2025 Volume 66 Issue 1 Pages 60-65
The distribution of Ni, Co, and Cu between the Al2O3–Li2O–CaO slag or the Al2O3–Li2O–SiO2 slag and the Ni–Co–Cu alloy at 1623 K was investigated. In the oxygen partial pressure (pO2) range of 10−14 to 10−12, the distribution ratio, LX, of component X (X: Co, Ni, or Cu) between the Al2O3–Li2O–CaO slag and the molten Ni–Co–Cu alloy was LNi < LCu < LCo. The valences of Ni and Co in the Al2O3–Li2O–CaO slag system are mainly divalent, and that of Cu is mainly monovalent, as determined from the log pO2 dependence of log LX in the pO2 range of 10−14 to 10−12. The pO2 of the Al2O3–Li2O–CaO system is three to five orders of magnitude lower than that of the Al2O3–Li2O–SiO2 system and the previously reported slag system without Li2O, even though the distribution ratios (LNi, LCo, and LCu) are similar. The activity coefficients of Ni, Co, and Cu in the Al2O3–Li2O–CaO system are all lower than those in the Al2O3–Li2O–SiO2 system and the previously reported slag system without Li2O. In particular, [γCoO] is approximately two orders of magnitude lower.
Fig. 10 Comparison of γCoO determined in this study at 1623 K with previously reported values.
In recent years, the electrification of automobiles has been expected to progress along with the rise in environmental issues worldwide, and the demand for lithium-ion batteries (LIBs), the power sources of these vehicles, is increasing. Ni, Co, and Li contained in LIBs pose risks such as resource depletion and uneven regional distribution, and resource circulation involving recycling is required.
The main constituents of LIBs are the cathode and anode materials, separators, and electrolytes. LIBs are manufactured by coating the cathode and anode active materials on a foil, inserting a separator between them, placing them into an outer can, and then filling the can with the electrolyte and sealing it [1, 2]. The positive electrode consists of an Al foil current-collector coated with a cathode material composed of Li, Ni, Co, Mn, and Al powders. The negative electrode consists of Cu foil or another current collector coated mainly with graphite powder. A polyolefin film with micropores is used as the separator, an organic electrolyte containing a lithium salt (e.g., LiPF6) is used as the electrolyte, and Al or Fe is used as the outer can.
To recover the valuable materials contained in LIBs, processes for recovering them from waste LIBs (used LIBs or defective products generated during the LIB manufacturing process) have been investigated, including physical sorting, pyrometallurgy, hydrometallurgy, and other methods [3–6]. Sumitomo Metal Mining Co., Ltd. is developing a new process that combines pyrometallurgy with hydrometallurgy to recover Cu, Ni, Co, and Li [7]. In a pilot-scale pyrometallurgical process, LIBs are melted to remove most of the impurities as slag, and the resulting Ni–Co–Cu alloy with only a few impurities is purified using a hydrometallurgical process to recover a battery-grade nickel–cobalt mixture. In the pyrometallurgical process, when the elements contained in LIBs are melted in the presence of carbon, elements such as Al and Li, which have strong affinities for oxygen and are thus easily oxidized, are distributed in the slag phase as oxides, whereas elements such as Ni, Co, and Cu, which are easily reduced by carbon, are distributed in the metallic phase [8]. It is important to understand the distribution behavior of these components in the slag and metallic phases when considering the operating conditions.
Many studies have examined the distribution of Ni, Co, and Cu between the slag and metallic phases, focusing on conventional slag systems, whose main components are FeOx, SiO2, Al2O3, CaO, and MgO. For example, studies have been conducted on the distribution between the slag phase and the metallic phase in the FeOx–SiO2 [9, 10], FeOx–SiO2–Al2O3 [11], FeOx–MgO–SiO2 [12, 13], FeOx–MgO–SiO2–Al2O3 [13], FeOx–MgO–SiO2–CaO [14], FeOx–CaO [15], FeOx–CaO–SiO2 [16], and the CaO–SiO2–Al2O3 [17] systems. Ren et al. studied the MnO–SiO2-based [18] and the MnO–SiO2–Al2O3-based [18, 19] slags for the treatment of waste LIBs. However, none of the above studies discussed slag systems containing Li2O.
When waste LIBs are subjected to reduction melting treatment, Li is distributed in the slag, producing slag that contains a lot of Li2O. However, as mentioned above, there are no reports on the distribution of Ni, Co, and Cu in slag systems containing Li2O. In this study, we investigated the distribution of Ni, Co, and Cu between the Al2O3–Li2O–CaO slag, which is the basic form for the waste LIB melting process currently under development, and the molten Ni–Co–Cu alloy at 1623 K. For comparison, we also determined the distribution of Ni, Co, and Cu between the Al2O3–Li2O–SiO2 slag and the Ni–Co–Cu alloy.
The reagents, Al2O3 (special grade) and Li2CO3 (99.0 mass%), were weighed, mixed, and placed in an Al2O3 crucible. The mixture was melted in an electric furnace at 1623 K under N2 gas for 2 h and then cooled to produce a synthetic slag sample of 57%Al2O3–43%Li2O. Approximately 30 g of a metallic powder sample was prepared by weighing and mixing reagent-grade Cu powder (99.5 mass%), Ni powder (99.7 mass%), and Co powder (99.5 mass%) to obtain a specified composition, and then placing the mixture in an MgO crucible (φ38 × φ33 × 45 mm). Approximately 30 g of a slag sample was placed on top of the metallic powder sample, which was prepared by adding CaO (≥98.0 mass%) or SiO2 (special grade) and Cu2O (99.5 mass%) to the synthetic slag described above to obtain the specified composition. The MgO crucible containing the sample was placed in an Al2O3 crucible (Fig. 1), which was covered and placed in an electric furnace. The crucible was maintained at 1623 K for 2 h while supplying Ar gas flowing at a rate of 3 L/min to melt the sample. After the holding period, the following oxygen concentration cell containing a MgO-stabilized ZrO2 solid electrolyte was immersed in the molten sample to measure the electromotive force, and the oxygen partial pressure was calculated using eq. (1).
Schematic of experimental equipment.
Pt/Re/Cr-Cr2O3/ZrO2-MgO/sample/Re/Pt
| \begin{equation} E = -\frac{\text{R}T}{n\text{F}}\ln\frac{p_{\text{O2}}^{\text{R}}}{p_{\text{O2}}^{\text{S}}} \end{equation} | (1) |
Here, the electromotive force is E (V), the gas constant is R = 8.314 (J mol−1K−1), the temperature is T (K), the amount of electrons is n (mol), the Faraday constant is F = 96 500 (J mol−1V−1), the oxygen partial pressure in the reference electrode is pO2R, the oxygen partial pressure in the sample electrode is pO2S, and the oxygen partial pressure pO2 is a dimensionless number obtained by dividing the oxygen partial pressure by the standard pressure of 1.01325 × 105 Pa. An oxygen concentration cell was constructed as shown in Fig. 2. The lead wire consisted of a Pt wire with a Re wire welded to the end, and the reference electrode was a MgO-stabilized ZrO2 tube containing Cr powder (99.9 mass%) and Cr2O3 powder (>98.5 mass%) mixed at a mass ratio of Cr:Cr2O3 = 3:1. The electromotive force was measured by connecting the sample electrode and reference electrode to the positive and negative sides of the data logger, respectively. The oxygen partial pressure pO2R in the reference electrode was calculated from the standard free energy of the reaction using the following reaction formula:
| \begin{equation} \text{4/3Cr(s)} + \text{O$_{2}$(g)} = \text{2/3Cr$_{2}$O$_{3}$(s)} \end{equation} | (2) |
| \begin{equation} \varDelta G^{\circ}/\text{J} = -754\,900 + 170\ T\ [20] \end{equation} | (3) |
After measuring the oxygen partial pressure, the sample was removed and cooled in water together with the crucible, and the elemental distribution was determined by chemical analysis of the obtained alloy and slag. Tests were conducted by changing the oxygen partial pressure in the melt by varying the amount of Cu2O added from 0 g to 3.6 g, and the relationship between the measured oxygen partial pressure and the elemental distribution was obtained.
Schematic of oxygen concentration cell.
Table 1 lists the results of the oxygen partial pressure measurements and the component concentrations of each phase, as obtained by chemical analysis of the slag and alloy phases. In this experiment, MgO was dissolved in the slag because a MgO crucible was used. Figure 3 shows the MgO concentration vs. Li2O and CaO concentrations in the slag. As shown in the figure, the solubility of MgO in the Al2O3–Li2O–CaO slag is lower than that in the Al2O3–Li2O–SiO2 slag. In addition, the MgO concentration in the slag tends to decrease with increasing concentrations of Li2O and CaO in the slag. Because MgO is a basic oxide, it is highly soluble in the Al2O3–Li2O–SiO2 system, which contains the acidic oxide SiO2. Conversely, MgO is poorly soluble in the more basic Al2O3–Li2O–CaO slag, which contains high concentrations of Li2O and CaO.
;%0A%09%09%09newWindow.document.open();%0A%09%09%09newWindow.document.write('<img src=%22./Graphics/MT-M2024120tab01.jpg%22>');%0A%09%09%09newWindow.document.close();%0A%09%09)
MgO solubility in Al2O3–Li2O–CaO and Al2O3–Li2O–SiO2 slags at 1623 K.
Figures 4 and 5 show the relationship between pO2 and the distribution ratio, LX, of component X (Ni, Co, or Cu). Here, LX is defined as LX = (mass%X in slag)/[mass%X in metal]. As shown in Fig. 4, for the Al2O3–Li2O–CaO slags, the log LX dependence on log pO2 yielded slopes of 0.58 and 0.56 for Co and Ni, respectively, which is close to 1/2, and a slope of 0.28 for Cu, which is close to 1/4. In addition, within the range of oxygen partial pressures used in this study, the distribution ratio was Ni < Cu < Co, which reflects the increasing ease of distribution in the slag. If the equilibrium reaction of component X between the slag and metallic phases is expressed by eq. (4), and the equilibrium constant KX is expressed by eq. (5), the LX can be derived using eq. (6).
| \begin{equation} \text{X(l)} + \text{$\nu$/4 O$_{2}$(g)} = \text{XO$_{\nu/2}$(l)} \end{equation} | (4) |
| \begin{equation} K_{\text{X}} = \frac{a_{\text{XO}_{\nu/2}}}{a_{\text{X}} \cdot p_{\text{O${_{2}}$}}^{\nu/4}} \end{equation} | (5) |
| \begin{equation} \log L_{\text{X}} = \frac{\nu}{4}\log p_{\text{O${_{2}}$}} + \log K_{\text{X}} + \log\frac{[\gamma_{\text{X}}]}{(\gamma_{\text{XO${_{\nu/2}}$}})} + \log\frac{(n_{\text{T}})}{[n_{\text{T}}]} \end{equation} | (6) |
where [γX] and (γXOν/2) are the activity coefficients of X in the alloy and XOν/2 in the slag, respectively, and (nT) and [nT] are the total moles of substances contained in 100 g of slag and alloy, respectively. From eq. (6), the oxide form of X in the slag, XOν/2, can be calculated from the slope ν/4 of the relationship between log pO2 and log LX. In Figs. 4 and 5, the oxidized forms of Ni and Co are considered to be NiO, CoO, and CuO0.5, respectively, based on the log pO2 dependence of log LX. The experimental results of this study are compared with the distribution ratios of Ni and Co between the MnO–SiO2–Al2O3 slag and the molten Cu–Ni or Cu–Co systems reported by Ren et al.; the distribution ratio of Ni between the FeOx–SiO2 and the molten Au–Ni–Fe systems reported by Grimsey; the distribution ratio of Co between the FeOx–MgO–SiO2–CaO and the molten Cu–Ni–Fe systems reported by Kitamura et al.; the distribution ratio of Cu between FeOx–CaO and molten Cu reported by Takeda et al.; and the distribution ratio of Cu between the CaO–SiO2–Al2O3 and the molten Cu–Pb reported by Ishikawa et al. The comparison is shown in Figs. 6, 7, and 8. As shown, the pO2 of the Al2O3–Li2O–CaO system is three to five orders of magnitude lower than that of the Al2O3–Li2O–SiO2 system and the previously reported slag system without Li2O, even though the distribution ratios (LNi, LCo, and LCu) are similar. The reason for this difference is unclear, but it is thought to arise from differences in the slag and alloy systems.
Distribution ratios of Ni, Co, and Cu between Al2O3–Li2O–CaO slag and Ni–Co–Cu alloy at 1623 K and varying oxygen partial pressures.
Distribution ratios of Ni, Co, and Cu between Al2O3–Li2O–SiO2 slag and Ni–Co–Cu alloy at 1623 K and varying oxygen partial pressures.
Comparison of log LNi determined at 1623 K in this study with previously reported values.
Comparison of log LCo determined in this study at 1623 K with previously reported values.
Comparison of log LCu determined in this study at 1623 K with previously reported values.
In our study, the activity coefficients of NiO, CoO, and CuO0.5 were calculated using eq. (6) based on our experimentally determined values of LNi, LCo, and LCu, and compared our results with previously published data. The equilibrium constant KX in eq. (6) was calculated from the following reaction and its standard free energy change, ΔG°.
| \begin{equation} \text{Ni(l)} + \text{1/2O$_{2}$(g)} = \text{NiO(s)} \end{equation} | (7) |
| \begin{equation} \varDelta G^{\circ}/\text{J} = -250\,500 + 93.97\ T\ [21] \end{equation} | (8) |
| \begin{equation} \text{Co(l)} + \text{1/2O$_{2}$(g)} = \text{CoO(s)} \end{equation} | (9) |
| \begin{equation} \varDelta G^{\circ}/\text{J} = -249\,900 + 79.62\ T\ [21] \end{equation} | (10) |
| \begin{equation} \text{Cu(l)} + \text{1/4O$_{2}$(g)} = \text{CuO$_{0.5}$(l)} \end{equation} | (11) |
| \begin{equation} \varDelta G^{\circ}/\text{J} = -63\,620 + 23.43\ T\ [21] \end{equation} | (12) |
In addition, because no reports of [γCo], [γNi], and [γCu] in the molten Ni–Co–Cu alloys were found in the literature, the thermodynamic calculation software FactSage [22] was used to calculate the liquid-based activities of Ni, Co, and Cu for the composition and temperature used in this experiment, and the obtained activity values were used to calculate [γCo], [γNi], and [γCu]. The database FSstel was used for the FactSage calculations. Table 2 lists the activity coefficients, where (γNiO) and (γCoO) are based on solid standards, and (γCuO0.5) is based on liquid standards. In the Al2O3–Li2O–CaO system, (γNiO), (γCoO), and (γCuO0.5) were approximately 0.05–0.16, 0.02–0.06, and 0.4–0.9, respectively, while in the Al2O3–Li2O–SiO2 system, (γNiO), (γCoO), and (γCuO0.5) were approximately 4–18, 2–8, and 7–16, respectively. Figures 9, 10, and 11 compare the (γNiO), (γCoO), and (γCuO0.5) values obtained in this study, respectively, with those reported in the literature. The (γNiO), (γCoO), and (γCuO0.5) values for the Al2O3–Li2O–CaO system were all lower than those reported previously, where (γCoO) in particular was approximately two orders of magnitude lower than the literature values. By contrast, the (γNiO), (γCoO), and (γCuO0.5) values for the Al2O3–Li2O–SiO2 system were almost the same as those reported previously. Although NiO and CoO are generally considered to be basic oxides, the extremely high basicity of the Al2O3–Li2O–CaO slag may cause NiO and CoO to partially combine with O2−, resulting in acidic oxide-like behavior, which may explain the small activity coefficient values. On the other hand, the addition of SiO2 lowers the basicity, so NiO and CoO behave as basic oxides as usual, and the activity coefficients are thought to be similar to those in the previous study.
;%0A%09%09%09newWindow.document.open();%0A%09%09%09newWindow.document.write('<img src=%22./Graphics/MT-M2024120tab02.jpg%22>');%0A%09%09%09newWindow.document.close();%0A%09%09)
Comparison of γNiO determined in this study at 1623 K with previously reported values.
Comparison of γCoO determined in this study at 1623 K with previously reported values.
Comparison of γCuO0.5 determined in this study at 1623 K with previously reported values.
The following findings regarding the distribution of Ni, Co, and Cu between the Ni–Co–Cu alloy and the Al2O3–Li2O–CaO slag or the Al2O3–Li2O–SiO2 slag were obtained:
