The process of sample treatment, direct reduction, and smelting separation are shown in Figure 1. First, copper slag, lignite, and CaO were thoroughly blended. The mass ratio of copper slag and lignite is based on the molar ratio of “O” and “C”. The mass ratio of the copper slag to CaO is based on the molar ratio of SiO₂ and CaO. In this experiment, the molar ratios of C/O and CaO/SiO₂, regarded as the optimum reaction conditions, were all 1.0, and based on previous experimental results (Zuo et al., 2017; Zuo et al., 2020b). The masses of lignite and CaO were calculated based on Eqs 7, 8. A carbonaceous pellet with copper slag and CaO was prepared in a spherical shape using a ball-press machine. The maximum diameter of the pellets was 21 mm. The mass of each pellet was approximately 10 g. The pressure setting of the briquetting machine was set to 30 MPa. m lignite = FeO% / 100 ×m copper   slag × 12 / 72 + Fe 3 O 4 % / 100 ×m copper   slag × 48 / 232 (7) m CaO = SiO 2 % / 100 ×m copper   slag × 56 / 60 (8) where m _lignite is the mass of lignite, g; m _{copper slag} is the mass of copper slag, g; m _CaO is the mass of CaO, g; Fe₃O₄% is the mass percentage of Fe₃O₄ in copper slag, and %; FeO% is the mass percentage of FeO in the copper slag, %.

The pellets were then placed in a corundum crucible in a reduction furnace. The temperature was programmed to heat and kept at 1,150°C for 30 min to achieve the reduction reaction of the copper slag.

Finally, the pellets were placed in a graphite crucible in a smelting-separation furnace with a mass of 460 g. Ar was blown into the melting copper slag using a crucible tube, as shown in Figure 1. Pellets were heated to 1,350°C and held for 15 min. The sample was then cooled in a furnace. The Ar crucible tube was removed from the liquid level and allowed to flow until the temperature of the furnace reached 100°C.

To analyze the feasibility of this technology, economic estimation was calculated based on experimental data and energy consumption of each industrial scale process (Zuo et al., 2021b; Gu, 2021). The calculation results are based on the current price of raw materials and converted into Unites States dollars. For this three steps, the energy consumption is in the form of electricity. In this paper, the energy consumption is transformed into mass of standard coal. By calculation, we can see that this method has great economic benefits. On average, Unites States $ 25.056 can be obtained for each treatment of 100 kg copper slag.

Graphite crucible was cut by a chainsaw to obtain slag and ingot iron. The mass of slag, ingot iron, and carburizing powder was then tested. Then, chemical composition, mineral phase, element distribution, and crystal structure of slag, collected dust or iron ingot were analyzed by XRD, and SEM metallurgical microscope.

When cooled to room temperature, the copper slag was stratified and solidified. By cutting, the original characteristics of the slag after stratification were retained. A photograph of the copper slag after the layering process is shown in Figure 2. It was eroded to varying degrees inside and outside of the graphite crucible. On the outside of the graphite crucible, oxygen in the air contacted the graphite crucible and carried out oxidation erosion on the outside surface. As shown in Eq. 9, the oxidation erosion of graphite proceeds at the separation temperature. On the inside of the graphite crucible, Ar gas was protected from the surface of the crucible to some extent. However, erosion occurred mainly because of the reaction of metal oxides with graphite on the crucible surface, as shown in Eq. 10. “MeO” represents any metal oxide. The inside erosion phenomenon occurs only when they are in contact with one another. 2 C + O 2 = 2 CO ( g ) Δ G 1350 °C = - 121 . 5 kJ / mol (9) MeO + C = Me + CO ( g ) (10)

After reduction and solidification, there was a distinct stratification in the copper slag. The reduction slag is on the top of the solid copper slag, and ingot iron is on the bottom of the solid copper slag. The reduction slag is black brown in color, and tiny pores are present inside the sample. This means that the gas was produced by the reduction reaction, and volatiles were generated.

In addition, on the surface of the reduction slag, there appeared a slightly yellow material. This is due to the release of sulfur from the copper slag, which is removed by the gas. At the bottom, the iron ingot is shaped like a flying saucer. There is an interfacial exudate at the top of the flying saucer, which was analyzed to be residual carbon by XRD, and as shown in Figure 2. The residual carbon oozed at the interface between the iron ingot and the reducing slag. This reveals the direction of material diffusion in the smelting reduction process to some extent.

During the melting separation process, volatile substances are produced as gas dust in the molten pool. These volatiles were deposited on a corundum tube, as shown in Figure 2. Analyzed by XRD, the main components of some dusts are CuS and ZnO. Therefore, it is very important to collect soot volatilized from the copper slag in the molten state to reduce environmental pollution. The mass of the reduction slag, ingot iron, carbon, and volatiles were obtained. The mass percentages of copper slag, lignite, and CaO before reduction were 314.60, 52.61, and 93.92 g, respectively. Mass percentage of reduction slag, ingot iron, carbon, and volatiles (carried dust in gas: CuS, ZnO; gas: CO, CO₂) after reduction were 238.5, 222.6, 1.12, and 1.88 g, respectively.

Reactions in the copper slag reduction process were analyzed based on chemical composition tests. Based on the results, we determined the element type and mass content in the sample. The reactions of iron oxide containing a typical reduction reaction are shown in Eq. 10. Mass distribution of Fe, Cu, Zn, S, Pb, as typical elements, and were analyzed as shown in Figures 3, 4. In copper slag, the phases of Fe element are FeO, Fe₃O₄, and MFe. 86.71% of Fe element moved into iron ingot in the phase of MFe as shown in Figures 3, 4. Some FeO still entered into the reduction slag and ingot respectively. The results confirmed the reduction reaction sequence of iron oxides (Fe₃O₄-FeO-Fe) to a certain extent.

In copper slag, the phases of Cu element are Cu, CuO, and CuS. 10.17, 86.70, and 3.13% of Cu element moved into reduction slag, iron ingot and somke in the phase of (CuS, Cu), (Cu), and (CuS) respectively, as shown in Figures 4A,D. Although only a small part of CuS enters the flue gas, its polluting characteristics should be paied more attention on recovery study. In the reducing atmosphere, CuO phase has been transformed into Cu phase completely. This is because that, copper is preferentially reduced in the reduction process compared with FeO or Fe₃O₄.

In copper slag, the phase of Zn element is merely ZnO. As shown in Figures 4B, 5, due to the volatile characteristics of ZnO at high temperature, 94.43% of ZnO entered into the flue gas. Due to the simple element migration, the recovery purity of ZnO powder in flue gas will be relatively high.

In copper slag, the phases of S element are S and CuS. As shown in Figures 4D, 5, the distribution of S was obvious. 64.58% of S existed in reduction slag in the phase of CuS and 22.86% of S existed in smoke in the phase of CuS. To capture such volatile solid pollutants, spray soot or spray is usually used to collect such soot, and such as Venturi tube (Bal and Meikap, 2017). However, it should be noted that due to the entry of sulfur-containing substances in copper slag, the corrosion of scrubber needs attention (Sarkar et al., 2007).

In copper slag, the phases of Pb element are Pb and PbO. As shown in Figures 4C, 5, the distribution of Pb was derived by its gravity. 92.98% of Pb element went into iron ingot.

The Gibbs free energy of the reaction of C with oxide materials in copper slag varies with temperature, as shown in Figure 6. We should declare that not all elements in the slag exist in the oxidation state; we only discuss the possible reduction situation as far as possible. As for SiO₂, Al₂O₃, CaO, and MgO, reactions of C with these oxide materials cannot occur when the temperature is higher than 1,150°C. The other reaction of oxide materials with C can take place at this temperature. In other words, during the direct reduction and smelting separation process, Si, Al, Ca, and Mg elements exist in the oxide state, while the other elements cannot stably exist in that state.

Metallographic microscope photos of the copper slag, reduction slag, and ingot are shown in Figure 7. Because of the corrosive effect that occurs during the sample making process, we can find the boundary profile of the metal oxide phase and matrix in the copper slag. At different magnifications, we found that the phases in the copper slag were embedded into the matrix like jelly. After reduction, the metal oxides are stripped from the matrix, and the boundary contour becomes clearer. They are a mixture of iron oxide and slag. As shown in Table 1, in the obtained iron ingot, the black matrix material is mainly fayalite, and which is not fully reacted and is distributed in the iron metal. Some black matrix materials are dots, and some are lines.

A typical circular area element analysis of the iron ingot is shown in Figure 8A, based on SEM-EDS. In the reduction process, the matrix of fayalite shrinks continuously. As circular areas 1# and 2#, two typical circular fayalite matrices, the radius of the circle is small. The reduction degree of 2# was higher than that of 1#. Chemical analysis of the points was shown in Table 2. The Fe atoms acquired were 54.54 and 50.83% for 2# and 1#, respectively. Through the element distribution and the changes in the size of the circles in these two areas, we can determine the migration rule of elements (O, Al, Si, Fe, and S) inside the circle. Fe and O are mainly present in the matrix of the black circle in the form of FeO. The small holes in the circle matrix are caused by the release of O atoms and migrate to the produced gas. The Fe atom migrates from the inside to the outside of the circle matrix continuously. Al and Si are all present in the slag phase. The S element exists in the circle matrix at the initial stage of reduction, and the concentration of S decreases at the later reduction stage. Another typical circle matrix region, such as 3#, is shown in Figure 8B. Atom percentages of Fe is 44.22%. The results further confirmed the migration order of Fe and O and confirmed the regional variation relationship of the matrix. It should be noted that the proportions of Cu and S atoms were significantly higher than elsewhere in 1#, 2#, and 3#. This indicates that during the reduction process, Cu and S are in the CuS phase, and part of the CuS will migrate to the produced gas.

The element migration behavior during the separation process is accompanied by a partial reduction reaction. During melting separation, the liquid is divided into three layers: a reduction slag layer, a reactive boundary layer, and an iron ingot layer. The element migration process and mechanism follow the rules described below. A diagram of the element migration mechanism of smelting copper slag in a reducing atmosphere is shown in Figure 9.

Because melting separation is carried out in a graphite crucible, C powder participates in a secondary reduction reaction during the separation process. C reacts with reducible “O” atoms, which exist in 2FeO·SiO₂, Fe₃O₄, CuO forms, and CO is generated and overflows the liquid layer.

According to the analysis of the chemical composition and phase in the slag, it can be found that the Fe atom in the copper slag mainly exists in the 2FeO·SiO₂ and Fe₃O₄ phases. After reduction by C or CO, metallic iron is produced and migrates downward to the iron ingot layer because of its higher density. The Cu in the melting liquid is in the form of CuO and CuS. Like Fe, CuO is reduced by C, and metallic copper is produced and migrates downward to the iron ingot layer because of its higher density. The reduction reaction is more likely to occur than the iron oxide reduction reaction.

S exists mainly in the form of CuS. Based on the chemical composition analysis, the weight percentages of S in the copper slag, reduction slag, and iron ingot were 0.84, 1.20, and 0.051%, respectively. This indicates that the migration of S in the liquid gradually moves from diffusion to the upper reduction slag layer. Through the chemical composition analysis of the upper layer reduction slag and the phase analysis of the products in the flue gas, it was found that there was a large amount of ZnO and CuS dust in the smoke. Thus, much of the ZnO and CuS spilled from the liquid into the smoke.

As for Si, it exists mainly in the form of 2FeO·SiO₂. When CaO is in contact with 2FeO·SiO₂, desilication and slagging reactions occurred, as shown in Eq. 11. After the reaction, CaO·SiO₂ is generated and ascends to the upper reduction slag layer because of its lower density. CaO + 2 FeO·SiO 2 + 2 C = 2 Fe + 2 CO + CaO·SiO 2 (11)