Selective Recovery and Enrichment of Cobalt from Cobalt-Containing Slag by Carbothermal Reduction

Abstract

Cobalt ore resources are relatively scarce; thus, the recycling of cobalt-containing slag is highly significant in the economy and society. In this study, the effects of reduction temperature, the reduction agent ratio, reduction time, and particle size on the grade and recovery rate of cobalt in a concentrate were systematically investigated during the carbothermal reduction of cobalt-containing slag. The results revealed that the grades of cobalt, iron, and copper in the concentrate after magnetic separation were 4.02%, 2.48%, and 81.33%, respectively, and the recoveries were 94.17%, 74.80%, and 53.27%, respectively, under the reduction temperature of 1150 °C, the reduction agent ratio of 40%, the reduction time of 2 h, and the particle size of −3.0 mm. Furthermore, through static reduction roasting in a muffle furnace and dynamic reduction roasting in a rotary kiln followed by magnetic separation, a stable cobalt grade, high selective recovery, and effective enrichment were achieved under optimal conditions.

1. Introduction

Cobalt is an important, strategic non-ferrous metal widely used in fields such as battery materials, high-temperature alloys, hard alloys, magnetic materials, and catalyst [1,2]. With the increasing global demand for clean energy and electric vehicles, the status and future prospects of cobalt resources have attracted widespread attention [3,4].

From the perspective of cobalt ore resources, pure cobalt ores are rare, and cobalt is almost entirely present as a by-product in associated ores, with a very low concentration at 25 ppm in the Earth’s crust [5,6]. The main types of cobalt ores include the following: sandstone-type copper ores, lateritic nickel ores, and magmatic nickel–copper sulfide deposits [7]. Currently, the majority of the globally proven cobalt resources are contained in nickel laterite deposits, with the remainder mostly found in nickel–copper sulfide deposits [8,9]. These deposits are primarily located in countries such as Australia, Canada, and Russia [10]. Additionally, cobalt resources in the Democratic Republic of Congo (DRC) and the Republic of Zambia are mainly found in sedimentary copper deposits [11]. The globally proven terrestrial cobalt resources amount to approximately 25 million tonnes, with reserves of 7.2 million tonnes. During the smelting of these ores, cobalt is typically extracted as a by-product. In China, cobalt ore is distributed over a wide area [12], but cobalt resources remain relatively scarce, manifested in small reserves, low ore grades, a high proportion of low-grade ores, and a large number of associated ores [13,14,15]. This results in a certain degree of dependence on external sources for China’s cobalt supply [16]. Meanwhile, driven by the development of the domestic lithium battery and alloy smelting industries, China’s consumption of cobalt products has experienced rapid growth, increasing from 3800 tons in 2000 to 119,000 tons in 2022 [17,18]. Therefore, the recovery of cobalt from secondary resources, such as cobalt-containing slag, has become an urgent issue [19,20].

Cobalt-containing slag primarily originates from the smelting processes of cobalt and nickel ores, including high-temperature smelting, leaching, and electrolysis [21,22]. During the smelting process, the extraction efficiency of cobalt is often low, resulting in the generation of large amounts of cobalt-bearing waste slag [23]. Depending on the specific smelting process, the composition of the slag varies, typically containing various metal oxides (such as CoO, NiO, FeO, SiO₂, etc.) and other inorganic substances [24]. Recent studies have shown that the proper treatment and utilization of cobalt-containing slag can not only reduce waste accumulation but also enable resource recovery [25]. Currently, the treatment methods for cobalt-containing slag are classified into hydrometallurgical and pyrometallurgical processes [26,27]. Among them, pyrometallurgical processes are favored over hydrometallurgical ones due to their simplicity [28] and absence of extensive reagent consumption, and they remain the predominant approach for processing cobalt-containing slag [29,30]. The carbothermal reduction roasting process involves the addition of reductants to treat cobalt-containing slag at elevated temperatures. This promotes the selective reduction of cobalt and iron oxides, thereby creating favorable conditions for subsequent separation through magnetic or hydrometallurgical techniques [31,32].

This study investigates the process of selectively recovering cobalt from cobalt-containing slag via carbothermic roasting. The effects of reduction temperature, reductant ratio, reduction time, and particle size on the cobalt grade and recovery rate are examined, with both static reduction roasting in a muffle furnace and dynamic reduction roasting in a rotary kiln used for verification. Through analytical methods such as optical microscopy and scanning electron microscopy (SEM), the phase transformations during the roasting process are clarified, providing a foundational study for the industrial-scale comprehensive recovery of cobalt and other valuable metals from cobalt-containing slag.

2. Materials and Methods

2.1. Materials

The raw material of the cobalt-containing slag was obtained from the Zambia Chambishi Copper Smelting Co., Ltd., Kitwe, Zambia (CCS). The chemical compositions of the three samples were determined by ICP and XRF analyses, and the results are presented in

Table 1. Chemical composition of the cobalt-containing slag (wt. %).

Serial Number	Chemical Composition
	Si	Al	Ca	Mg	S	Cu	Co	Fe
1	13.96	0.516	0.456	0.353	0.060	1.01	1.27	47.29
2	14.07	0.444	0.420	0.360	0.058	1.03	1.29	46.92
3	14.00	0.964	0.438	0.354	0.061	1.03	1.27	47.06

. As listed in Table 1, the main metal elements in cobalt-containing slag are Fe, Cu, and Co, with an Fe content of 46.92~47.29%, a Cu content of 1.01~1.03%, and a Co content of 1.27~1.29%. It illustrates that Fe, Cu, and Co are uniformly distributed in the slag. The mineral content of cobalt-containing slag was statistically analyzed by optical microscope, and the results are shown in

Table 2. Distribution of cobalt in cobalt-containing slag (%).

Mineral	Content	Cobalt Distribution Rate
Cobalt–nickel sulfide	0.10	2.52
Copper sulfide	1.87	0
Magnetite	5.01	0.49
Pyrrhotite	0.02	0.06
Metallic copper	0.05	0
Iron-containing silicate	77.88	96.93
Vitreous mineral	15.07	0
Cobalt–iron alloy	-	-

. It can be observed that the primary minerals in the cobalt-containing slag are iron silicate minerals, glassy minerals, magnetite, and copper sulfide minerals.

The XRD analysis of the cobalt-containing slag is displayed in Figure 1, where the main minerals are iron-bearing silicate minerals, magnetite, and cobalt-bearing magnetite. Cobalt-containing slag was analyzed using a scanning electron microscope (SEM) and an energy-dispersive spectrometer, and the results are shown in Figure 2. It can be seen that the phases of Fe and Co are highly overlapped and have a good correlation, indicating that there is a dispersion of Co in the ferrosilicate minerals. Si and Co also have a certain correlation, indicating that the Co distribution in the vitreous minerals may be lower.

The reduction agent used in the reduction roasting experiments of the cobalt-containing slag was anthracite. Its industrial analysis was carried out in accordance with the [33], and the results are presented in

Table 3. Industrial analysis results of the reduction agent.

Fixed Carbon/wt. %	Volatile Matter/wt. %	Ash Content/wt. %	Moisture Content/wt. %
67.21	17.76	13.48	1.55

. The high fixed carbon content of the qualified anthracite ensures the efficient reduction of cobalt oxides, while the moderate volatile matter content contributes to establishing a reducing atmosphere. However, the ash content of 13.48% requires attention due to its potential impact on the melting behavior of the slag during the roasting process. These characteristics indicate that this reduction agent is suitable for the reduction roasting of cobalt-containing slag.

2.2. Reduction Roasting Experiments and Analysis Method

2.2.1. Pretreatment of Cobalt-Containing Slag

The results of the particle size screening experiment of the cobalt-containing slag are seen in

Table 4. The results of the particle size screening experiment.

Particle Size	Weight/g	Distribution Rate/%	Positive Accumulation/%	Negative Accumulation/%
+4.00	17.67	1.39	1.39	100.00
−4.00 + 3.35	40.05	3.15	4.54	98.61
−3.35 + 3.00	117.81	9.27	13.81	95.46
−3.00 + 2.00	298.58	23.50	37.31	86.19
−2.00 + 1.00	260.19	20.48	57.79	62.69
−1.00 + 0.5	186.05	14.64	72.43	42.21
−0.5 + 0.15	133.12	10.48	82.91	27.57
−0.15 + 0.106	77.36	6.09	89.00	17.09
−0.106 + 0.075	54.74	4.30	93.30	11.00
−0.075	85.02	6.70	100.00	6.70
Total	1270.59	100.00	—	—

. The proportion of cobalt-containing slag with a particle size above +3 mm is 13.81%, and the proportion of cobalt-containing slag with a particle size of −3 mm to +1 mm is higher, accounting for 43.98%. Since the coarse particle size of raw cobalt-containing slag fails to meet the requirements for subsequent reduction experiments, the samples need to be pretreated. The pretreatment process of an experimental sample of cobalt-containing slag is shown in Figure 3. The cobalt-containing slag was crushed by a jaw crusher (RK/PEF 100 × 100, Wuhan Exploration Machinery Co., Ltd., Wuhan, China) and a roller crusher (RK/PG-φ200 × 125, Wuhan Exploration Machinery Co., Ltd., Wuhan, China), resulting in a particle size of −3 mm, with 80% of the cobalt-containing slag falling within this range. The cobalt-containing slag was thoroughly mixed to ensure the uniform distribution of particles of different sizes. The mixture was then split using a sample splitter, and samples were taken for further analysis.

2.2.2. Reduction Roasting Experiments

In each experiment, 50 g of cobalt-containing slag was mixed with a reducing agent, with anthracite used as the reductant. The amount of anthracite was varied between 30% and 50% of the weight of the cobalt-containing slag. The thoroughly mixed sample was placed in a crucible and then transferred to a muffle furnace. The furnace was heated to the experimental target temperature at a heating rate of 10 °C/min at room temperature, where carbothermal reduction roasting was carried out. After reduction roasting, the sample was taken out at the experimental target temperature and naturally cooled to room temperature in air. The roasting product was then crushed and ball-milled to achieve a particle size distribution where 75% passed −0.075 mm, while the remaining 25% ranged between −2 mm and −0.075 mm. Finally, the cobalt concentrate was obtained through magnetic separation, with the magnetic field intensity set to 95.54 kA/m. The experimental process of reduction roasting is shown in Figure 4.

Static reduction experiment: The 300 g of cobalt-containing slag was mixed with 90–150 g of anthracite. The mixture was placed in an 800 mL crucible and transferred to a muffle furnace. The furnace was heated to the target temperature at a rate of 10 °C/min to perform carbothermal reduction roasting. After roasting, the sample was removed and allowed to cool naturally. Rotary kiln dynamic experiment: The 500 g of cobalt-containing slag was mixed with 150–250 g of anthracite and placed in a laboratory rotary kiln with a diameter of 90 mm. The kiln was heated to the target temperature at a rate of 10 °C/min to carry out dynamic reduction roasting. After roasting, the sample was removed and allowed to cool naturally. The roasted samples obtained from both static and dynamic reduction roasting were subjected to ball-milling to achieve a particle size of −0.075 mm (accounting for 75% of the total particles). Magnetic separation was then performed at a magnetic field intensity of 95.54 kA/m to obtain cobalt concentrate.

The cobalt, iron, and copper contents in the cobalt-containing slag and the magnetic separation concentrate were measured to calculate the recovery rates of Co, Fe, and Cu. The measured concentrations of Co, Fe, and Cu in the cobalt concentrate represent the grades of these elements. The recovery rates of Co, Fe, and Cu were calculated using the following:

$$\mathrm{R}=1-{\displaystyle \frac{{\mathrm{m}}_1{\mathrm{c}}_1}{{\mathrm{m}}_0{\mathrm{c}}_0}}$$

where m₀ and m₁ represent the mass of the cobalt-containing and residue slag, respectively; c₀ represents the Co, Fe, and Cu contents in the converter slag; c₁ represents the Co, Fe, and Cu contents in the residue slag.

2.2.3. Analytical Methods

The chemical component of the cobalt-containing slag samples was determined by inductively coupled plasma atomic emission spectrum (ICP-AES, IRIS Advantage Radial, PerkinElmer, Waltham, MA, USA) and X-ray fluorescence spectroscopy (XRF, Axios advanced, Thermo Fisher Scientific, Waltham, MA, USA). And the main phase compositions of the cobalt-containing slag were detected by an X-ray diffractometer (XRD, Simens D500 automatic X-ray diffractometer, Siemens AG, Berlin, Germany) with a copper target and operated at 40 kV and 250 mA in step mode with a 0.02° 2θ step and a count time of 0.5 s per step over a 2θ range from 10° to 80°. The morphology and the distribution of elements in cobalt-containing slag and roasting products were observed under a scanning electron microscope (SEM, JEOL, Tokyo, Japan) equipped with an energy-dispersive spectrometer (EDS, BRUKER, Bremen, Germany), operated at 20.0 kV after gold spraying of the samples. The cobalt-containing slag and roasting product samples were observed in the optical microscope (Leica DM4500P, Leica Camera AG, Solms, Germany) by using the reflected plane polarized light. According to the different colors of minerals and pores presented in the optical micrographs, the software of Image-Pro Plus 6.0 was used to quantitatively determine the mineral compositions via the area calculation method.

3. Results and Discussion

3.1. Thermodynamic Analysis

The composition of cobalt-containing slag is complex, and the reactions of Fe₃O₄, FeO, Fe₂SiO₄, CoFe₂O₄, CoO, and CuO with fixed carbon at atmospheric pressure are shown in Equations (1)–(6). The Gibbs free energy variations in these reactions with temperature were calculated using HSC Chemistry 6 software, and the results are presented in Figure 5a. Based on the Gibbs free energy values, the initial reduction temperatures of Fe₃O₄, FeO, and CoO by carbon were determined to be 675 °C, 722 °C, and 498 °C, respectively. Theoretically, CuO can be reduced to metallic Cu at room temperature. Furthermore, according to the equilibrium diagrams in Figure 5c,d, the primary attachment carriers of Co, such as iron silicates and cobalt–iron oxides, can undergo reduction reactions at 792 °C and 722 °C, respectively, reducing Co and Fe to their metallic forms.

During the reduction process, carbon undergoes the Boudouard reaction, generating CO gas (Equation (7)), which subsequently acts as a reducing agent for metal oxides, as illustrated in Equations (8)–(11). As shown in Figure 5b, calculations indicate that the Boudouard reaction becomes thermodynamically favorable at temperatures above 700 °C. As the equilibrium partial pressure of CO₂ in the system increases, it promotes the gasification of carbon, shifting the reduction pathway of metal oxides from solid–solid reduction to gas–solid reduction, thereby enhancing the overall reduction efficiency.

2Fe₃O₄ + C = 6FeO + CO₂(g)

(1)

FeO + C = Fe + CO

(2)

Fe₂SiO₄ + 2C = 2Fe + SiO₂ + 2CO(g)

(3)

CoFe₂O₄ + 4C = Co + 2Fe + 4CO(g)

(4)

CoO + C = Co + CO(g)

(5)

CuO + C = Cu + CO(g)

(6)

C + CO₂(g) = 2CO(g)

(7)

Fe₃O₄ + CO(g) = 3FeO + CO₂(g)

(8)

3FeO + CO(g) = Fe+ CO₂(g)

(9)

CoO + CO(g) = Co + CO₂(g)

(10)

CuO + C = Cu + CO₂(g)

(11)

3.2. Effect of Reduction Temperature

The selection of the reduction temperature is crucial. The reasonable selection of the reduction temperature is directly related to the metal grade and metal recovery rate of the magnetic separation concentrate. When the reduction agent ratio was fixed at 40%, the reduction time was 2 h, and the particle size of the cobalt-containing slag was −3 mm, the effect of the reduction temperature on the recovery rates of Co, Cu, and Fe and the grades of Co, Cu, and Fe was investigated. The result is shown in Figure 6.

It can be observed that with the increase in temperature, the grades of Co, Cu, and Fe showed an upward trend, reaching high points of 4.02%, 2.48%, and 81.33%, respectively, at 1150 °C. As the temperature rose, the grades of Co, Cu, and Fe all decreased. The recovery rates of the Co, Cu, and Fe metals increased with the increasing temperature, but the increase was smaller between 1150 °C and 1200 °C. The reduction in cobalt from its oxide form was nearly complete at between 1000 °C and 1150 °C, resulting in a sharp increase in the cobalt recovery rate. Therefore, considering both the metal grades and metal recovery rate indicators, the reduction temperature was selected as 1150 °C.

3.3. Effect of Reduction Agent Ratio

The appropriate reduction agent ratio can regulate the reduction degree of Co and Fe to increase the enrichment ratio of cobalt concentrate. The effects of the reducing agent ratio on the recoveries of Co, Cu, and Fe and the grades of Co, Cu, and Fe were investigated when the reduction temperature was 1150 °C, the reduction time was 2 h, and the particle size of the cobalt-containing slag was −3 mm.

It can be seen from Figure 7 that with the increase in the reduction agent ratio, the Co grade of concentrate has a slight change of about 4%, which shows a slight downward trend. The Co grade gradually decreased as the reducing agent ratio increased to 40% and then remained unchanged. The Fe grade reached 81.33% when the reducing agent ratio was 40%. The recovery rates of Co, Cu, and Fe first gradually increased and then decreased with the increase in the amount of reduction agent added. When the reduction agent ratio was 40%, the highest recovery rates of Co, Cu, and Fe were 94.17%, 74.8%, and 53.27%, respectively. When the reduction agent ratio exceeded 40%, the excessive reduction of iron oxides led to the formation of non-magnetic phases, which are hard to separate during the subsequent magnetic separation process, resulting in a decrease in the iron grade. Once cobalt oxide is reduced to metallic cobalt, its stability is high, making it resistant to further reduction into other lower-valent compounds, thus maintaining a stable cobalt grade. Therefore, the reduction agent ratio was determined to be 40%.

3.4. Effects of Reduction Time

Metal oxides begin to be reduced under specific temperature conditions, and the extent of reduction is closely related to the reduction time. When the reduction temperature is fixed at 1150 °C, the reduction agent ratio is 40%, and the particle size of the cobalt-containing slag is −3 mm, the effect of the reduction time on the recovery rates of Co, Cu, and Fe and the grades of Co, Cu, and Fe is investigated.

It can be seen from the analysis of Figure 8 that the grades of Co, Cu, and Fe showed an upward trend before 2 h, reached the highest of 4.02%, 2.48%, and 81.33%, respectively, at 2 h, then showed an inflection point at 2 h, and then declined. The recovery rate of Cu and Fe was increasing. The recovery rate of Co first increased with time and then decreased slightly after 2 h. Therefore, the reduction time was determined to be 2 h.

3.5. Effects of Particle Size

As can be seen from the data in Figure 9, there is no significant difference in the reduction roasting effect when the material particle size is −3 mm, −6 mm, −10 mm, or −13 mm. The Co grade of the concentrate and the Co metal recovery rate are basically the same. The Co grade is basically around 4%, and the Co recovery rate is around 95%. Smaller particle sizes generally improve recovery efficiency by providing a larger reaction surface area and enhancing the diffusion capacity of the reducing agent. However, within the particle size range of −3 mm to −13 mm, no significant difference in cobalt recovery was observed. This suggests that within this range, the particle size of the cobalt-containing slag has little impact on the efficiency of the reduction process. The Cu and Fe grades also do not change much. The Cu and Fe recovery rates first increase and then slightly decrease as the particle size of the cobalt-containing slag decreases. Considering the metal grade and recovery rate, in order to maintain the consistency of the experiment, the −3.0 mm material particle size was selected.

3.6. Reduction Verification Experiment

Under the optimal conditions of the reduction temperature of 1150 °C, the reduction agent ratio of 40%, the reduction time of 2 h, and the particle size of −3 mm, the static reduction experiment of three samples of cobalt-containing slag was carried out in the muffle furnace, and the experiment results are shown in Figure 10. As can be seen from Figure 10, the cobalt grade is between 3.93 and 4.30%, and the recovery rate is between 90.49% and 92.66% in the three static reduction experiments under the best conditions. At the same time, the rotary kiln dynamic reduction experiment was carried out under the same optimal reduction conditions. It can be seen in Figure 11 that the cobalt grade and cobalt recovery rate reached 4.26–4.41% and 90.67–93.67%, respectively. The main reason for the better results of the dynamic reduction experiment is that the mass and heat transfer of materials in the high-temperature reduction environment during the rotation of the rotary kiln are higher than in the muffle furnace. In summary, the static and dynamic experiment results are stable and can provide technical support for industrialization.

3.7. Migration of Cobalt

The cobalt-containing slag and the roasting product were detected by optical microscopy, and the mineral content was statistically analyzed. Then, the cobalt distribution rate was obtained by combining the statistical results of mineral content and scanning electron microscopy spectroscopy, and the results are shown in

Table 5. Distribution of cobalt in cobalt-containing slag and roasting product (%).

Mineral	Content of Cobalt-Containing Slag	Cobalt Distribution Rate	Content of Roasting Product	Cobalt Distribution Rate
Cobalt–nickel sulfide	0.10	2.52	-	-
Copper sulfide	1.87	0	0.75	0.45
Magnetite	5.01	0.49	0.83	0
Pyrrhotite	0.02	0.06	-	-
Metallic copper	0.05	0	0.08	0.02
Iron-containing silicate	77.88	96.93	51.13	15.79
Vitreous mineral	15.07	0	28.87	1.64
Cobalt–iron alloy	-	-	18.34	82.10

and Figure 12. It can be seen that the main minerals of cobalt-containing slag are iron-containing silicate, vitreous minerals, magnetite, and copper sulfide minerals, followed by a small amount of cobalt–nickel sulfide, copper metal, and pyrrhotite. The iron-containing silicate and vitreous minerals in cobalt-containing slag are cemented to form a matrix, and copper sulfide minerals and magnetite are embedded in the matrix in granular forms of varying thicknesses. Cobalt–nickel sulfide and pyrrhotite are mostly wrapped in fine particles in copper sulfide minerals, and copper metal is distributed in copper sulfide minerals and the matrix.

As shown in the table, the cobalt in the cobalt-containing slag is mainly attached to the iron-containing silicate, and the distribution rate of cobalt is as high as 96.93%. The cobalt–iron alloy accounted for 18.34% of the total mineral content in the roasting products, the iron-containing silicate content was 51.13%, but the cobalt distribution rate was low. Cobalt was mainly distributed in cobalt–iron alloy after reduction roasting, accounting for 82.10%, which indicates that cobalt oxide is reduced and dissociated from iron-containing silicate at a high temperature during the reduction roasting process. Cobalt forms a strong magnetic cobalt–iron alloy with iron, which can be recovered by magnetic enrichment.

4. Conclusions

(1)

The primary minerals in cobalt-containing slag include iron silicate minerals, glassy minerals, magnetite, and copper sulfide minerals, with smaller amounts of cobalt-nickel sulfide minerals, metallic copper, and pyrrhotite. The main cobalt-bearing minerals are cobalt–nickel sulfide minerals, pyrrhotite, iron silicate minerals, and magnetite. Cobalt is predominantly hosted in the iron silicate minerals.

(2)

The optimal process conditions for the carbothermic reduction of cobalt-containing slag, based on experimental results, are as follows: a reduction temperature of 1150 °C, a reductant ratio of 40%, a reduction time of 2 h, a particle size of −3 mm, a grinding fineness of −0.075 mm with an 80% passing rate, and a magnetic field strength of 95.54 kA/m. The results show that the grades of cobalt, copper, and iron in the concentrate after magnetic separation are 4.02%, 2.48%, and 81.33%, respectively, and the recoveries are 94.17%, 74.80%, and 53.27%.

(3)

The static reduction in a muffle furnace and the dynamic reduction in a rotary kiln were validated under the optimal conditions, resulting in stable cobalt grade and recovery rates, which are suitable for industrial applications. The cobalt grade is between 3.93 and 4.30%, and the recovery rate is between 90.49% and 92.66% according to the static reduction experiment. The cobalt grade and cobalt recovery rate reached 4.26–4.41% and 90.67–93.67%, respectively, with the rotary kiln dynamic reduction experiment.