Initial Purification of the Leaching Solution
The leaching solution (PLS1) was rich in battery metals (Ni, Co, Mn and Li) and impurities like Fe, Al and Cu (shown in Table I). Based upon previous research,19 Cu was first recovered as high-purity copper powder by cementation, i.e., utilization of redox replacement phenomenon. In this study, copper powders were recovered using 1.3 times the stoichiometric amount of iron powder at pH 2, which gave a recovery of > 99% (supplementary Fig.S-1, in online supplementary materials) and impurities within the copper product of ca. 0.2% (supplementary Fig.S-2). The resultant copper-free solution (PLS2), that contained less than 20 ppm Cu, was then subjected to oxidization (H2O2: 2% v/v; pH ~ 2) and neutralization at pH 3.5 (supplementary Fig.S-3), which resulted in the majority of Fe (ca. 90%), some Al (ca. 20%) and a trace amount of Co (0.5%) being removed. Through these purification steps, the concentrations of the impurity elements in the solution were reduced to 0.02 g/L Cu, 0.48 g/L Fe and 1.65 g/L Al (PLS3). Further details about the initial purification process can be found in the supporting information.
Mn Separation from Other Elements by Solvent Extraction
The commercial solvent, D2EHPA, has been widely used for the purification of Ni and Co sulfate-containing solutions. For example, Cook and Szmokaluk have reported that the Mn concentration in Co-rich raffinate (ca. 10 g/L Co) could be reduced to a level of a few ppm by multistage solvent extraction process with D2EHPA.20 According to the literature, extraction of metals in D2EHPA follows the order: Fe3+ > Zn2+ ≈ Ca2+ > Fe2+ > Cu2+ ≈ Mn2+ > Co2+ ≈ Ni2+ > Mg2+ > Li+ based on their pH50 values.18,21 From the data, it appears that D2EHPA should provide a good separation of Mn from elements like Co, Ni and Li. Nevertheless, as the pH50 values for metals are usually obtained from ideal single element solutions, use of multi-metal mixtures typical for hydrometallurgical solutions may result in non-ideal behavior.
Batch Extraction Experiments with D2EHPA
Figure 1a, b, and c shows the extraction of Co, Ni, Mn, Al, Fe and Li from PLS3 as a function of equilibrium pH, A/O ratios and D2EHPA concentration. Furthermore, extraction isotherms of Mn, Fe and Al are displayed in Fig. 1d at pH 3.2 (25°C). From the results illustrated in Fig. 1a, b, and c the extraction of metals in D2EHPA is seen to follow the order: Fe > Al > Mn > Co > Ni ≈ Li, with the extraction of Co, Ni and Li observed to be much lower than that of Fe, Al, and Mn. This suggests that the selective extraction of Fe, Al, and Mn from Co, Ni and Li can be achieved under optimized conditions. For example, extraction of Fe, Al and Mn were in excess of 80%, whereas Co, Ni, and Li extractions remained below 1% at pH 3.2 with A/O = 1:2 and 0.4 M D2EHPA. Based upon the equilibrium results in Fig. 1c, the extraction isotherms for Fe, Al and Mn were plotted in Fig. 1d. A sharp and linear extraction isotherm for Fe and Al can be observed, indicating that D2EHPA has a very strong extraction ability towards Fe and Al under the investigated conditions. Interestingly, Mn extraction shows an upward trend when the A/O ratio is below 1:1, and then it decreases substantially as the A/O ratio is increased to 2:1. This finding probably results from a lack of reactive sites for complexation and the increase in competitive extraction of Al and Fe as the A/O proportions are increased to 2:1. In the case of the other elements (Co, Ni and Li), their extraction isotherms are not presented as their extractions were of an insufficient level to form an isotherm.
The corresponding separation factors of Mn in relation to the other five elements are listed in supplementary Table S-I. Separation factors of Mn over Co, Ni and Li were determined to be in the hundreds, which indicates a good separation towards Mn. In comparison, the low separation factors of Mn/Fe (< 0.1) highlights that there is a good separation of Fe over Mn. Conversely, the separation factors of Mn over Al are close to 1.0, which suggests a difficult separation between these elements by D2EHPA.
Multiple Stage Extraction
Based upon the above results of batch experiments, a three-stage continuous counter-current extraction process was simulated under the condition of A/O = 1:2, D2EHPA = 0.4 M, pH 3.2, t = 15 min. In order to simulate the change of metal concentration in organic phase during the three-stage solvent extraction process, Series 1 was conducted with the same organic sample consecutively contacted by three new PLS3 solutions. The results in Table II show that Fe, Al and Mn were clearly extracted to the organic phase and that their concentration in the organic phase increases with the increased number of extraction stages. Also, Co extraction was evident in the first SX stage (loading up to 281 ppm); however, with further contact with fresh PLS3, Co was crowded back to the PLS. The same phenomenon was also observed with Ni and Li. After three contacts with the PLS3, the organic phases were loaded with 0.57 g/L of Fe, 1.20 g/L of Al, 1.59 g/L Mn and less than 0.02 g/L of Co, Li and Ni. These findings indicate that the extracted or entrained Co, Ni and Li in the organic phases can be effectively crowded out by Fe, Al and Mn.
Table II Metal concentration in organic (Series 1) and aqueous solutions (Series 2) during multiple stage extraction at pH 3.2 (A/O = 1:2, t = 15 min) In contrast, Series 2 simulated the change of aqueous solution with the same PLS3 sample contacted by new organic samples three times. As observed, the concentration of Fe, Al and Mn decreased to a few ppm after three extraction stages with fresh organic solutions. The Co, Ni and Li concentration remained almost the same after the first two stages of extraction. Nevertheless, at the 3rd stage of contact with fresh organic solvent, approximately 38% of Co and 5% of Ni and Li were co-extracted. The high extraction of Co, Ni, and Li during the 3rd stage extraction can be attributed to the availability of too many complexation sites within the organic solvent and reduced competition from Fe, Al and Mn.
Overall, it can be concluded from the results of Series 1 and 2 that more than 99% of Mn, Al and Fe could be extracted from the aqueous solution by the multistage extraction experiments and that most of the co-extracted Co, Ni, and Li could be crowded out from the loaded organic phases by Mn, Al and Fe.
Stripping Process of Loaded D2EHPA
Scrubbing Process
The aim of scrubbing the organic phase is to replace the co-extracted or mechanically entrained Co, Ni or Li together with Mn. The loaded organic from Series 1 was contacted twice by pure MnSO4 solution (Mn: 4 g/L) under the conditions of A/O = 1:10, t = 5 min. The initial pH of the MnSO4 scrubbing solution was 4 and no further pH adjustment was performed during the scrubbing process. Table III shows the chemical composition of the scrubbing solutions (S1, and S2) and the resultant organic solution (O1). As observed from Table III, after two scrubbing stages, the co-extracted Co, Ni, and Li in the organic solution were almost completely replaced by Mn, resulting in an organic solvent (O1) with almost no Co, Ni, and Li (< 0.003 g/L) and scrubbing solutions (S1 and S2) that contain Mn, Co, Ni and Li. Moreover, the Al concentrations in S1 and S2 were found to be ca. 0.2 g/L and 0.1 g/L, respectively, which suggests that part of the Al (ca. 2.3%) was also substituted by Mn in the organic phase indicating that a scrubbing solution with lower Mn concentration (< 4 g/L) should be used. The scrubbing solution produced could be circulated back into the solvent extraction process, whereas the resultant organic solution can be subjected to a selective stripping process.
Table III Concentration of elements in the two-stage scrubbing solutions (S1 and S2) and the resultant organic solution (O1) Selective Stripping of Mn
Stripping behaviors of Fe, Al, and Mn from the resultant organic phase after scrubbing (O1; Table III) were investigated in terms of H2SO4 concentration (0.01–1.0 mol/L), A/O ratio and stripping time. It can be clearly seen from Fig. 2a, that Mn is stripped in preference to Al and Fe over the range of H2SO4 concentration investigated and that the following sequence of Mn > Al < Fe (A/O = 1:1, t = 30 min) is established. Complete Mn stripping could be achieved with 0.5 mol/L H2SO4, which also results in Al stripping of ca. 40% and Fe stripping of only 0.2%. Nevertheless, with further increases of H2SO4 concentration from 0.5 mol/L to 2.0 mol/L the level of Al stripping remains at around 40%, which implies that the other ca. 60% of the extracted Al ions have a higher affinity towards D2EHPA, and that stronger stripping agents like HCl need to be adopted in order to achieve complete Al stripping. In contrast, in the case of Fe, the stripping yields in the acid concentrations investigated remained relatively low at < 3%.
Figure 2b and c illustrates the influence of A/O ratios and stripping time with 0.5 mol/L H2SO4. As observed in Fig. 2b, as the stripping time is extended, Al and Mn show similar behavior and both reach equilibrium after 10 min. In Fig. 2c, an increase of the A/O ratios has a limited influence on Mn stripping, which maintains a high level of ca. 98%, whereas Al stripping decreases by nearly 20% from ca. 45% (A/O = 1:1) to 25% (A/O = 1:8). The variation of separation factors with H2SO4 concentration is presented in Fig. 2d. The results show that the separation factor of Mn towards Al increases with increasing H2SO4 concentration, whilst that of Al towards Fe shows an initial rise in the range of 0.01–0.5 mol/L, before a dramatic decrease is observed at higher acid concentrations. In order to produce a stripping solution that allows a high Mn extraction and low associated impurities, the optimal H2SO4 concentration is suggested to be in the region of 0.2–0.5 mol/L. Overall, selective stripping with H2SO4 = 0.5 mol/L, A/O = 1:8, t = 10 min, allows ~ 99% Mn and 25% Al to be stripped, to give a high-Mn-low-impurity solution composed of approximately 12 g/L Mn and 2 g/L Al and almost no other impurity metals (Fe, Ni, Co: < 0.025 g/L, Li: < 0.01 g/L).
Mn Recovery by Oxidative Precipitation
The resultant Mn-rich solution was subjected to oxidative MnO2 precipitation with KMnO4, as outlined in reaction (4) (T = 298 K).
$$ {\text{Mn}}^{{ 2 + }}+ {\text{2MnO}}_{ 4}^{ - }+ {\text{2H}}_{2} {\text{O}} = {\text{5MnO}}_{ 2} ( {\text{s) + 4H}}^{ + } \quad\Delta G = - 143.6\,{\text{kJ/mol}}. $$
(4)
It can be seen from Fig. 3a that Mn recovery increased with the higher amounts of KMnO4 and that up to 99% Mn could be recovered with an equal stoichiometric amount of KMnO4, as would be expected from the Mn oxidization reaction (4). The resultant filtrate contains < 18 ppm Mn and the purity of the MnO2 product is up to 99.5%; impurity elements like Co, Fe, and Al are all less than 0.02% (Fig. 3b). In contrast, Fig. 3c shows the direct precipitation of Mn from non-purified solution (PLS3) in the presence of other elements, which has been previously reported as a method to recover Mn from LIB leaching solution.22 It can be seen that Mn precipitation recovery from PLS3 is obviously lower compared to the purified stripping solution (Fig. 3a) with the same stoichiometric consumption of KMnO4. From PLS3, the maximum Mn precipitation achieved is only ca. 80% when 1.5 times the stoichiometric amount of KMnO4 was consumed. Moreover, cobalt precipitation is also observed to increase with higher levels of KMnO4 additions, and with cobalt precipitation levels of > 8% with 1.5 times the stoichiometry of KMnO4 this results in impure MnO2 products with ~ 9% Co and 0.7% Fe (Fig. 3d). Furthermore, a more detailed examination of the SEM images (Fig. 3b and d) shows that there are distinct morphological differences between the MnO2 products formed depending on the solution. For example, the MnO2 produced from the purified stripping solution has a more uniform particle size (< 2 µm) than that produced from PLS3, which has a much more inhomogeneous appearance as a result of the relatively high level of impurities like Co.
Figure 4 displays the Eh-pH diagram of the Mn-Co-H2O system for 0.05 M Mn and 0.35 M Co at 25°C (HSC 9.2.6; Outotec Technologies). The upper left green region is the overlapping domain of soluble Co (II) and Mn precipitate below the O2 formation line (upper dashed line), in which the thermodynamics predict selective precipitation of Mn without the co-precipitation of Co. Such a narrow operating range of Eh and pH values suggests the difficulty in obtaining high-purity MnO2 in the presence of Co ions, which are prone to be oxidized as a Co3O4 precipitate via reaction (5) (298 K):
$$ 9 {\text{Co}}^{{ 2 { + }}}+ {\text{ 2MnO}}_{ 4}^{ - }+ {\text{ 8H}}_{2} {\text{O}} = {\text{2MnO}}_{ 2} ( {\text{s) + 3Co}}_{ 3} {\text{O}}_{ 4} {\text{ + 16H}}^{ + } \quad\Delta G = - 29.3\,{\text{kJ/mol}} $$
(5)
In principle, a high-purity MnO2 product can be produced from the acid leaching solution of LIB waste by a combination of solvent extraction with D2EHPA, selective stripping and oxidative precipitation, as presented in Fig. 5. Note that this study does not include details of the recovery of Co, Ni, and Li, which can be achieved by state-of-art SX-EW (solvent extraction and electrowinning) processes, followed by Li recovery from the raffinate as Li3PO4 or Li2CO3.4,23 One advantage of the current approach is that, after partial neutralization of the Fe and Al, the rest of these impurities can be co-removed from the PLS along with Mn removal, already at a low pH (pH ~ 3.2), and thus the co-precipitation of Co, Ni and Li into the Fe and Al precipitate can be avoided.24 The Mn recovery methodology proposed has excellent potential, as the technology required for the different unit processes is readily available and in use on an industrial scale for other applications.