Recovery of Palladium and Silver from Copper Sludge and Spent Petrochemical Catalysts via Effective Pyrometallurgical Processing

Abstract

Copper-containing sludge and spent petrochemical catalyst (SPC) were investigated for recovering palladium (Pd) and silver (Ag). Increasing the mixing ratio of alumina-based SPC leads to reduced recovery rates at 1500 °C. Specifically, as the SPC mixing ratio increases from 10% to 30%, the recovery rate of Pd and Ag sharply decreases to 62.1% and 91.0%, respectively. This is attributed to an increase in the slag viscosity as well as to the higher sulfur content in the metal phase by decreasing the CaO/Al₂O₃ ratio of the slag. An increase in the slag viscosity causes a decrease in the metal recovery, as it lowers the settling velocity of metal droplets, resulting in imperfect metal separation, i.e., an increase in physical loss. Additionally, the presence of sulfur at the slag–metal interface was found to reduce interfacial tension, facilitating the entrapment of copper droplets within the slag. This further hindered phase separation and contributed to an increase in physical loss. This study highlights that physical loss is more serious in metal recovery rather than chemical loss, which is dependent on the thermochemical solubility of the target metals in the slag. The results emphasize the need for the precise control of slag properties to maximize the metal recovery processes in conjunction with a mitigation of CO₂ emissions.

1. Introduction

Rapid industrialization has resulted in the production of millions of tonnes of waste worldwide. The global production of electronic waste in 2022 reached approximately 62 million tonnes (Mt), with only 22.3% being documented as properly managed through collection and recycling processes [1]. The management of solid and household waste around industrial sites is often poorly designed, resulting in disposal in landfills. Such poor management methods threaten human health and can lead to severe ecological harm. The improper management of waste can result in contamination from metals, which, upon leaching into the environment, can cause significant ecological problems. However, these wastes also serve as potential sources of valuable metals, functioning as anthropogenic ores for the recovery of heavy metals. Spent catalysts, discarded batteries, red mud, municipal solid waste incineration ash, electroplating residue (sludge), and electronic scrap (e-scrap) contain a variety of elements, including Au, Ag, and Pd, which are ideal for recycling [2].

Recycling valuable metals is more energy-efficient and environmentally friendly than extracting metals from natural ore. As shown in Figure 1, producing valuable metals like Pd and Ag releases 9380 tons and 144 tons of CO₂ per ton of metal, respectively [3,4]. Consequently, the primary production of metals needed for electrical and electronic equipment (EEE) from natural resources results in the emission of 23.4 million tonnes of CO₂, which is nearly 1/2000 of the global CO₂ emissions. Consequently, approaches to managing industrial waste, including reuse, remanufacturing, and materials recovery, have received significant interest. These strategies enhance long-term economic efficiency and enable metal recovery enterprises to achieve sustained profitability [5]. Specifically, as shown in Figure 2, the recovery rate of valuable metals from the end-of-life (EOL) material sources are relatively high (90–99%) in building and construction, and in the automotive industries, engineering alloys, and lead batteries. However, the recovery of Al from packaging and that of Cu, Au, Ag, and Pd from e-scrap is not more than about 60% and 35%, respectively.

The annual production of municipal solid waste worldwide is expected to reach around 3.5 billion tonnes by 2050 [6], which highlights the necessity for improved recovery technologies. Various methods, such as pyrometallurgical, hydrometallurgical, electrochemical, and bio-hydrometallurgical processes, are used to extract valuable metals, which are crucial for reducing the environmental impact of waste and for utilizing valuable resources from these waste streams. Pyrometallurgical recovery entails the thermal treatment of ore and waste materials containing metals, which results in physical and chemical transformations that enable the extraction of valuable metals.

From the perspective of pyrometallurgical processing, extensive studies have been conducted on the behavior and recovery of valuable metals. Avarmaa et al. [7] conducted a study to determine the distribution coefficients for the valuable metals Rh, Ag, Pt, Au, and Pd between slag and copper matte within FeO_x/SiO₂ slag at 1250–1350 °C. In their findings, the solubility of silver in the slag ranged from 40 to 80 ppm, and for other valuable metals, solubility in iron silicate slag ranged from 1 to 20 ppm, decreasing with increasing matte grade. The study found that the distribution coefficients of these metals increased with a matte grade and were minimally affected by temperature. The highest distribution coefficients were observed for rhodium and the lowest for silver. Wan et al. [8] investigated the behavior of copper concentrate, iron oxide–silica slag, and PMs (Au, Ag, Pt, and Pd) by pre-mixing and reacting them at a typical copper smelting temperature of 1300 °C in both air and argon atmospheres. They reported that, as the molten matte and slag began to form, the studied valuable metals almost instantly migrated to the matte phase. The order of the distribution coefficients $L^{\raisebox{1ex}{$m$}\!\left/ \!\raisebox{-1ex}{$s$}\right.}\left(Me\right)$ in both air and argon atmospheres was observed to be Pd, followed by Pt, Au, and Ag.

Jiang et al. [9] carried out smelting by mixing the leaching residue with NiSO₄, a reducing agent such as carbon, a slagging agent (e.g., CaO, SiO₂, and Al₂O₃), and flux (e.g., Na₂CO₃, CaF₂ and Na₂B₄O₇) in various ratios at 1400 °C for 2 h. They then performed acid leaching on the obtained alloy to determine the recovery rates of PGMs, which were 90.13% for Pt, 91.02% for Pd, and 90.54% for Rh under optimal smelting conditions. After acid leaching, the enrichment ratio of PGMs increased by 8.66 times compared to the raw materials. Ghodrat et al. [10] investigated the thermodynamic analysis of secondary copper smelting using PCBs, and predicted the recovery rates for two defined scenarios, SCE1 (copper recycling with addition of PCB waste) and SCE2 (general copper recycling) at temperatures of 1100–1600 °C and oxygen partial pressures of p(O₂) = 10⁻⁷–10⁻¹⁰ atm. The study found that the copper recovery rate in the SCE1 scenario was 83.3%, higher than the 78.8% in the SCE2 scenario. Additionally, the recovery rates of valuable metals (Ag, Au) were predicted to be 88.5% and 96.5%, respectively. The analysis highlighted that optimal reduction conditions ($p_{O_2}={10}^{-8}\mathrm{atm}$) and temperatures above 1300 °C are desirable for the process.

Park [11] conducted a fundamental study on the recovery behavior of Au, Ag, and Ni using CaO–Al₂O₃–SiO₂ and CaO–SiO₂–CaF₂ slag from waste PCB at high temperatures. This study aimed to extract valuable metals like silver and gold, along with essential rare metals such as nickel. The calcium fluorosilicate slag containing CaF₂ was controlled to have a lower melting point and viscosity than calcium aluminosilicate slag, resulting in higher recovery of Au, Ag, and Ni. Kim et al. [12] tested mixtures of incinerated PCB and automobile catalysts in ratios of 70:30, 78:22, and 84:16, melting at 1250 °C for 60 min. The most effective slag composition was 28CaO–45SiO₂–17Al₂O₃–10MgO (wt%), with recoveries of 99.0% for Au, 98.0% for Pd, and 98.5% for Pt. They demonstrated that valuable metals are recoverable from spent PCB and automobile catalysts by utilizing metals from PCB as a collecting bath and automobile catalysts as a slag former. Park et al. [13] conducted a smelting experiment to recover valuable metals, such as gold, silver, nickel, and tin, from the insoluble residue of waste mobile phones. They used different synthetic flux compositions with varying CaO/SiO₂ ratios to control the slag composition. A high-silica flux (34CaO/61SiO₂/5Al₂O₃, wt%) showed poor recovery rates for Au, Ni, and Sn because of its high viscosity. Martinez et al. [14] investigated the recovery of platinum group metals (PGM) from waste automobile catalyst monolithic honeycomb using a method combining pyrometallurgical and electrolysis steps. The pyrometallurgical step enriched the PGM and prepared conductive material for the electrolysis, which was performed in a molten chloride electrolyte, achieving a recovery rate of nearly 100% PGM.

The present study focused on extracting valuable metals from industrial waste using a pyrometallurgical process. Two industrial waste materials, i.e., copper-containing sludge and spent petrochemical catalyst (SPC) were used to recover silver and palladium, respectively. We focused on the physicochemical properties of the slag that affect the recovery rates of target elements.

2. Experimental Procedure

The industrial waste used in the present study was copper sludge and spent petrochemical catalyst (SPC), with an image of each shown in Figure 3. The X-ray diffraction patterns of each material are shown in Figure 4. The chemical compositions of the industrial waste were examined using an inductively coupled plasma atomic emission spectroscopy (ICP-AES, SPECTRO ARCOS, Kleve, Germany) and a combustion analyzer (CS800, ELTRA, Haan, Germany). Details of the findings are presented in

Table 1. Composition of copper sludge and spent petrochemical catalysts (wt%).

	Al2O3	CaCO3	Cu2O	Fe2O3	SnO2	P2O5	SiO2	MgO	C	S	Ag	Pd
Copper sludge	0.6	29.9	24.7	22.3	6.7	3.5	2.6	1.7	5.6	2.4	0.014	-
Spent catalyst	99.7	-	-	-	-	-	-	-	-	-	-	0.32

. The Al₂O₃-based SPC was mixed with copper-containing sludge at a proportion varied from 10% to 30% SPC. The objective of this mixing condition is to ensure that the Al₂O₃ content in the slag does not exceed 30 wt% in order to avoid a significant rise in the melting point of the slag.

Three experiments were conducted utilizing a high-frequency induction furnace equipped with a graphite heating component. A schematic diagram of the experimental apparatus is presented in Figure 5. To ensure efficient heating, 100 g of a blend consisting of copper sludge and SPC, 60 g of copper powder to create a 5 mm melting pool, and about 3 g of solid carbon to facilitate the reduction reaction were placed in a fused magnesia crucible (60 × 50 × 120, mm) with a graphite heating element. The furnace was supplied with purified Ar gas at a flow rate of 500 mL/min, regulated by a mass flow controller. After maintaining the reduction reaction for 60 min at 1500 °C, the samples were cooled in an Ar gas atmosphere.

To perform the chemical analysis, the slag was pulverized to a fine powder with particle sizes lower than 100 μm. Valuable metal concentration and the final composition of the slag were analyzed using ICP-AES and CS800.

For SEM imaging of the slag, samples were cold mounted in molds and the surfaces were ground using sandpaper. Metal droplets dispersed in the slag were examined with field emission scanning electron microscopy and energy-dispersive X-ray spectroscopy (FE-SEM and EDS, Nova Nano SEM 450, FEI, Waltham, MA, USA).

3. Results and Discussion

All analytical data for the metal ingots and slag obtained from the smelting experiments are listed in

Table 2. Composition of metal ingot after the experiments.

Run ID		Metal Composition (wt%)					(ppm)
		Cu	Fe	Sn	S	P	Pd	Ag
SPC 10%	Cu-rich phase	71.45	20.27	7.25	0.61	0.36	410	150
SPC 20%	Cu-rich phase	79.41	9.95	9.03	0.82	0.59	710	1108
	Fe-rich phase	17.38	72.24	0.18	0.27	9.93	-	-
SPC 30%	Cu-rich phase	81.86	8.77	7.14	1.56	0.58	970	100
	Fe-rich phase	20.44	68.79	0.45	0.36	9.96	-	-

and

Table 3. Composition of slag phase after the experiments.

Run ID	Slag Composition (wt%)								(ppm)
	T.Cu	MgO	Al2O3	SiO2	CaO	FeO	P2O5	S	Pd	Ag
SPC 10%	0.63	6.8	20.1	5.4	52.4	7.0	5.6	2.1	0.1	0.2
SPC 20%	0.77	3.0	35.2	4.8	47.4	3.8	2.8	2.2	0.3	0.9
SPC 30%	1.79	3.0	46.5	2.8	35.5	5.3	2.3	2.8	60.0	1.0

, respectively.

After the experiment, quenched samples were obtained. As shown in Figure 6, the metal ingot, primarily composed of a copper-rich phase and an iron-rich phase, and the slag were successfully separated. The amount of copper present in the slag is crucial because copper metal droplets significantly influence the overall metal recovery rate.

In the current study, metal recovery was evaluated by employing Equation (1) [11,12,13,14,15,16].

$$\mathrm{Recovery} \mathrm{rate} \mathrm{of} i \left(\%\right)={\displaystyle \frac{{\left[\mathrm{wt}\%i\right]}_{\mathrm{metal}}}{{\left[\mathrm{wt}\%i\right]}_{\mathrm{metal}}+{\left(\mathrm{wt}\%i\right)}_{\mathrm{slag}}}}\times 100$$

(1)

The recovery rate of Pd and Ag calculated from Equation (1) and the content of each element in metal ingot and slag given in Table 2 and Table 3 is presented in Figure 7. It is demonstrated that, as the SPC ratio increases, the recovery rate of Pd and Ag decreases. A decreasing tendency in the recovery rate is significant in the case of the Pd, i.e., it drops sharply when the SPC ratio increases to 30%. This trend aligns with the changes in copper content in the slag, i.e., the total copper content in the slag increases from about 0.6 wt% to 1.8 wt% with increasing the ratio of SPC from 10% to 30% (Table 3). Hence, it is suggested that the copper content in the slag is a critical factor influencing the recovery rates of valuable metals.

The copper content in the slag is closely related to the slag viscosity, which directly impacts the efficiency of physical separation between the slag and the metal. In particular, an increase in the SPC ratio results in a decrease in the CaO/Al₂O₃ (=C/A) ratio of the slag from C/A = 2.6 to 0.8 (Table 3), and this change likely influenced the viscosity and physical properties of the slag. Several researchers have investigated how the structure affects the viscosity of calcium aluminosilicate melts with low silica content (SiO₂ < 10 wt%), which is similar to the present slag system [17,18,19]. It was found that increasing the C/A ratio causes a sharp decrease in the proportion of polymerized aluminate structure unit $Q_{ Al}^4$ (non-bridging oxygen per aluminum, NBO/Al = 0), while the amounts of simpler units $Q_{ Al}^2$ (NBO/Al = 2) and $Q_{ Al}^3$ (NBO/Al = 1) rise, which in turn reduces the viscosity of the slag. Consequently, the relatively low C/A ratio of the slag in the 30% SPC led to a substantial rise in viscosity, which trapped the copper droplets within the slag.

The smelting process uses gravitational forces to separate liquid metal from slag. The efficiency of metal recovery is determined by physical separation along with chemical reactions. A lower C/A ratio for the slag in the 30% SPC mixing condition resulted in the higher viscosity, which in turn led to a decreased recovery rate for valuable metals. The copper droplets entrapped and dispersed in the slag phase indicates an incomplete metal separation form the slag, as shown in Figure 8. A comparison between the 20% SPC mixing condition (Figure 8a), which had a higher recovery rate (99.6% for Pd, 94.6% for Ag, Figure 7), and the 30% SPC mixing condition (Figure 8b), which had a lower recovery rate (62.1% for Pd, 91.0% for Ag, Figure 7), shows that the 30% SPC mixing condition resulted in the larger (several tens of micrometers with spherical morphology) copper droplets trapped in the slag compared to that of 20% SPC mixing condition with the relatively smaller (several micrometers with spherical morphology) copper droplets.

As shown in Figure 8, EDX analysis qualitatively confirmed that Pd and Ag have relatively high solubility in the liquid copper phase. Alternatively, the chemical solubility of Pd (=3 to 320 ppm) in the Fe_tO-SiO₂-CaO-Al₂O₃-MgO slag at 1600 °C was measured as a function of slag composition as well as oxygen partial pressure [20]. From the equilibrium solubility data, the following dissolution reaction was proposed.

$$\mathrm{Pd}\left(l\right)+{\displaystyle \frac{1}{4}}{\mathrm{O}}_2\left(g\right)+{\displaystyle \frac{3}{2}}{\left({\mathrm{O}}^{2-}\right)}_{\mathrm{slag}}={\left({\mathrm{PdO}}_2^{3-}\right)}_{\mathrm{slag}}$$

(2)

This reaction indicates that the chemical solubility of Pd in the slag increases with increasing oxygen partial pressure ($p_{{\mathrm{O}}_2}$) and the activity of free oxygen ion (${\mathrm{a}}_{{\mathrm{O}}^{2-}}$), i.e., which is qualitatively proportional to that of CaO (Equation (3)) at a fixed temperature, assuming that the activity of the ${\mathrm{Ca}}^{2+}$ ion would not be seriously changed by slag composition [20,21,22].

$$\left(\mathrm{CaO}\right)=\left({\mathrm{Ca}}^{2+}\right)+\left({\mathrm{O}}^{2-}\right), K_{\left(3\right)}={\displaystyle \frac{a_{{\mathrm{Ca}}^{2+}} · a_{{\mathrm{O}}^{2-}}}{a_{\mathrm{CaO}}}}$$

(3)

Therefore, from Equation (2), the chemical dissolution of Pd (as a form of ionic complex, ${\mathrm{PdO}}_2^{3-}$) in the slag decreases with decreasing C/A ratio, i.e., with increasing SCP mixing ratio in the present study. However, as shown in Figure 6, the recovery rate of Pd sharply decreased by increasing the mixing ratio of SPC to 30%. This means that the physical loss of Pd as a form of Cu(-Pd) metal droplets trapped and dispersed in the slag is much higher than the chemical loss of Pd due to an ionic dissolution based on Equation (2). A similar understanding is available for the changes in the recovery rate of Ag against the SPC mixing ratio.

The main reason for metal loss in commercial smelters is the physical entrapment of copper droplets in the slag [13,23]. Therefore, we calculated the terminal settling velocity of metal droplets in the slag. The relationship among gravitational, buoyant, and frictional forces acting on the particles was considered. To estimate the terminal velocity of metallic phases settling through molten slag, assuming spherical particles and a low Reynolds number (Re < 1) under laminar flow regime, Stokes’ law was applied, as shown in Equation (4) [11,12,13,24,25,26].

$$u_t={\displaystyle \frac{2}{9}}{\displaystyle \frac{r^2·\left({\rho }_{\mathrm{metal}}-{\rho }_{\mathrm{slag}}\right)·g}{\eta }}$$

(4)

where, $u_t$ denotes the terminal velocity (m/s), $\rho$ is the density (kg/m³), $r$ is the radius of the metal droplets (m), $g$ is the gravitational constant (m/s²), and $\eta$ is the slag viscosity (dPa·s = poise). In the present study, FactSage^TM (version 8.2, ESM Software, Hamilton, OH), a commercial thermochemical computing software, was used to calculate the slag viscosity. The software has been widely used to predict the physicochemical properties of the slag in ferrous and non-ferrous metallurgical systems [20,21,22,23,24,27,28,29,30,31,32,33,34,35,36,37,38,39,40,41,42,43]. The viscosities of the slag from 10%, 20%, and 30% SPC additions were 0.8 poise, 1.6 poise, and 2.5 poise at 1500 °C, respectively. Since copper was the primary element in the metal droplets, the density of liquid copper (7680 kg/m³) was used [44]. After the experiment, the slag composition was based on the CaO/Al₂O₃ based system (Table 3), the density values of the CaO/Al₂O₃ melt at 1500 °C were obtained from the literature, i.e., 2680 kg/m³, 2700 kg/m³, and 2750 kg/m³ for SPC ratios of 10%, 20%, and 30%, respectively [45]. Using Equation (4), the terminal velocities for metal droplets of various sizes (from 10 μm to 100 μm) in the CaO/Al₂O₃ based slag system were calculated. The results are presented in Figure 9. The most pivotal property in the settling process is slag viscosity. The settling velocity increases with increasing metallic particle size at a given slag composition and decreases by adding the SPC from 10% to 30%, which contributes to an increase in the viscosity of molten slag from about 0.8 to 2.5 poise.

Finally, as illustrated in Figure 10, the presence of sulfur at the interface between slag and metal was confirmed, substantiating its function as a surface-active element in liquid metals. The surface-active elements exert an influence on the interatomic bond at the metal surface, leading to a decrease in surface tension [46]. Then, the presence of sulfur at the slag–metal interface contributes to a reduction in the slag–metal interfacial tension. As the interfacial tension decreases, copper droplets become more readily entrapped within the slag due to an instability of the interface, impeding their efficient separation [47,48,49,50,51,52,53,54]. Consequently, this interfacial phenomenon is potentially another factor affecting the recovery rate of valuable metals in the smelting process. Hence, more specific investigations for the slag–metal interfacial phenomena are necessary in further works.

In summary, the thermophysical properties, such as slag viscosity, surface or interfacial tensions, in conjunction with thermochemical solubility of target elements in the slag, contribute to the recovery rate of valuable metals in terms of physical loss and chemical loss, respectively. To maximize the recovery of target metals, the lower slag viscosity, resulting in the higher terminal velocity of metal droplets, as well as the removal of sulfur, resulting in a higher surface or interfacial tension, are necessary for minimizing a physical loss. Alternatively, the lower solubility of target metals in the slag is required to minimize a chemical loss [20,24]. A graphical summary of the metal loss mechanism according to the slag properties is schematically shown in Figure 11.

4. Conclusions

The current study explored the optimal slag conditions for the pyrometallurgical processing to recover palladium (Pd) and silver (Ag) from hazardous industrial wastes such as copper containing sludge and spent petrochemical catalyst (SPC) at 1500 °C. The major findings can be summarized as follows.

• Effective separation of slag and metal is essential for maximizing metal recovery rate, which requires a high settling velocity of metal droplets. This study demonstrates that slag viscosity significantly influences settling velocity, with lower viscosity facilitating higher settling velocity, and thus more efficient phase separation and metal recovery.
• As the mixing ratio of SPC increases from 10% to 30%, the recovery rate of Pd and Ag sharply decreases to 62.1% and 91.0%, respectively. This is attributed to the increase in slag viscosity as well as to the higher sulfur content in the metal phase, resulting in a decrease in the slag–metal interfacial tension. The lower interfacial tension enhances the instability of the interface, resulting in an emulsification of copper droplets in the slag phase, and hence hindering the metal–slag separation, ultimately decreasing the recovery of Pd and Ag.
• Even though the chemical solubility of target precious metals (i.e., chemical loss) decreases with the decreasing value of the CaO/Al₂O₃ ratio of the slag (i.e., with increasing SPC mixing ratio), an increase in the viscosity of the slag, and thus lower settling velocity of metal particles in slag phase (e.g., 0.1 mm/s for 50 μm diameter droplets for 30% SPC mixing with slag viscosity of 2.5 Poise). results in the higher physical loss. Hence, from the present experimental findings, physical loss is a much more dominant factor affecting the metal recovery rate rather than chemical loss.