Xiong
Xiao
a,
Billy W.
Hoogendoorn
a,
Yiqian
Ma
b,
Suchithra
Ashoka Sahadevan
c,
James M.
Gardner
c,
Kerstin
Forsberg
b and
Richard T.
Olsson
*a
aDepartment of Fiber and Polymer Technology, KTH Royal Institute of Technology, Teknikringen 56, 11428 Stockholm, Sweden. E-mail: rols@kth.se
bDepartment of Chemical Engineering, KTH Royal Institute of Technology, Teknikringen 42, 11428 Stockholm, Sweden
cDepartment of Chemistry, KTH Royal Institute of Technology, Teknikringen 30, 11428 Stockholm, Sweden
First published on 29th September 2021
An ultrasound-assisted extraction (leaching) method of valuable metals from discarded lithium-ion batteries (LiBs) is reported. Mild organic citric or acetic acids were used as leaching agents for a more environmentally-friendly recovery of the lithium, nickel, cobalt, and manganese from the discharged and crushed lithium nickel-manganese-cobalt oxide (NMC) LiBs. The extraction was performed with the presence/absence of continuous ultrasound (US) energy supplied by a 110 W ultrasonic bath. The effect of temperature (30–70 °C), reducing agent concentration (H2O2: 0–2.0 vol%), as well as choice of specific acid on the metal dissolution were investigated. The US leaching decreased the leaching time by more than 50% and improved the leached percentage of Li, Mn, Co, and Ni due to the local heat and improved mass transfer between solid and liquid interfaces in the process. The X-ray diffraction results of residues from the US leaching further confirmed an improved dissolution of the crushed layered NMC structure, resulting in the significant improvement of the leached amounts of the valuable metals. Furthermore, it is demonstrated that using citric acid eliminated the need of additional reducing agents and suppressed the dissolution of copper (Cu) due to its inhibitor effect on the Cu surface, i.e. compared with using acetic acid as leaching reagent. Overall, it is shown that recovery of the battery metals can be facilitated and carried out in a more energy-efficient manner at low temperatures (50 °C) using ultrasound to improve metal ions mass transportation in the residue layers of the NMC during the organic acid leaching.
Previously, ultrasound was demonstrated as an efficient and green technology that can be used in the process of extraction from biological samples. Most ultrasonic baths send mechanical pressure waves with frequencies greater than 20 kHz through the liquid medium. By sending the ultrasonic waves through an aqueous medium, microbubbles are formed.15 The collapse of the bubbles generates local temperatures of 5000 K along with the production of highly reactive free radicals.16 The events will cause physical effects, such as turbulence, shear forces, shock waves, and microjets, resulting in increased mass transfer while degassing the solution of the dissolved gases. Ultrasound may however also allow for improving mass transfer in the extraction of valuable battery metals to an extent that more environment-friendly organic acids can be used in the recovery of the metals. So far, investigations on the extent of the improved efficiency of metal extraction from used LiBs using ultrasound are scarce.12,13,17,18 In these studies, the ultrasound extraction of metals from LiBs focuses on the dissolution of LiCoO2, LiNiO2, and LiMn2O4, and occasionally reports on the leaching kinetics of these materials with the complexity of added reducing agents. The more environment-friendly acids, so far reported, include formic acid,19 lactic acid,14 acetic acid,20 and citric acid.21 In the absence of ultrasound, these weaker organic acids would normally be regarded as insufficiently strong for efficient extractions and require the use of additional chemicals or pre-treatments steps before the actual extraction experiments (leaching process) to reach recovery levels >95%, see Table S1.† In LiBs recycling, low and high-power ultrasound vibration was used to assist the electrode delamination process during disassembly of batteries.22
In this work, the ultrasound-assisted extraction of the metals from lithium nickel–manganese–cobalt oxide (NMC) batteries (the most common lithium battery type) is reported for the first time, and the kinetics of the extraction is described in detail. The difference between leaching with, and without ultrasound, using weak acetic or citric acids is shown. Citric acid was used as a natural, mildly reducing (E0 = −0.18 V vs. NHE), weak triprotic acid that is commercially produced by fermentation.18,23 It is ‘green’, abundant, comparatively inexpensive (to other organic acids), and has been widely used by the food and beverages industry, and as a non-toxic household detergent for years.24 The citric acid was compared with acetic acid due to its inexpensive nature at high purity, and high availability, although not as acidic as the citric acid i.e. being the sole organic acid with a pKa value of 4.7625 (see Table S2†). The batteries used were shredded 24 kW h Volvo C30 NMC batteries. The leaching processing was however carried out as a universal approach, making the demonstrated methods applicable to work with other battery types. It is shown that ultrasound significantly increases the leaching rate of the battery metals during the processing. The leaching time was reduced by more than 50% when using a standard laboratory ultrasonic bath, as opposed to when carrying out traditional extraction in a stirred reactor. Almost complete metal ion extraction is demonstrated, using the acids in combination with the ultrasound. The citric acid further improved the leached percentage of lithium, nickel, cobalt, and manganese while suppressing the dissolution of Cu, when compared with acetic acid. The results were confirmed with multiple measurements and characterizations of all battery residues before, and after, the leaching process.
![]() | ||
Fig. 1 (a) Barrels of discharged and ground batteries, (b) SEM images of fine ground LiBs powder, and (c) the fine ground powder in high magnification, (d) size distributions of fine ground powder derived from >50 micrographs using min. 600 particles for accurate statistics, (e) metal elements content variation (ICP-OES measurements) from top to bottom of the Barrel 1 and 2, respectively, (f) EDS maps taken from (c). Triple samples tested on each location in the barrels confirmed the nonuniform powder composition from the natural fractionation that had occurred in the coarse ground powder, demonstrated by the error bars in Fig. 1e. |
![]() | (1) |
![]() | (2) |
Fig. 1e shows the elemental presence in the top and bottom powder fractions of the barrels for the fine and coarse ground powder, as a comparison. The coarse ground powder contained a markedly higher content of aluminium and copper compared to the more valuable metals Co, Ni, and Mn (right hand side columns, Fig. 1e). The coarse ground powder also demonstrated a wider particle size distribution (Fig. S2b†), which explained a more extensive fractionation that had occurred during the barrel transportation. It was thus evident from the elemental analysis and microscopy that only the fine ground powder, with almost identical metal composition on the top and bottom (left hand side columns, Fig. 1e) was useful as starting material for developing optimized extraction protocols, i.e. due to its more uniform composition. The fine ground powder was therefore further analysed and used in all the remaining experiments. The larger size particles in Fig. 1b (representing less than 5% of the particle fraction, although a considerable part of the metallic mass content) was however firstly filtered off and identified as composed of aluminium or copper (Fig. S2†). The larger sized particles were accordingly excluded from the extraction protocols focusing on the active battery materials.
Fig. 1c shows a magnified area in Fig. 1b, displaying the finer particles, with the dominant phase represented in particle sizes by the histogram shown in Fig. 1d. Fig. 1f shows the EDS mapping images of the same area displayed in Fig. 1c. The detected elements were Co, Ni, and Mn (top row), while the bottom row shows C, O, and F. The latter elements were remains from the conductive additives and binders used in the LiBs fabrication process. Lithium is not shown in Fig. 1f since it was not detected by the normal EDS probe analysis due to its low characteristic radiation energy.27 For comparison, Fig. S3† shows the EDS mapping of the residue of the same LiBs powder after aqua regia leaching digestion, demonstrating that the aqua regia leaching digestion could effectively be used for complete extraction (100%) of the Ni, Co, to obtain the reference values synonymous with complete extraction of all active battery metals from the remaining carbon residuals.
Fig. 2 shows the TGA curves and XRD patterns of the fine ground LiBs powder before and after aqua regia leaching digestion. TGA was carried out under O2 flow to ensure that all organic materials in the fine ground LiBs powder were removed, and its relative content could be determined. The results show that before aqua regia leaching digestion about 61.8 wt% residue remained after heating to 950 °C, which was associated with the metal fraction of the battery powder. After aqua regia leaching digestion, less than 1 wt% material remained although a small amount of ash existed. The results were consistent with an almost complete oxidation of the organic materials (Fig. 2a), at a temperature above 500 °C. Moreover, XRD characterization was used to investigate the crystalline phases in the fine ground powder before and after the aqua regia leaching digestion.
The peaks in the diffractograms before leaching originated from the layered NMC phase (active material), Al, Cu, and graphite, respectively, Fig. 2b (top). After the aqua regia leaching digestion, the remaining material was graphite and polypropylene (PP). The PP came from the separator/battery casing, see Fig. 2b (bottom). The TGA and XRD of powders before and after aqua regia leaching agreed with the EDS mapping and confirmed that a full extraction of the metal elements had been carried out.
ICP-OES was used to carry out the quantitative analysis of the aqua regia leachate composition. The concentration (mmol g−1) and standard deviation of all metals are summarized and shown in Table 1. Fe was detected in the fine ground LiBs powder at a concentration of 0.048 mmol (pristine powder) g−1, which is not included in Table 1 due to its comparably small amount. The metal mole-to-mole ratio of the active material Li:
Mn
:
Co
:
Ni was ca. 1.289
:
0.315
:
0.319
:
0.366, reflecting that the composition of active material (LiMn0.315Co0.319Ni0.366O2) and electrolyte in the hybrid EV battery was closed to LiNi1/3Co1/3Mn1/3O2. The phase composition was also in agreement with the diffraction pattern in Fig. 2b (top), although the additional 29% Li originated from the electrolyte (LiPF6) that was present in the sample. The Al and Cu in Fig. 2b originated from the current collector and were not shown in the EDS mapping (Fig. 1f) due to their large particle sizes and coincidentally were found in similar amounts as the targeted metal ions (Ni, Co, and Mn). In total, the weight of Al, Cu, and the NMC phase occupied ca. 65.7 wt% of the fine ground powder, which was in good agreement with the remaining residue percentage (61.8 wt%, Fig. 2a) obtained from the gravimetric measurements.
In summary, the characterization in terms of the natural fractionation and deviation of uniformity (coarse vs. fine ground powder), the phase characterization, and the chemical composition allowed only the fine ground powder to be identified as a useful reference material. The coarse ground powder was always displaying different powder characteristics at different locations in the barrel and was therefore discarded as a source for more experiments.
ICP-OES result | Metal elements (mmol (pristine powder) g−1) | |||||
---|---|---|---|---|---|---|
Li | Mn | Co | Ni | Al | Cu | |
Comment: The data for each ion was acquired as repeated twice with separate ICP-OES measurements. The deviation was maximum ±3% with an exception of Cu (maximum ±7%). | ||||||
Oil bath, 323 K | 89% | 89% | 87% | 91% | 48% | 83% |
US, 323 K | 94% | 96% | 100% | 99% | 60% | 48% |
Oil bath, 343 K | 90% | 91% | 93% | 92% | 78% | 54% |
Table 3 show over time that the active battery metals could be extracted much faster using ultrasound. The required times for reaching 20%, 50%, 80%, and 90% of the total amount of extractable metal ions (Li, Ni, Co, and Mn) are shown in Table 3.
Times (t, min) | Li | Mn | Co | Ni |
---|---|---|---|---|
Comment: The data in the table with parentheses represent the required time in an oil bath. Data from Fig. 3, see dashed lines for 80 and 90%. | ||||
t 20% | 20 (20) | 44 (40) | 43 (45) | 47 (41) |
t 50% | 75 (75) | 140 (250) | 160 (290) | 150 (240) |
t 80% | 270 (650) | 430 (975) | 420 (1125) | 420 (940) |
t 90% | 600 (>1440) | 790 (>1440) | 685 (>1440) | 675 (1440) |
Among the metals, the fastest extraction was observed for the lithium, showing that 80% of the total amount of extractable Li (t80%) could be reached within 270 min. The same extent of extraction, using traditional stirring, required over 650 min to be completed. At the same time, 420 min was needed to extract 80% of the cobalt (t80%) using the ultrasound in comparison to 1125 min using the traditional extraction, see Table 3 and Fig. 3.
The X-ray diffraction patterns of residues, as dependent on the leaching time with/without ultrasound, are shown in Fig. S5† and Fig. 4. From 0 to 720 min, a shift of the 19.05° peak (representing the NMC interlayer direction), occurred from a higher to a lower angle (Δ 0.8°, see Fig. 4a). The shift was caused by the proton exchange accompanied with water molecules28 intercalated within the NMC layered space, resulting in an expansion of the interlayer distance, see Fig. 4b. With an additional increase in the leaching time to 960 min, the peak shifted back to a higher angle due to the destruction of the layered NMC phase and the formation of a spinel-type LiMn2O4. The spinel-type LiMn2O4 can retain its structure even when exposed to harsh HCl acid solutions.29 After 1440 min leaching, the diffractogram peaks associated with the LiMn2O4 spinel structure disappeared in the residues from the US leaching method while they remained in the residues from the oil bath leaching at 323 K. The results suggested that the ultrasound not only assisted in the diffusion of the valuable metal ions into the solution, but also had an impact on the dissolution of the more stable LiMn2O4 spinel structure.
![]() | ||
Fig. 4 (a) X-ray diffraction patterns of residues in the leaching system with/without ultrasound depending on leaching time at 323 K. (b) Crystal structures of LiCo1/3Ni1/3Mn1/3O2 (NMC) and (c) spinel-type LiMn2O4 (draw by VESTA30), Li (green ball/octahedral), Ni/NiO6 (grey ball/octahedral), Co (blue ball/octahedral), and Mn (purple ball/octahedral), vacancy (white ball). |
The only exception from the general observation that ultrasound assisted leaching outperformed the traditional oil bath leaching was the Cu element. The leached percentage of Cu at 1440 min was supressed during the leaching, see Table 2. This was related to the depletion of dissolved O2 in the solution (caused by the intense ultrasound), which have been reported to interact with Cu and citric acid in conditions without any oxidizing agents.31 For this reason, ultrasound has traditionally also been used for degassing and release of gases from solutions.32,33 A lower O2 concentration consequently interfered with the Cu oxidation and dissolution mechanisms as previously reported.31 This would be in line with the results showing that extraction carried out at higher temperature in the oil bath, see Table 2 (343 K) also suppressed Cu leaching since a higher temperature results in a lower percentage of naturally dissolved O2.34
(1) Liquid boundary layer mass transfer contribution (eqn (3)):
![]() | (3) |
(2) Surface chemical reaction contribution (eqn (4)):
![]() | (4) |
(3) Residue layer diffusion contribution (eqn (5)):
![]() | (5) |
X = k1t | (6) |
1 − (1 − X)1/3 = k2t | (7) |
1 − 3(1 − X)2/3 + 2(1 − X) = k3t | (8) |
![]() | ||
Fig. 5 Kinetics analysis during the leaching of the fine ground LiBs powder with/without ultrasound, adapting leaching time data to the residue layer diffusion contribution (eqn (5)) of the shrinking core model. Conditions: citric acid, 1.5 mol L−1; S/L, 25 g L−1, H2O2, 0 vol%; temperature, 323 K; stirring speed, 500 rpm. |
The results show that for both systems (with/without ultrasound), the fitting lines of Li, Mn, Co, Ni, and Al were in agreement with the residue layer diffusion contribution to the shrinking core model. A linear relationship (k3) with a higher regression coefficient (R2) value (≥0.9855) could be established, whereas no relationships could be established for any of the other contributions to the shrinking core model (Fig. S6†).
The specific R2 values are shown in Table 4 for the different ions. The only exception was the leaching of Cu that instead reasonably well correlated with the reaction control contribution, showing a R2 value of 0.989 when the extraction was done in absence of ultrasound (Fig. S6†). This result again suggested that having oxygen dissolved in the solution contributed to a uniformly progressing Cu dissolution reaction (see section 3.2). Overall, it was evident that the rate limiting contribution to the US leaching behavior could be assigned to the residue layer diffusion (eqn (5) and (8)), while the ultrasound favoring the residue layer diffusion may also be used to suppress the dissolution of copper, compare Table 2.
Li (1st stage) | Li (2nd stage) | Mn | Co | Ni | Al | |||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|
k 3 × 10−4 (min−1) | R 2 | k 3 × 10−4 (min−1) | R 2 | k 3 × 10−4 (min−1) | R 2 | k 3 × 10−4 (min−1) | R 2 | k 3 × 10−4 (min−1) | R 2 | k 3 × 10−4 (min−1) | R 2 | |
Comment: Copper was excluded in the table because of its low fitting R2 value from the US leaching system, regardless of the control models described in the shrinking core model. Conditions: citric acid, 1.5 mol L−1; S/L, 25 g L−1, temperature, 323 K; stirring speed, 500 rpm. | ||||||||||||
Without US | 15.6 | 0.9916 | 3.52 | 0.9917 | 4.41 | 0.9980 | 3.81 | 0.9981 | 4.36 | 0.9981 | 0.79 | 0.9855 |
With US | 15.3 | 0.9814 | 7.28 | 0.9763 | 8.75 | 0.9944 | 9.30 | 0.9945 | 9.47 | 0.9956 | 0.99 | 0.9982 |
Table 4 shows the k3 values for Li, Mn, Co, Ni, and Al, respectively, as well as their related R2 values from the fitted slopes based on residue layer diffusion contribution (Fig. 5). The k3 values represent the release rate of the metal ions as the steepness of the linear slope in Fig. 5. It is shown that the Li extraction that required a two-stage fitting of the residue layer diffusion contribution in the 1st stage leaching demonstrated k3 values that were independent of the leaching method (oil bath or ultrasound), and ca. 2–4 times higher than other metal ions in the NMC. This was caused by some of the lithium originating from the electrolyte, making it easier to release to the acid solution. During the 2nd stage, the release of the lithium was dominantly from the solid phase and therefore occurred at a slower rate, with k3 value similar to the Mn, Co, and Ni metal ions.
Notably, although the k3 values for Mn, Co, and Ni were similar (Fig. 5 and Table 4), ultrasound always yielded higher release rates (regardless of the specific metal ion) suggesting that this release was caused by ultrasound impact on the solid phase. This observation was supported by the observation of almost identical k3 values for Li during the 1st stage leaching, compare 15.6 × 10−4 to 15.3 × 10−4 min−1 (without and with US).
Fig. 6 shows the SEM micrographs of the blackmass residue after the leaching had been carried out, with or without ultrasound, for the leaching times; 1.5 h, 3 h, and 6 h. The corresponding average values of extracted metal ions (Mn, Co and Ni) are included as marked ‘Ext. xx%’ in the upper righthand corners of the micrographs. The morphology of the NMC particles was initially made up of larger ca. 10 μm spherical secondary particles shown in Fig. 6a (white arrows), which consisted of smaller aggregated primary particles of irregular sizes (Fig. 6b). The primary particles ranged in size from ca. 0.5–3.0 μm in diameter, and the boundaries between these particles are distinguishable as fused surfaces (red arrows) in Fig. 6b. The surface of the primary particles was smooth although covered by significantly smaller particles with approximate size of 40–100 nm (Fig. 6b). With ultrasound treatment during the leaching, the secondary particles completely fell apart into the primary particles after 1.5 h and all the adhesion between primary particles were lost (Fig. S7c†). Without ultrasound, the secondary particles still held together for longer time than 3 h, i.e. the positioning of the primary particles remained almost the same in the larger secondary particles, see Fig. 6d, f and Fig. S7d.† Accordingly, a significant portion of the NMC surfaces were inaccessible to the acid during the early hours of the leaching when performed in absence of US. The extended particle size decrease in the US system consequently resulted in enhanced residue layer diffusion for the ultrasound system, see eqn (8). It is noteworthy that the smaller 40–100 nm particles disappeared completely in both cases. It is suggested that the disappearance of these particles followed a similar dissolution behavior as the aluminum, see Fig. 5, which showed the best fitting to a linear dissolution behavior from 0 to 1440 min (R2 = 0.9982, Table 4). Another observation was that the morphology of the particles exposed to ultrasound had developed more faceted edges, compare Fig. 6c and d. These facetted edges of the samples became most pronounced with longer extraction times (>6 h), while appearing only after 3 h with for the traditional extraction without ultrasound. Fig. 7a and b show high-resolution images of typical NMC primary particles with the layered grains after 3 h of extraction. The stacked sheet appearance was suggested to exist as a consequence of a more intense metal ion drainage in the surface of the particles, see arrow in Fig. 7b.
![]() | ||
Fig. 6 Scanning electron micrographs of the pristine LiBs powder before the ultrasound extraction/leaching of battery metals (a and b), and after (c–h) citric acid leaching. The top row shows ultrasound assisted leaching while bottom row shows the oil bath leaching for different times, where c and d is 1.5 h, e and f 3 h, and g and h 6 h. The average values (displayed as inserts in each micrograph) of the extracted metal ions (excluding Li) were derived from Fig. 3. |
![]() | ||
Fig. 7 Scanning electron micrographs of the primary particle of the layered NMC grains after 3 hours of extraction. Note the visible channel formations along different facets (a and b). |
It should be noted that lithium-proton ion exchange dominated in the earlier stages of extraction. Thus H+/H3O+ replaced Li+ sites, leaving primarily a Ni/Mn/CoO6 built layered structure (Fig. 7b) with less lithium present than represented by the crystal structure in Fig. 4b. This intercalation of H3O+ into the interlayers to some extent resulted in the shift of (003) to the lower 2 theta angle (Fig. 4a). It is suggested that although the morphological differences shown in Fig. 6 and 7 primarily related to initially breaking up the secondary particles, a more efficient metal ion drainage of the primary particles was always present when the ultrasound was used. This was evidenced by k3 values remaining constant for leaching times >3 h up to 24 h (1440 min) as shown in Fig. 3.
Table 5 summarize the leached percentages of all the metals at 1440 min. At 303 K only 71% Li, 53% Mn, 52% Co, and 54% Ni were released, while at 333 K ca. 20% more Li were released and about 40% more of the divalent Mn, Co, and Ni.
Leached percentage | Metal elements | |||||
---|---|---|---|---|---|---|
Li | Mn | Co | Ni | Al | Cu | |
Comment: The data for each ion was acquired as repeated more than twice. The deviation was maximum ±3% with an exception of Cu (maximum ±7%). | ||||||
303 K | 71% | 53% | 52% | 54% | 28% | 54% |
313 K | 77% | 72% | 66% | 70% | 34% | 73% |
323 K | 89% | 89% | 87% | 91% | 48% | 83% |
333 K | 92% | 92% | 93% | 93% | 58% | 77% |
343 K | 90% | 91% | 93% | 92% | 78% | 54% |
A further increase in temperature to 343 K resulted in more leached Al at 1440 min, whereas that of Li, Mn, Co, and Ni remained essentially the same and showed no additional leached amount of metal ions. The only exception from the general behavior that an increase in temperature favored more dissolved metal ions was for the copper (Cu). Here the leached percentage reached a maximum of 83% at 323 K and then gradually decreased to 54% when temperature was increased to 343 K. The explanation was argued related to the depletion of the dissolved O2 causing a supressed dissolution of Cu as described in section 3.2.
The activation energies (Ea) were calculated according to eqn (9) to investigate if there was a correlation between the minimum energy required to release of Li, Ni, Co, and Mn and the total amount of leached metal ions, as well as rate of the release of metal ions in the NMC grain structure. The activation energies were derived for the oil bath system since the Ea values were required to be calculated for uniform temperature conditions, which could not be established for the US system.15,16
![]() | (9) |
Fig. 9 shows that the following Ea values could be established for the residue layer diffusion of the different ions: 67.9 (Li 1st stage), 39.6 (Li 2nd stage), 55.7 (Mn), 56.6 (Co), and 57.2 (Ni) KJ mol−1, respectively. The activation energy for the 1st stage Li leaching was in fact significantly higher than other metal elements implying that a slower leaching rate should have existed although the opposite was observed. It should be highlighted here on the basis of the Li ion location in the NMC phase that the proton exchange process likely has a different reaction/diffusion path as compared to the other metal ions due to its location within the interlayer space of the NMC. This is in agreement with the structural data, wherein direct bonding holds Co/Mn/Ni in an oxide-rich coordination sphere, as shown in Fig. 4b. The 1st stage Li leaching was also reflecting contributions from both the electrolyte Li ion release process and the proton exchange process within the Ni/Co/MnO6 layers (see Fig. 4b and Fig. 7). It was calculated that approximately 50% of the Li leaching connected to the 1st stage leaching was related to the electrolyte releasing process, see further section 3.1. It is noteworthy that while the electrolyte release process may have had a small contribution to the derived Ea value for the 1st Li leaching part, the residue layer diffusion for the first Ea value may have required an even higher activation energy due to the intense electrostatic attractions in the interlayer space of the NMC structure. On the contrary, during the 2nd stage of Li leaching, Li showed a diffusion rate similar to the Mn, Co and Ni metals ions (Table S3†), although the estimated Ea was lower, i.e. 39.6 kJ mol−1. A suggested explanation would be that the layered structure of the NMC started to dissociate at this stage as it was converted into the spinel structure (see Fig. 4c), thereby facilitating the release of Li. At this stage, the activation energy based on the surface residue layer diffusion was on the order of Ni > Co > Mn > Li (2nd stage), which shows that the binding between Ni-citrate is the strongest while that of Li-citrate is the weakest among all the metal ions in NMC. A prevailing composition of Ni-citrate may thus exist in the residue layer interface surrounding the dissolving NMC grains.
The R2 values for the Arrhenius plots of Cu and Al, shown in Fig. S10 and Table S4,† are significantly lower than the other elements. Since the Arrhenius equation assumes that the activation energy is indenpendent on temperature,37 this observation suggests that the leaching mechanisms of Al and Cu were temperature dependent to a larger degree than for the NMC-metals. The deviation from the Arrhenius equation can especially be expected for Cu, since the temperature directly affects the concentration of dissolved oxygen in the solution.
Leached percentage | Metal elements | |||||
---|---|---|---|---|---|---|
Li | Ni | Co | Mn | Al | Cu | |
1.5 mol L−1 citric acid | 89% | 89% | 87% | 91% | 48% | 83% |
4.5 mol L−1 acetic acid | 83% | 75% | 72% | 73% | 53% | 86% |
The effect of the strength of the organic acid on the leaching behavior was investigated using a weaker organic acid, namely the monocarboxylic acetic acid (pKa = 4.76).25 4.5 mol L−1 acetic acid was used to normalize the concentration for the reduced number of carboxylic groups. Fig. 10 shows the leaching behavior when using 4.5 mol L−1 acetic acid and 1.5 mol L−1 citric acid at 323 K. During the first 150 min, the leaching behaviors of the acids were almost similar, thereafter the leached percentage of Li, and especially Mn, Co, and Ni showed to be significantly higher for the citric acid, although the higher number of carboxylic groups had been compensated for with 3 times higher concentration of acetic acid. Thus, without any additional reducing agent, 1.5 M citric acid (pH ≈ 1.14) showed better transition metal (Ni, Mn, Co) leaching performance compared to the 4.5 M acetic acid (pH ≈ 1.74). The leached percentage of Cu in 1.5 mol L−1 citric acid was, however, lower than in 4.5 mol L−1 acetic acid during the entire course of the reaction. This may be explained as a result from that citric acid has been reported as an efficient chelating agent to suppress the Cu dissolution in acid.38
The leaching behavior with/without ultrasound using acetic acid was also investigated. As shown in Fig. S11,† the leached percentage of the targeted metal ions were significantly improved with ultrasound also for the acetic acid, reaching ca. 96% ± 2% (Table 7). The result further highlighted the impact of ultrasound in improving the extraction of metals from the blackmass, irrespectively of the acid species. This might be due to the formation of hydroxyl radicals in the US leaching system, possibly allowing continuous formation of hydrogen peroxides.39 Fig. S12† shows the overlapping nature of the citric and the acetic acid during the entire leaching cycles, suggesting that the primary difference between the acids was their leaching rate (speed) under the influence of the ultrasound.
Leached percentage | Metal elements | |||||
---|---|---|---|---|---|---|
Li | Ni | Co | Mn | Al | Cu | |
Oil bath, 323 K | 83% | 75% | 72% | 73% | 53% | 86% |
US, 323 K | 94% | 97% | 97% | 95% | 60% | 73% |
The Ea values for leaching of Li (1st stage), Li (2nd stage), Mn, Co, and Ni were 70.1, 41.5, 44.4, 45.1, and 44.1 kJ mol−1, respectively (see Fig. S13, S14, S15, and Tables S5, S6† for their derivation). The activation energies for the citric and the acetic acid were thus similar for the Li (1st stage) and Li (2nd stage), while for Mn, Co, and Ni the Ea values were more than 10 kJ mol−1 lower for the acetic acid as compared to the citric acid. The reason to the lower activation energy for the divalent metal ions when using acetic acid is presently unknown. It was speculated being a consequence of a less restricted diffusion through the residue layers on the surface of the NMC grains. This was based on the presumption that the complex formation of the extracted divalent metal ions (Mn, Co, and Ni) and the citric acid likely formed stronger complexes by directing 2 oxygen rich carboxylic groups of the acid towards the positively charged metal ions, while leaving one carboxyl group facing outwards towards the bulk acid solution. This configuration stands in contrast to the nonpolar methyl group facing outwards in the case of the acetic acid when its carboxyl group in turned towards the metal ions.
The viscosities of the aqueous solutions of 4.5 mol L−1 acetic acid and 1.5 mol L−1 citric acid, before and after the 24 h leaching process, were taken as an indicative measure of this possible complex binding and coordination of the divalent metal ions (Table 8). The viscosity of the aqueous solution of the 1.5 mol L−1 citric acid was significantly higher than that of the 4.5 mol L−1 acetic acid. The differences in viscosity before and after (Δ) leaching of the 1.5 mol L−1 citric acid solution was: Δ = 0.33 mm2 s−1, which in turn also was higher than that of 4.5 mol L−1 acetic acid solution: Δ = 0.17 mm2 s−1. The higher viscosity of the citric acid system was suggested to reflect the stronger metal–organic ion complex binding. This would be in agreement with the higher activation energy required for the diffusion of transition metal ions through the residue layer in the 1.5 mol L−1 citric acid.
Acid species | 1.5 mol L−1 citric acid | 4.5 mol L−1 acetic acid | ||
---|---|---|---|---|
Before | After | Before | After | |
Comment: The data for each viscosity was acquired as repeated five times measurements at 298 K. The deviation was maximum ±0.03%. | ||||
Viscosity (mm2 s−1) | 1.67 | 2.00 | 1.33 | 1.50 |
Fig. S18† shows the curves in Fig. S16† as fitted to the residue layer diffusion contribution of the shrinking core model. The leaching processes of all NMC metal elements could be divided into two stages in the presence of H2O2, with the corresponding k3 values being summarized in Table S7.† When 1.5 vol% of H2O2 was added, the k3 values of the transition metal elements (Ni, Mn, and Co) were ca. 2–3 times higher during the 1st stage (initial 150 min) than when no H2O2 was added. The increased rate constants were synonymous with a faster release of the structurally bonded metal ions. The increase of the k3 value in the 1st stage was attributed to:
• The H2O2 reducing Co and Mn from high valence values to low valence ones, which improved the ability of acid to solvate the metal ions,
• The decomposition of the H2O2 resulting in the formation of bubbles, which improves the mixing of the solution, contributing to a facilitated mass transfer and a higher leaching rate.
After 150 min (the 2nd stage), the k3 values decreased and became similar, regardless of H2O2 concentration. It was suggested that H2O2 mainly worked during the beginning of the leaching process due to its depletion. Overall, it was concluded that the addition of H2O2 enhanced the leaching rate of transition metal elements, whereas the total amount of metal ions decreased at excessively high H2O2 concentrations due to the formation of intermediate hydroxide phases (e.g., Cu(OH)2).
Footnote |
† Electronic supplementary information (ESI) available. See DOI: 10.1039/d1gc02693c |
This journal is © The Royal Society of Chemistry 2021 |