Battery modules compositionSome very prevalent devices with LiBs are laptops, smartwatches, wireless earbuds, powerbanks and mobile (smart) phones1. From each of these applications, multiple battery modules have been disassembled. In Fig. 1, one example of a battery module (sometimes with respective device) can be seen for each mentioned application. The battery module is made of five major components: the battery cell(s), battery casing, the battery management system (BMS), wiring and separators or glues. The battery cells contain the active materials and are used for the storage of electrical energy. The wiring connects multiple cells to the BMS, which in turn ensures safe charge and discharge of the battery cells. The separators are used to avoid short circuiting by direct contact of the cells, and the glue keeps the cells in place during use. Lastly, the casing encapsulates and protects all other components. After disassembly, the average relative weights of all components from the battery modules were calculated. These are compared in Fig. 2 for the main five components.Figure 1Battery modules (some of which with device) before and after disassembly. Represented are a (disassembled) laptop battery module (a, b), smart watch (c, d), set of wireless earphones (e, f, g), powerbank (h, i) and mobile phone module (j, k). For all sub images, the numbers represent the same component: 1. BMS, 2. Battery casing, 3. Battery cells, 4. Wiring, 5. Glue and separating components.Figure 2Proportional composition of battery modules from laptops, smartwatches, earbuds, powerbanks and (smart)phones (sm. phone). Note that the x-axis starts at 50 wt%.The laptops (3 devices) researched in this study all included a rather easily removeable module, one of which is shown in Fig. 1a. Figure 1b shows this same battery module after disassembly. For this application the cells, either 18650 type as seen in the picture or pouch type, take up most of the weight (85 wt%). This is followed by the battery casing with 11 wt%. The BMS (1.5 wt%), glue and separators (1.0 wt%) and wiring (0.8 wt%) are only minor components.The disassembled smartwatches (4 devices) use a smaller battery module with one cell (Fig. 1c). These modules consist of a pouch type cell (93 wt%), to which a BMS (5 wt%) is attached. The battery casing consists of a plastic foil (1.2 wt%) wrapped over the BMS. Hard plastic casings, such as seen in the laptop battery modules, are not present. The wiring and glue or separators have a very low relative weight of 0.7 and 0.2 wt% respectively. An example of the disassembled module is shown in Fig. 1d.All disassembled earbud sets (4 devices) (Fig. 1e) contain three battery modules: one large module (Fig. 1f) and two smaller modules (Fig. 1g). These mostly consist of a cell (87 wt%) to which a BMS is attached (9 wt%). It must be noted that the BMSs shown in Fig. 1f and g are not present in all modules that were retrieved. In two units, the cells are directly attached to the circuitry in the earplugs or the casing, avoiding the need for an extra BMS. The battery casing consists, similarly to the modules present in smartwatches, of a foil wrapped around the module (1.2 wt%). The wiring and glue or separators contribute 2.2 and 0.3 wt% respectively.An example of a powerbank is shown in Fig. 1h. Figure 1i shows this powerbank after disassembly. The disassembled powerbanks (5 devices) contain one to three cells, either 18650 or pouch type, which take up 71 wt% on average. The casing and BMS take up 22 and 6.6 wt% respectively. The wiring and glue or separators are only minor components, taking up 0.5 and 0.3 wt% respectively.The array of (smart) phones (10 devices) that were disassembled contain two types of battery modules, all with one single pouch type cell. One module type is similar to the modules found in the smartwatches and earbuds, albeit larger. The second type is shown in Fig. 1j and is easily removeable from the device. These were generally found in the older (smart) phones (2008–2015), but it also depended on the brand. The module after disassembly can be seen in Fig. 1k. The cell (94 wt%) is similar in shape to the pouch cells seen in Fig. 1i, except for the connectors. Instead of a broad wire, flat connecting points are used on which the BMS (1.7 wt%) is directly applied. Some modules such as the one shown make use of wiring (0.7 wt%) to attach the BMS to (one of) the connecting points. The module casing consists of a sticker wrapped around the cell, in combination with plastic parts to cap off the top and bottom of the module (3.7 wt%). Glue and separators are used very scarcely (< 0.1 wt%).When comparing the different applications with each other, major variations can be seen. The powerbanks and laptop battery modules have multiple cells which need to be kept in place, connected with wiring and isolated by rubber pads to avoid short circuiting. Therefore, the relative weight of the cells is lower (71–85 wt%) than for applications with just one cell per module (87–94 wt%). The smaller modules consist of one cell (87–94 wt%) to which the other parts are attached. These modules are generally not easily removed from the device since they are kept in place by a glue. In this manner, the device itself also acts as protective cover. Therefore, the need for casing, internal wiring and separators is limited. This results in a large weight contribution of the battery casing in the applications with multiple cells (11–22 wt%) compared to applications with one cell per module (1–4 wt%). Divergent are the easily removable batteries from (older) mobile phones, as seen in Fig. 1j, of which the casing contains hard plastic parts to protect the BMS during removal and placement. This results in the higher casing contribution for the mobile phones. The BMS is relatively large for smaller modules, resulting in a high weight contribution for the earbuds (9.1 wt%) and smartwatches (4.7 wt%). The BMS contribution is lower for the larger modules from smartphones (1.7 wt%) and laptops (1.9 wt%). This does not hold for the powerbanks, since the BMS often also harbours a screen and multiple charging ports resulting in a higher relative weight. The two other components (wiring, glue and separators) are similar for all categories and mostly depend on the size of the module. It must be noted that the limited number of samples in this study provides a good illustration of the typical components in consumer e-waste stream containing LIBs, but is not exhaustive and further variation can be expected.Battery cells compositionAfter removing the cells from the battery modules, they were placed in the salt electrolyte for discharging before disassembly. Three typical examples can be seen in Fig. 3, before and after disassembly. The two most popular types of cells, cylindrical (typically 18650) and pouch cells, are shown. Although battery cells vary in size and shape, their basic structure is identical. Generally, a battery cell consists of the cell casing, an anode (graphite (C) on copper foil), a cathode (lithium metal oxide on aluminium foil) and a polymer separator with a supporting electrolyte in which lithium ions move1. The electrode foils (Cu, Al) are present to transfer electrons from and to the electrode materials via an external circuit. All components are indicated by the numbers 1–4 in Fig. 3. The average proportional composition of both cell types is compared in Fig. 4.Figure 318650 type (a, b) and two pouch type (c–f) battery cells before and after disassembly. For all sub images, the numbers represent the same component. 1. Cell casing, 2. Anode, 3. Cathode, 4. Separator.Figure 4Average proportional composition of pouch and 18650 type cells. Subtraction of the component weights from the initial cell weight results in the loss category. It includes evaporated electrolyte and potential loss of (active) materials during cell opening.Figures 3a and b show a 18650 battery cell before and after disassembly, respectively. In these cells, the cathode is most prevalent (44 wt%), followed by the anode (32 wt%) and cell casing (17 wt%). The separator and loss have a low contribution of 3.6 and 2.7 wt%, respectively. Figure 3c and e both show pouch type cells. Figure 3d and f show these same cells after opening. Within these cells, the cathode has again the highest component share (51 wt%), followed by the anode (38 wt%). The cell casing, separator and loss all have a low contribution, being 3.3, 4.1 and 3.5 wt%, respectively.When comparing the two cell types, the main difference is found in the contribution of the cell casing. This is much larger for the 18650 cells, which have a steel casing, compared to the pouch cells, with a plastic one. When the cell casing is excluded, the relative share of the other components is the same for both cell types. The composition of the 18650 cells found here is in good agreement with earlier studies when considering the contribution of the cell casing and separator41,45. Differences are found when comparing the anode and cathode, as well as the electrolyte. The latter is not considered in our study, since it is impregnated on the separator, cathode and anode. Also, a part of it evaporates after opening the cells (this is incorporated in the “loss” category). The electrolyte is therefore not selectively removed and weighed. This results in the anode and cathode categories being larger since they include some remaining electrolyte. Also, the amount of cathode or anode material that is attached to the collector foil can vary amongst different manufacturers, resulting in a different component share46.Black mass composition and morphologyTo study the variety in composition and morphology of BM from Li-ion battery waste streams in consumer electronics, samples from the previously disassembled cells were characterized. The SEM results can be seen in Fig. 5. The explanation of these results, as well as the prevalent phases according to XRD, are presented in Table S1. Their elemental compositions according to ICP-OES are depicted in Table 1.Figure 5SEM images of BM retrieved manually from batteries (a–h) and industry (i). Individual grains were characterised by SEM–EDX.Table 1 Elemental composition (wt%) according to ICP-OES analysis of various manually acquired BM types, and one industrial sample.The first two samples, which are shown in Fig. 5a and b and labelled NMC 1 and NMC 2 respectively, both consist of a mixed lithium metal (nickel, cobalt and manganese) oxide. They are present as small particles (1–4 µm), as well as larger coagulates of these particles (8–20 µm). XRD-analysis confirms the presence of a mixed lithium metal oxide in both NMC 1 and NMC 2. ICP-OES-analysis (Table 1) shows that Ni is the most prevalent element in both samples (29–35 wt%), followed by Mn (17–18 wt%) and Co (10–12 wt%). The samples consist of around 5.5 wt% Li. Al is present as impurity in both samples (0.2–0.4 wt%). 31–36 wt% is taken up by other compounds or elements, such as graphite, binder and O.The samples shown in Fig. 5c and d consist of an oxide of Li and Co, labelled as LCO 1 and LCO 2, respectively. The grain size ranges from 2 to 40 µm in LCO 1, and from 10 to 30 µm in LCO 2. According to XRD-analysis, both samples consist of LiCoO2, however LCO 1 also contains CoO2. ICP-OES-analysis shows that cobalt is the most prevalent element in both BMs (58–59 wt%), followed by Li (6.5–6.6 wt%). LCO 1 contains a small amount of Al as impurity (3.5 wt%), while LCO 2 contains some Ni (1.5 wt%). Both samples also contain 32–33 wt% of other compounds or elements.Figure 5e, f and g show samples that consist of blended cathode material, called LMO + NMC, LCO + NMC 1 and LCO + NMC 2 respectively. These blended cathode materials are a more recent development in LiB technology, and will therefore be increasingly expected in e-waste in the coming decades47,48. Cathode blending is done to complement positive aspects of certain cathode chemistries, while also mitigating their drawbacks48. XRD-analysis indicates that the first sample consists of LiMn2O4 + Li1.2Mn0.6Ni0.2O2, whereas the other two samples consist of LiCoO2 and LiNixMnyCozO2. According to ICP-OES-analysis, LMO + NMC mostly consists of Mn (41 wt%), followed by Ni (12 wt%), Co (4.6 wt%) and Li (4.4 wt%). Al is present as impurity (2.5 wt%). Both LCO + NMC samples mostly consist of Co (54–53 wt%). Ni (6.6–6.3 wt%) and Mn (3.1–3.4 wt%) are present in lower amounts. Both also contain a small amount of Al (0.1–1 wt%). Other compounds take up 34.2, 29.2 and 29.1 wt% respectively.Figure 5h shows BM that is retrieved from lithium iron phosphate (LFP) battery cells. This type of cathode material is increasingly used due to its higher stability during charge and discharge and its lower cost49,50. However, its chemical properties also result in a lower capacity49,50. The grains look more like flakes and range from 1 to 15 µm in size. According to XRD-analysis, this sample consists of LiFePO4, FePO4 and graphite. ICP-OES-analysis shows that of the LFP black mass, 33 wt% is Fe, and 52 wt% is PO4. The Li content is 4.1 wt% and Al is present as impurity (0.3 wt%). 10.8 wt% of the black mass is taken up by graphite, O and others.Lastly, an industrial BM is shown in Fig. 5i. Its pre-treatment differs from the one applied to earlier mentioned BMs; the battery cells are shredded entirely instead of disassembled manually (the latter including separation of anode and cathode). Also, the BM is pyrolyzed to remove organic components such as the binder and electrolyte components. Therefore, more impurities are to be expected. The BM consists of a multitude of compounds, such as NMC, LCO and LFP as well as graphite, which is used as anode active material and is not present in previous samples. Particles of Al and Cu foils are also present. These result from breakage of the cathode and anode collector foils, respectively. Al was present in some previous samples, whereas Cu was not. XRD-analysis confirms the presence of these phases, as well as Cu0.2MnNi5.8O8, Cu0.85Fe0.1O and Li2CO3. Also, a part of the Ni is present in metallic form. This is due to it being present in the metallic shell of the 18650 cells, which is not removed in industrial pre-treatment. According to ICP-OES-analysis, Co is the most prevalent element (18 wt%) in the industrial sample, followed by Ni (12 wt%), Mn (4.7 wt%) and Li (4.2 wt%). Al (2.8 wt%) and Cu (2.1 wt%) are both present in a small amount, as well as Fe (0.4 wt%). Other compounds and elements, such as graphite, binder, and O take up 55.4 wt%.When comparing all these different BMs, it can be seen that there is a number of different cathode chemistries present in various electronic devices. Samples can be high in Co (LCO), Mn (LMO), Ni (NMC 1 and 2) or contain Fe and P (LFP). Sometimes, a blend of different chemistries is used (LMO + NMC, LCO + NMC 1 and 2). No direct link is found between the type of device and the cathode chemistry, suggesting that sorting per device will not reduce the variation in chemistries. Other elements such as Al are often present as impurities. It is important to note that large standard deviations of the Al content were found since in some batteries, the collector foil was so thin that it broke by the ultrasonic vibrations during the liberation process.The industrial sample seems to be a mixture of all other investigated battery types, as Li, Co, Ni, Mn, Al, and Fe are all present. Mechanical processing and pyrolysis are generally able to remove most of the plastics originating from the module casings, glue and separating components as well as the pouch cell casings and separators42. The metallic Fe, which can originate from the cell casing of cylindrical cells, can be largely removed by magnetic separation42. In addition, it contains impurities of Cu and Al, likely from the current collectors. These are, on industrial scale, separated by multiple sieving steps and electrostatic separation, however, certain level of impurities is always present after mechanical processing42. Lastly, the industrial BM contains a lot of graphite, as seen on the SEM-image (Fig. 5i). This is the active material of the anode, which could be separated by flotation or removed by heat-treatment42. Alternatively, it can also be removed as residue after leaching42.Effect of black mass composition and H2O2 addition on leachingA selection of BMs that were retrieved in the previous sections, as well as a pristine sample, were submitted to the same leaching conditions. This provides insight into the influence of BM composition on the leaching efficiency of the contained elements. The following conditions were chosen as a benchmark based on a literature survey20,22,28,51. All samples were submitted to a lixiviant with 2 M H2SO4 for 2 h at 50 °C and S/L of 60 g/L. The addition of H2O2 is varied from 0 to 4 vol% to study the required extent for a reductive agent, which improves dissolution of the TMs52. In order to perform leaching experiments on multiple BMs, only 0, 1 and 4 vol% of H2O2 addition were applied. The driving reaction in this leaching system is described in Eq. (2) 22. Note that this reaction assumes an equal presence of Ni, Co and Mn, which is not always the case in this study.$$6 Li{Ni}_\frac{1}{3}{Mn}_\frac{1}{3}{Co}_\frac{1}{3}{O}_{2 s}+9{H}_{2}S{O}_{4 aq}+3 {H}_{2}{O}_{2 aq}\to 3 {Li}_{2}S{O}_{4 aq}+2 NiS{O}_{4 aq}+2 CoS{O}_{4 aq}+2 MnS{O}_{4 aq}+12 {H}_{2}O+3 {O}_{2 g}$$
(2)
The results of the leaching experiments for the different chemistries are illustrated in Fig. 6. Since the chosen manual liberation method resulted in a generally low amount of impurities (mostly Al), only the leaching efficiencies of the relevant elements; Li, Ni, Mn and Co are shown. It is important to note that sometimes the leaching efficiency exceeds 100%, which is not possible in practice. Evaporation of water from the PLS during the experimental phase could lead to an overestimation of elemental concentrations. Also, while both initial composition and leaching results were performed multiple times, the inherent heterogeneous nature of BM leads to an estimated error of 4–5%.Figure 6Leaching efficiencies of Li, Ni, Mn and Co from different BM samples as a function of the amount of H2O2 addition. Other leaching conditions are 2 M H2SO4, S/L = 60 g/L, T = 50 °C, t = 120 min.Without addition of H2O2, the leaching efficiency of all elements except Li is very poor. Li is leached between 68 and 90%, whereas dissolution of the TMs ranges between 25 and 45%. An increase in the H2O2 concentration generally improves the leaching efficiency of all relevant elements. This is already seen when 1 vol% of H2O2 solution is added, at which point the leaching efficiencies increase from 72 to 95% for Li and from 40 to 70% for the TMs. However, 100% leaching efficiency is not reached for all samples, despite their similar trends, even at 4 vol% H2O2. This is observed in the leaching of both mixed LCO and NMC chemistries, showing much lower efficiencies (approx. 88%, 89%, 93% and 97% for Li, Co, Ni and Mn, respectively), but comparable to each other. On the other hand, the two NMC samples (NMC 1 and NMC 2) show much more promising results, but with a slight discrepancy between the two samples (100% and 95% leaching efficiency, respectively) for TMs, while Li is leached for 100% in both cases. The pristine NMC 532 shows the best leaching behaviour, even at low (or zero) H2O2 concentration. At 4 vol% H2O2, both the TMs and Li are fully leached.The observed trend concerning the H2O2 concentration, as well as the distinct leaching behaviour of Li compared to the TMs is in agreement with earlier studies22,28. It is to be expected that 100% leaching efficiency will be achieved for all BMs with further increase of the H2O2 concentrations53. Noticeably, however, the leaching trend of the industrial BM is completely different from all other BM samples. Strangely, Li, Co and Mn are fully leached without addition of any H2O2. However, at this point, Ni is only leached for 61% and does not exceed 80% at 4 vol% H2O2 solution. The high observed leaching efficiencies can be attributed to the low metal concentration in the feed, as seen in Table 1. It contains a lot of graphite, which does not dissolve during leaching but does contribute to the S/L, which was kept constant in all leaching experiments. Therefore, the total amount of metals in the feed is much lower relative to other samples. This dilution of the target metals by other components, such as graphite, plastics, binder and current collectors (Al, Cu) is also reported in other studies54. Also, it is known that pyrolysis of BMs containing graphite can induce reactions between the present metals and graphite, causing pre-reduction of the TMs55. As a consequence, leaching efficiencies are much higher at a lower addition of H2O2. Ni is the only element that does not fully dissolve, which is a result of it being present in metallic form, as observed earlier.Li leaching generally shows a slightly different trend compared to the transition metals. At lower H2O2 concentrations it is not completely leached, but it varies from 70 to 90%. This is due to the good solubility of Li+ and the lower binding energy of Li and O, compared to the other TMs28,56. However, for Li to be fully leached, the solid particles need to be broken down first, enabling contact between the Li present inside these particles and the dissolving acid. On the other hand, the TMs dissolve most efficiently in oxidation state + II31. However, these are also present in state + III (Ni and Co) and + IV (Co and Mn)57,58. Hence, they need to be reduced before dissolution, which is manifested by the reaction with H2O253. When the average oxidation state is higher, a larger amount of H2O2 is needed to reduce the TMs to the + II state31. The leaching results suggest a different magnitude of H2O2 consumption for the tested BMs, and therefore a different distribution of the oxidation numbers of the TMs in these samples. To study this in more detail, some of these samples have been subjected to leaching with an excess of H2O2 in combination with KMnO4-titration.When leaching with 10 vol% H2O2 solution, all the Li and TMs in NMC 1, NMC 2, the industrial BM and the pristine NMC 532 were transferred to the PLS according to ICP-OES analysis. Afterwards, the remaining H2O2 was titrated with a KMnO4 solution to calculate the H2O2 consumption, which is expressed as moles of H2O2 consumed per mole of TMs. Results show that the dissolution of NMC 2 consumes more H2O2 compared to NMC 1 (0.91 and 0.87, respectively). Dissolution of the industrial BM requires about 3 times more H2O2 (2.7 mol per mole of TMs). The dissolution of the pristine NMC 532 results in the lowest ratio (0.67). These differences in consumption of the reducing agent indicate, according to Eq. (2), the difference in oxidation states of the TMs in the NMC samples. The fact that the industrial BM requires so much reducing agent compared to the other samples is unexpected. One explanation is the high content of impurities which can react with H2O2, such as Cu, Al and Fe59,60,61. Therefore, the high consumption is not only related to the oxidation number of the target TMs and cannot be directly compared with the other samples. The H2O2 consumption of NMC 2, NMC 1 and pristine (NMC 532) is in agreements with the leaching results observed earlier, as leaching efficiencies follows the same respective order NMC 2 > NMC 1 > pristine concluded for increasing oxidation states of TMs. The titration experiments suggest a higher average oxidation states of Ni and Co in the used samples (NMC 1 and 2) compared to pristine material, which directly influences their leaching efficiencies.Generally, the leaching experiments show that it is possible to leach all the different chemistries together at high H2O2 concentration. It has to be noted that not all CAM chemistries available on the market were used in the leaching experiments. These may be present in an industrially processed BM, as is shown earlier, and lead to even more divergence in leaching efficiencies. To accommodate for variations in chemistries, a high amounts of reagents is needed, putting extra pressure on the sustainability of the recycling process as a whole. Also important to note is that in the leaching of the industrial BM, impurities end up in the PLS. Table S2 in the supplementary information shows the composition of the PLS, acquired after leaching the industrial BM with 4 vol% H2O2-solution. It can be seen that Cu, Al and Fe contaminate this PLS. In order for any recycled Li, Co, Ni and Mn products to meet industrial standards, these elements need to be removed. There are various methods to serve this purpose, such as solvent extraction and subsequent stripping and scrubbing steps, precipitation and recrystallization or electrowinning62. However, these steps can be energy intensive and produce (indirect) emissions.
A closer look at lithium-ion batteries in E-waste and the potential for a universal hydrometallurgical recycling process
