid.

Fig. 1

Schematic of the experimental setup

Table I Experimental Results for the Cu–CuO_0.5 System at 1523 K

Table II Experimental Results for the Pb–PbO System at 1523 K

Fig. 2

Binary phase diagrams: (a) Cu₂O–SiO₂, (b) Cu₂O–CaO, (c) PbO–SiO₂, and (d) PbO–CaO.^{[25,26,27,28]}

To confirm whether equilibrium was sufficiently reached with a holding time of 4 h, experiments were conducted with 100 mass pct CuO_0.5 slag for 10 h and 24 h. The same results were obtained in both experiments, indicating that equilibrium was reached by holding the slag for 4 h.

Results and Discussion

Solubility of MgO in Slags

The MgO crucible was partially dissolved in the slag and equilibrated. Figure 3 shows the relationship between the concentration of MgO and the concentration of SiO₂ or CaO in the slag. MgO barely dissolves, with a maximum of 0.34 mass pct, in the CuO_0.5, CuO_0.5–CaO, PbO, and PbO–CaO slags. In contrast, a significant increase in the MgO concentration is observed with increasing concentration of SiO₂ in the CuO_0.5–SiO₂ and PbO–SiO₂ slags. This is because MgO and CaO are basic oxides, and SiO₂ is an acidic oxide. Therefore, the use of CaO is speculated to suppress the reduction in the service life of the furnace refractory if the refractory is a basic oxide.

Fig. 3

Solubility of MgO in the slag at 1523 K

Oxygen Partial Pressure

The oxygen partial pressure in the system was indirectly measured via an oxygen-concentration cell using a MgO-stabilized ZrO₂ solid electrolyte. Figure 4 shows the relationship between the oxygen partial pressure and the concentration of SiO₂ or CaO in the slags.

Fig. 4

Relationship between the oxygen partial pressure and the concentration of SiO₂ or CaO in the slags

The dashed line in Figure 4 represents the oxygen partial pressure $({\text{log}}\,p_{{\text{O}}_{2}}=-3.82, -5.33)$, calculated using the standard Gibbs free energy change for reactions of [2] and [3] in the SGPS database of the thermodynamic computing system FactSage ver.8.1,^[29] when the pure liquid Cu and CuO_0.5 and the pure liquid Pb and PbO are in equilibrium at 1523 K, respectively. The oxygen partial pressure at equilibrium without SiO₂ or CaO agrees well with the calculated value. Furthermore, the oxygen partial pressure with the addition of CaO in the Cu-based system is lower than that with the addition of SiO₂. The oxygen partial pressure decreased with increasing concentration of SiO₂ in the Cu-based system. This is because the SiO₂ and CaO change the activities of CuO_0.5 and PbO. The validity of this result is discussed in Activity coefficients of section D.

$$ 4{\text{Cu }}\left( {\text{l}} \right) + {\text{O}}_{2} { }\left( {\text{g}} \right) = 4{\text{CuO}}_{0.5} { }\left( {\text{l}} \right):\;\Delta G^{0} = - 111311{ }\left( {\text{J}} \right){\text{ at }}1523{\text{ K}}^{{\left[ {29} \right]}}, $$

(2)

$$ 2{\text{Pb }}\left( {\text{l}} \right) + {\text{O}}_{2} { }\left( {\text{g}} \right) = 2{\text{PbO }}\left( {\text{l}} \right){ }:{ }\Delta G^{0} = - 155332{ }\left( {\text{J}} \right){\text{ at }}1523{\text{ K}}^{{\left[ {29} \right]}}. $$

(3)

Distributions of Rh, Pd, and Pt in the Alloy and Slag

The distribution ratios of Rh, Pd, and Pt between the slag and alloy were defined using Eq. [4]:

$$ L_{{\text{X}}}^{{{\text{s}}/{\text{m}}}} = \frac{{\left( {{\text{X }}\left( {{\text{mass pct}}} \right)\;{\text{in}}\;{\text{Slag}}} \right)}}{{\left( {{\text{X }}\left( {{\text{mass pct}}} \right)\;{\text{in}}\;{\text{Alloy}}} \right)}}, $$

(4)

where X = Rh, Pd, or Pt. Figures 5 and 6 depict the relationship between the distribution ratios of PGMs and the concentrations of SiO₂ or CaO in the CuO_0.5- and PbO-based slags, respectively.

Fig. 5

Distribution ratios of Rh, Pd, and Pt between liquid Cu and CuO_0.5-based slag at 1523 K: (a) CuO_0.5–SiO₂ slag system and (b) CuO_0.5–CaO slag system

Fig. 6

Distribution ratios of Rh, Pd, and Pt between liquid Pb and PbO-based slag at 1523 K: (a) PbO–SiO₂ slag system and (b) PbO–CaO slag system

The addition of SiO₂ or CaO (approximately 10 mass pct) to the Cu-based system decreases the distribution ratios of all PGMs to less than 1/10 of those without SiO₂ or CaO. Similarly, the addition of SiO₂ or CaO to the Pb-based system decreases the distribution ratios of all PGMs. Therefore, the addition of CaO or SiO₂ in the slag is suitable for reducing the dissolution of PGMs in the slag during the oxidation smelting process.

Activity Coefficient

The activity coefficients of each component in the slag were calculated to understand the interactions of each component. The activities used in this study were in the Raoult law standard state. The estimation of the activity of each component in the metal alloy requires thermodynamic data for the Cu–Rh–Pd–Pt and Pb–Rh–Pd–Pt systems; however, such thermodynamic data have not been measured. Therefore, the activities were considered consistent with the mole fractions based on the Rawls rule. The mole fractions of Cu and Pb were higher than 0.92; therefore, the activity coefficients ${\gamma }_{\mathrm{Pb}(\mathrm{l})}$ and ${\gamma }_{\mathrm{Cu}(\mathrm{l})}$ were assumed to be equal to one.

The activity coefficients of CuO_0.5 and PbO in the slag were calculated. The changes in standard Gibbs free energy ($\Delta {G}^{0})$ for the oxidation reactions of Cu and Pb are presented in Eqs. [2] and [3]. At equilibrium, the relationship between $\Delta {G}^{0}$ and the activity of Cu or Pb is expressed by Eqs. [5] and [6], respectively:

$$ \Delta G^{0}_{\left[ 2 \right]} = - {\text{R}}T{\text{ln}}\frac{{a_{{{\text{CuO}}_{0.5} { }\left( {\text{l}} \right)}}^{4} }}{{a_{{{\text{Cu }}\left( {\text{l}} \right)}}^{4} \cdot p_{{{\text{O}}_{2} }} }}, $$

(5)

$$ \Delta G^{0}_{\left[ 3 \right]} = - {\text{R}}T{\text{ln}}\frac{{a_{{{\text{PbO }}\left( {\text{l}} \right)}}^{2} }}{{a_{{{\text{Pb }}\left( {\text{l}} \right)}}^{2} \cdot p_{{{\text{O}}_{2} }} }}, $$

(6)

where $\Delta {G}^{0}$ is the change in the standard Gibbs free energy of reactions [2] and [3], $\mathrm{R}$ is the gas constant, $T$ is the absolute temperature, ${a}_{\mathrm{i}}$ is the activity of component i, and ${p}_{{\mathrm{O}}_{2}}$ is the oxygen partial pressure. The activity ${a}_{\mathrm{i}}$ is expressed by Eq. [7]:

$$ a_{{\text{i}}} = \gamma_{{\text{i}}} \cdot x_{{\text{i}}}, $$

(7)

where ${\gamma }_{\mathrm{i}}$ is the activity coefficient, and ${x}_{\mathrm{i}}$ is the molar fraction of component i. The activity coefficients of the liquid CuO_0.5 $({\gamma }_{{\mathrm{CuO}}_{0.5} \left(\mathrm{l}\right)})$ and PbO $({\gamma }_{\mathrm{PbO }\left(\mathrm{l}\right)})$ in the slag can be calculated using Eqs. [8] and [9], respectively:

$$ \gamma_{{{\text{CuO}}_{0.5} \left( {\text{l}} \right)}} = \frac{1}{{x_{{{\text{CuO}}_{0.5} }} }} \cdot \left\{ {x_{{{\text{Cu}}}}^{4} \cdot p_{{{\text{O}}_{2} }} \cdot {\text{exp}}\left( {\frac{{ - \Delta G^{0}_{\left[ 2 \right]} }}{{{\text{R}}T}}} \right)} \right\}^{\frac{1}{4}}, $$

(8)

$$ \gamma_{{{\text{PbO }}\left( {\text{l}} \right)}} = \frac{1}{{x_{{{\text{PbO}}}} }} \cdot \left\{ {x_{{{\text{Pb}}}}^{2} \cdot p_{{{\text{O}}_{2} }} \cdot {\text{exp}}\left( {\frac{{ - \Delta G^{0}_{\left[ 3 \right]} }}{{{\text{R}}T}}} \right)} \right\}^{\frac{1}{2}}. $$

(9)

Further, the activity coefficients of the PGM oxides in the slag were calculated. The interactions between PGMs were assumed negligible in each phase. Moreover, the oxidation states of the PGMs in the slag could not be determined. Therefore, considering the reactions of PGM oxides described in the thermodynamic database HSC ver.6.1,^[30] we assumed that PGMs dissolve in the slag as RhO_1.5, PdO, and PtO₂. $\Delta {G}^{0}$ for the oxidation reactions of Rh, Pd, and Pt are presented in Eqs. [10] through [12]. The thermodynamic database HSC provides data without any distinction between solid and liquid; therefore, all the metals and their corresponding oxides were assumed solids.

$$ \frac{4}{3}{\text{Rh }}\left( {\text{s}} \right) + {\text{O}}_{2} { }\left( {\text{g}} \right) = \frac{4}{3}{\text{RhO}}_{1.5} { }\left( {\text{s}} \right):\;\Delta G^{0} = 16839{ }\left( {\text{J}} \right)\;{\text{at}}\;{ }1523\;{\text{K}}^{{\left[ {30} \right]}}, $$

(10)

$$ 2{\text{Pd }}\left( {\text{s}} \right) + {\text{O}}_{2} { }\left( {\text{g}} \right) = 2{\text{PdO }}\left( {\text{s}} \right):\;\Delta G^{0} = 87898{ }\left( {\text{J}} \right)\;{\text{at}}\;1523\;{\text{K}}^{{\left[ {30} \right]}}, $$

(11)

$$ {\text{Pt }}\left( {\text{s}} \right) + {\text{O}}_{2} { }\left( {\text{g}} \right) = {\text{PtO}}_{2} { }\left( {\text{s}} \right):\;\Delta G^{0} = 100265\;\left( {\text{J}} \right)\;{\text{at}}\;1523\;{\text{K}}^{{\left[ {30} \right]}}. $$

(12)

The values of activity coefficients at infinite dilution for binary systems, adopted from the SGnobl values from FactSage ver.8.1, were used as the activity coefficients of Rh, Pd, and Pt in molten Cu and Pb.^[29] Table III lists the activity coefficient at infinite dilution in binary systems at 1523 K. The activity coefficients of PGM oxides in the slag were calculated using Eqs. [13] through [15] in the same manner as those of Cu and Pb:

$$ \gamma_{{{\text{RhO}}_{1.5} \left( {\text{s}} \right)}} = \frac{1}{{x_{{{\text{RhO}}_{1.5} }} }} \cdot \left\{ {a_{{{\text{Rh }}\left( {\text{s}} \right)}}^{\frac{4}{3}} \cdot p_{{{\text{O}}_{2} }} \cdot {\text{exp}}\left( {\frac{{ - \Delta G^{0}_{{\left[ {10} \right]}} }}{{{\text{R}}T}}} \right)} \right\}^{\frac{3}{4}}, $$

(13)

$$ \gamma_{{{\text{PdO}}\left( {\text{s}} \right)}} = \frac{1}{{x_{{{\text{PdO}}}} }} \cdot \left\{ {a_{{{\text{Pd }}\left( {\text{s}} \right)}}^{2} \cdot p_{{{\text{O}}_{2} }} \cdot {\text{exp}}\left( {\frac{{ - \Delta G^{0}_{{\left[ {11} \right]}} }}{{{\text{R}}T}}} \right)} \right\}^{\frac{1}{2}}, $$

(14)

$$ \gamma_{{{\text{PtO}}_{2} \left( {\text{s}} \right)}} = \frac{1}{{x_{{{\text{PtO}}_{2} }} }} \cdot \left\{ {a_{{{\text{Pt }}\left( {\text{s}} \right)}} \cdot p_{{{\text{O}}_{2} }} \cdot {\text{exp}}\left( {\frac{{ - \Delta G^{0}_{{\left[ {12} \right]}} }}{{{\text{R}}T}}} \right)} \right\}. $$

(15)

Table III Activity Coefficient at Infinite Dilution in Binary Systems at 1523 K^[29]

The PGM oxides are calculated as solids because thermodynamic data for liquids are not available. Tables IV and V list the calculated activity and activity coefficient of each oxide. Figures 7 and 8 depict the relationship between the activity and the molar fraction in the slag of the Cu–CuO_0.5 and Pb–PbO systems, respectively.

Table IV Activity and Activity Coefficient of the Cu–CuO_0.5 System at 1523 K

Table V Activity and Activity Coefficient of the Pb–PbO System at 1523 K

Fig. 7

Relationship between the activity and the mole fraction of CuO_0.5 in the CuO_0.5–SiO₂ and CuO_0.5–CaO systems

Fig. 8

Relationship between the activity and the mole fraction of PbO in the PbO–SiO₂ and PbO–CaO systems

Takeda et al.^[31] reported the activity of CuO_0.5 under MgO saturation at 1523 K for the CuO_0.5–SiO₂ system. Okajima et al.^[32] reported the activity of PbO at 1323 K for the PbO–CaO system. For the PbO–CaO system, experimental data could not be found, but solution thermodynamic data are available in FToxid in FactSage ver. 8.1.^[29] The results obtained in this study are in agreement with those in the aforementioned literature.^[29,31,32] The interaction between CuO_0.5 and SiO₂ is considered repulsive owing to the positively biased activity of CuO_0.5 in the CuO_0.5–SiO₂ system from the ideal value. In contrast, an attractive interaction is observed between CuO_0.5 and PbO because of their negatively biased activity in the CuO_0.5–CaO, PbO–SiO₂, and PbO–CaO systems.

Figures 9, 10, and 11 depict the activity of RhO_1.5, PdO, and PtO₂, respectively, as a function of the molar fraction of CuO_0.5 and PbO in the slags of the Cu–CuO_0.5 and Pb–PbO systems. The PGMs are speculated to be completely dissolved in the slag in the form of RhO_1.5, PdO, and PtO₂. In all combinations except for RhO_1.5 in the PbO–SiO₂ system, the activity coefficient of the PGM oxides increased with the coexistence of SiO₂ or CaO. This suggests that these PGM oxides are repulsive to SiO₂ and CaO in the slag.

Fig. 9

Activity coefficient of RhO_1.5 (s) at 1523 K: (a) CuO_0.5–SiO₂ and CuO_0.5–CaO systems and (b) PbO–SiO₂ and PbO–CaO systems

Fig. 10

Activity coefficient of PdO (s) at 1523 K: (a) CuO_0.5–SiO₂ and CuO_0.5–CaO systems and (b) PbO–SiO₂ and PbO–CaO systems

Fig. 11

Activity coefficient of PtO₂ (s) at 1523 K: (a) CuO_0.5–SiO₂ and CuO_0.5–CaO systems and (b) PbO–SiO₂ and PbO–CaO systems

Conclusions

The distribution behavior of Rh, Pd, and Pt during the oxidation smelting process was investigated for the recovery of PGMs. The effects of SiO₂ and CaO on the distributions of Rh, Pd, and Pt in the Cu–CuO_0.5 and Pb–PbO systems were investigated. The major conclusions are as follows:

• 1. MgO barely dissolved (a maximum of 0.34 mass pct) in the CuO_0.5, CuO_0.5–CaO, PbO, and PbO–CaO slags. In contrast, dissolution of MgO increased with increasing concentration of SiO₂ in the CuO_0.5–SiO₂ and PbO–SiO₂ slags.

• 2. The addition of SiO₂ or CaO reduced the oxygen partial pressure in the system. Furthermore, for all combinations of slag systems and PGMs, the addition of an optimum amount of SiO₂ or CaO reduced the concentrations of PGMs in the slags to less than approximately 1/10 of that without adding SiO₂ or CaO. The tendency for PGMs dissolved in the slags to increase with increasing oxygen partial pressure ^{[6,8,12,13,14,16,22]} and with increasing concentrations of Cu or Pb in the slags^{[12,13,16,22,24]} is consistent with previous studies.

Therefore, the addition of SiO₂ or CaO could reduce the loss of PGMs in slag during the oxidation smelting process as well as the amount of PGMs circulating in the recycling process.

• 3. The activity coefficient of each oxide in the slag was calculated, assuming the oxidation states of PGMs in the slag as RhO_1.5, PdO, and PtO₂. The addition of SiO₂ and CaO significantly changed the activity coefficient of each PGM oxide in the slag.