*3.3. InduMelt Experimients*

The last experimental series was conducted in the presented InduMelt reactor (Figure 2) to investigate the achievable transfer coefficients for Li, Ni, Co and Mn under the particular conditions of the reactor. The trend of the measured temperatures in- and outside of the reactor during one of the experiments is presented in Figure 8a. As explained in the materials and methods section, the slope of the outer s-type couples is used to control the temperature inside of the reactor after the operating temperature of the k-type couples is exceeded.

**Figure 8.** (**a**) Trend of reactor temperatures during IM\_NMC\_C. (**b**) Picture showing the crucible and the packed bed of graphite cubes with metal depositions after IM\_NMC\_C.

In Table 2 the compositions of the input mixtures for the InduMelt experiments are shown. For NMC\_C and NCA\_C the composition matches the stochiometric proportion of the used cathode materials (NCA, LiNi0.8Co0.15Al0.05O2; NMC, LiNi0.33Mn0.33Co0.33O2) with carbon addition of 10 w/%. If the whole added carbon is used and all oxides are removed the mass loss should accumulate to 40–46% of the input mass depending on

the amount of Li that can be removed. For AM, which is a mixture of different cathode materials from LIBs and considering its composition most likely also other battery types, the volatile components also accumulate to around 42 w/%.

**Table 2.** Chemical composition of the input mixtures for the InduMelt experiments. (w/%).


<sup>1</sup> Cu and other impurities are not specified here since they are not in focus of the experimental series. <sup>2</sup> Calculated on basis of the stochiometric Li-O2 ratio.

Because the aim of the experimental series is to investigate possible recovery and removal rates for certain metals contained in the cathode materials, Cu and other impurities of the sample AM are not further analyzed.

For the first InduMelt experiments with LIB cathode materials and black matter a maximum temperature of approx. 1550 ◦C was chosen. At this temperature, no further changes of the CSA or mass during the STA and heating microscope were observed and the expected metal alloy's melting point is also some ten degrees lower. This temperature was then held for approx. one hour before the heat input was stopped.

In Figure 8b the reactor after the experiment is shown. All graphite cubes were removed and cleaned from metal and slag deposits which were subsequently weighed. The individual mass of input material and product phases for each experiment can be seen from Table 3.

**Table 3.** Masses of the input sample and the obtained products in InduMelt experiments. (g).


<sup>1</sup> Neither metal accumulations nor slag depositions could be found.

The obtained product phases are subdivided into metal phase, slag phase and powder. On the first look at Table 3, one can see that the product distribution differs greatly between the experiments IM\_NMC\_C, IM\_NCA\_C and the experiment IM\_AM. Therefore, the results are presented and discussed separately.

For IM\_NCA\_C and IM\_NMC\_C the metal and slag phase accumulates at the bottom of the reactor or can be found as depositions on the graphite cubes and the crucible. To achieve the best mass balance possible, the depositions have been rubbed of the graphite cubes and the metal particles were magnetically separated. By this, 244.2 g respectively 267.3 g of a metal product, which—if we assume that the metal phase only consists of Ni, Co and Mn—accounts for 81% and, respectively, 91% of the said metals in the input material of IM\_NMC\_C and IM\_NCA\_C. According to the oxygen potentials of the metals, the slag phase should mainly consist of Li2O and Al2O3. With 37.7 g and 21.6 g of obtained slag for IM\_NMC\_C and IM\_NCA\_C compared to an input of approximately 36 g of pure Li alone one can say that this result looks promising, since the amount of oxygen—and of course Al—must also be taken into account. Furthermore, the refractory mortar and the crucible material also consist of Al2O3 and can take part in the reactions causing slag formation. Because this discussion is more complex than for the metal phase it will be continued later together with the chemical analyses of the phases. The powder phase of IM\_NMC\_C and IM\_NCA\_C is caused by abrasion during the removal of the small metal particles from the graphite cubes and therefore mainly consists of carbon. Summarized, the overall weight loss of IM\_NMC\_C and IM\_NCA\_C is 46.8% and, respectively, 41.4% of the input mass. If we assume that Li, O and C are the only volatile components in the

input material a maximum weight loss of 47.2% for IM\_NMC\_C and 50.2% for IM\_NCA\_C is achievable. For IM\_NMC\_C, the obtained slag phase is shown in Figure 9a, the metal accumulation in Figure 9b.

**Figure 9.** Obtained slag (**a**) and metal sample (**b**) from the experiment IM\_NMC\_C.

As can be seen, the separation of the metal and slag phase in IM\_NMC\_C for further chemical analysis was relatively easy since large specimens without fusions could be found. In contrast, the obtained products from IM\_NCA\_C were harder to separate as Figure 10a–d shows. Therefore, the ICP-MS analysis was performed for both, samples with and without inclusions, and the results weighted during data evaluation.

**Figure 10.** Obtained metal and slag samples from the IM\_NCA\_C experiment. (**a**) Metal sample 1 which is strongly fused with the produced slag. (**b**) Metal sample 2 with very little slag inclusions. (**c**) Slag sample 1 with metal depositions. (**d**) Slag sample 2 without inclusions or depositions.

To intensify this discussion, we need to look at the results of the chemical analysis, which were achieved by ICP-MS analysis. The discussion starts with the obtained metal phase from the experiments IM\_NMC\_C and IM\_NCA\_C for which the results are contained in Table 4.


**Table 4.** Mass fractions of certain metals in the obtained metal phases. (w/%).

<sup>1</sup> Species was not analyzed in this experiment. <sup>2</sup> Small inclusions of slag in the metal matrix need to be considered. <sup>3</sup> Slightly over-determined due to weighted consideration of residuals from the aqua regia digestion.

For IM\_NMC\_C the metal composition mostly matches the expected result. There is almost no Li and Al present in the metal alloy but Ni, Co and Mn. What is noticeable, however, is the significantly lower Mn content compared to Ni and Co. With an equal stoichiometric proportion and similar molecular weight—Mn is a little lighter—the difference should not be that high, which indicates that Mn also accumulates somewhere else than in the metal alloy.

As already explained, the sampling of NCA\_C was not trivial due to small slag inclusions within the metal particles. In order to increase the informative value, metal samples with (IM\_NCA\_C\_1) and without (IM\_NCA\_C\_2) small slag particles were analyzed. By this it can be stated that also for IM\_NCA\_C there was hardly an accumulation of Li and Al in the metal alloy that mainly consists of Ni and Co.

A complete mass balance is hardly feasible due to the difficult collection of the small metal particles. In future experiments and respective analyses, ICP-OES as well as XRD analysis methods will be used to balance all the elements included in greater detail. Nevertheless, compared to the initial amount in the input material it was possible to find around 90% of Ni and Co and 76% of Mn in the metal phase of IM\_NMC\_C as well as more than 90% of Ni and Co in the metal phase of IM\_NCA\_C.

In order to investigate the whereabouts of Mn, to clarify whether Ni and Co can also be found in the slag and to finally check the question of whether Li removal from the reactor could be achieved or not we now look at the slag analysis shown in Table 5.


**Table 5.** Mass fractions of certain metals in the obtained slag phase. (w/%).

Species was not analyzed in this experiment.

Beginning with IM\_NMC\_C it can be said that Ni does hardly accumulate in the slag while a significantly higher but still low amount of Co could be found. For Mn, from which only 76% of its initial input were found in the metal phase, can also not be found in the slag phase. Since Mn is very reactive and has several oxidation states it is likely that parts of it were removed from the reactor via the gas phase. For IM\_NMC\_C, analogous to the metal phase results, there are again two samples, IM\_NCA\_1 with metal particles and IM\_NCA\_2 without metal particles. The data shows that only a small amount of Ni and Co is found in the slag while Li and Al accumulate to higher extents.

If we now compare the amount of Li that was initially inserted in the experiments, which was approx. 36 g for IM\_NCA\_C and IM\_NMC\_C with the amount of Li that was found in the metal and slag phase, a lithium removal of 96.72 w/% for IM\_NCA\_C and 90.76 w/% for IM\_NMC\_C was achieved.

Before these results are finally summarized, we have to take a look at IM\_AM, which, as mentioned at the beginning, behaved differently than IM\_NCA\_C and IM\_NMC\_C. As can be seen in Table 3, neither a metal nor a slag accumulation was found but only a fine powder that was optically identical to the input material. The weight loss of 29,5% matches the initial carbon content exactly, which at first sight suggests that only the included carbon was burned in the reactor. However, analysis of the carbon content of the resulting powder

revealed a mass content of still 22.6%, which indicates that also in IM\_AM reduction reactions occurred. In the thermogravimetric analyses only a decrease in mass of 10% was achieved. This could be an indication that certain reactions proceed more slowly in AM and that longer holding times in the preliminary experiments would have provided better results, which is going to be investigated in the further course of the project. Furthermore, an increase of the average particle size was found that indicates at least an agglomeration of particles even if there was no molten phase. Because there was no slag or metal phase in IM\_AM, the results are discussed by a comparison of the chemical composition before and after the InduMelt experiment, which is shown in Table 6.

**Table 6.** Chemical composition of AM before and after the InduMelt experiment. (w/%).


<sup>1</sup> Total mass of input material: 561.9 g. <sup>2</sup> Total mass of product: 396.1 g.

The mass content of Ni, Co and Mn has risen by about 65% each which can only be caused by the mass loss of the sample. A statement about a possible discharge of Mn via the gas phase, as it was observed in IM\_NMC\_C, should not be made due to the already low concentration in IM\_AM. Lithium had an input mass of 13.59 g and was reduced to 3.04 g in the product powder, which corresponds to a decrease of 77.6 w/%. This value is significantly lower than with pure cathode materials but in the light of the different behavior of AM compared to NCA\_C and NMC\_C in all experimental series still a promising result.

To finally summarize the InduMelt experiments, one must notice that the difficulties to achieve a complete mass balance and the absence of an off-gas analysis lead to the fact that the absolute numbers should only be considered to a limited extent. However, it is not the claim of this work to precisely define transfer coefficients for all species in cathode materials respectively black matter, but to evaluate the magnitude of possible recovery rates for the valuable metals Ni, Co, Mn and Li by using the InduRed reactor technology. In view of this, these tendencies are summarized in Figure 11.

**Figure 11.** Qualitative consideration of the accumulation of Ni, Co, Mn and Li in the product phases obtained from the InduMelt experiments.

### **4. Conclusions**

The literature research clearly shows that the possibility of simultaneous lithium recovery with a pyrometallurgical process would close a large gap in the recycling chain. To evaluate if the presented InduRed technology can potentially provide a solution to this problem, a series of experiments have been conducted. By heating microscope experiments and simultaneous thermal analysis, the behavior of NCA and NMC cathode materials as well as black matter (AM) at high temperatures and under reducing conditions was investigated. The results showed that the significant reduction reaction between the lithium metal oxides and carbon take place between 800 ◦C and 1000 ◦C and that the produced metal alloy melts at approximately 1500 ◦C, which are technically feasible temperatures for the desired process.

Experiments, conducted in the InduMelt plant, a lab scale reactor modeled on the InduRed concept, were used to evaluate the transfer coefficients of Ni, Co, Mn and Li in qualitative terms. It was shown that Ni and Co seem to be fully recoverable by this technology while parts of manganese are removed from the reactor via the gas phase. For Li, which is considered to be the bottleneck of pyrometallurgical LIB recycling approaches, very promising results have been seen. In the InduMelt experiments with NCA and NMC more than 90%, respectively more than 75% in the experiment with black matter, of the initial Li were removed from the reactor. The fact that Li does neither accumulate in the slag nor in the metal phase indicates a high potential of the technology to enable new possibilities for Li recovery from the LIB waste stream. If Li is not obtained in small amounts in a slag phase, as in other processes, but can be collected in a separate material flow, its recovery from there can potentially be achieved with less effort and therefore represented more economically.

In order to better examine the removal of Li and Mn from the reactor, the experiments are going to be repeated using a gas vent with gas scrubbing. This should clarify in which form the Li can be obtained from the exhaust gas and how its recovery from there could be achieved. Furthermore, new cathode materials like NMC in other configurations (811, 622, 532 instead of 111) as well as LFP (lithium iron phosphate) are planned to be investigated regarding their suitability for treatment in the InduRed reactor.

The results from experiments with black matter (AM) showed some significant differences, which could partly be attributed to residues from the pre-treatment or excessively high temperatures during the thermal deactivation. Since the contrary behavior of AM in all experimental series cannot be fully elucidated with the available data, further research and experiments are necessary. In addition to that, it is planned to investigate black matter from different pre-treatment processes and the influence of interfering species like Cu or Al in general.

**Author Contributions:** Conceptualization, S.W.-K. and A.H.; methodology, S.W.-K.; investigation, S.W.-K., A.H. and C.P.; resources, S.W.-K., A.H. and C.P.; writing—original draft preparation, S.W.-K. writing—review and editing, S.W.-K., A.H., C.P. and H.R.; visualization, S.W.-K.; supervision, C.P. and H.R.; project administration, C.P.; funding acquisition, H.R. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research received no external funding.

**Institutional Review Board Statement:** Not applicable.

**Informed Consent Statement:** Not applicable.

**Data Availability Statement:** The data presented in this study are available on request from the corresponding author.

**Conflicts of Interest:** The authors declare no conflict of interest.

### **References**

