A green recycling process for spent lithium-ion batteries with extremely low chemical consumption

Highlights

• •A novel intermediate phase inserted with H⁺ was discovered and characterized.
• •A green recycling process for spent LIBs based on the novel intermediate phase.
• •Revealing the reaction mechanisms between different species of transition metals.
• •Achieving high utilization efficiency of leaching reagent.
• •High application range for different cathodes of spent LIBs.

Abstract

Selective recovery of lithium from spent cathodes is an attractive, green and efficient recycling method for spent lithium-ion batteries (LIBs). However, current technologies face numerous challenges including high reagent consumption, limited versatility and significant secondary pollution. In this study, we found that an intermediate phase formed during leaching exhibited high physiochemical stability, which hindered leaching of Li⁺ and led to increased reagent and energy consumption. A simple mechanical activation strategy was used to change the intermediate phase of the leaching process by activating the spent cathode without grinding additives. The energy input of the mechanical force resulted in defects in the activated waste, leading to a novel activated intermediate phases. Through experimental characterization and DFT analyses, this novel intermediate phase containing H⁺ was found to have a large specific surface area, large internal stress and low reaction energy, which enabled the leaching of the Li⁺. As a result, the method showed a high utilization efficiency of H⁺ in recycling most spent cathodes. In the end, we established a closed-loop recovery route for spent LIBs that exhibited exceptional efficiency for H⁺ utilization (> 97%), obviated the need for auxiliary reagents, and substantially reduced secondary pollutant generation.

Graphical abstract

1. Introduction

Recovery of valuable components from spent lithium-ion batteries (LIBs) is feasible, but there are still challenges in establishing green and rapid recycling processes [1], [2], [3]. Among these, recycling of Li from spent LIBs is still unsatisfactory due to its chemical reactivity and limitations of existing techniques. In traditional recycling processes, Li⁺ extraction usually occurs at the end of the process, leading to a leaching solution with a low concentration of Li⁺ but high concentrations of Na⁺ or other impurities [4], [5], [6]. Therefore, concentration and separation, which consume considerable energy, are needed to increase the extraction efficiency and purity of the Li₂CO₃ [7], [8], [9]. Moreover, high lithium losses (approximately 10- 30%) occurs during this process, such as during solvent extraction [10,11], which also reduces the efficiency of Li⁺ recycling.

In recent years, many technologies have been investigated to achieve selective/prior leaching of lithium with increased efficiencies, but there are still challenges in spite of the lower reaction temperatures, fewer additives and better applicability [12], [13], [14]. Selective leaching of Li via phase transitions is an important research direction. In these technologies, additives are used to transform the Li containing phase from waste LiCoO₂ to Li₂O, Li₂SO₄, LiCl or other water-soluble lithium salts through roasting or mechanical chemistry [15], [16], [17]. For example, in the carbothermic reduction process, the lithium contained in spent cathodes is transformed to Li₂CO₃, and the transition metals are transformed to metals or metal oxides [18], [19], [20]. In acidic or salt roasting, lithium is also transformed to Li₂SO₄ or LiCl [21], [22], [23]. After roasting, the lithium is extracted with water. The roasting process usually requires a high roasting temperature, and releases greenhouse gases and pollutants, such as NO₂ and CO [15,24]. Because of the transition metal oxides are reduced to metals, the extraction processes usually release H₂, which increases the operational risk of the technology. Mechanical chemistry enables the phase transitions under a mild conditions and at low temperatures [25]. NaCl and SiO₂ can be used as co-grinding reagents to achieve selective extraction of Li [26]. Dry ice, PVC and pure metals can also be used as co-grinding reagents to transform the phases of spent cathode materials [25,27,28]. However, complex additives and waste water limit application [26]. Direct preferential extraction of lithium from waste cathodes has been investigated, such as oxidative leaching, but mass production is difficult, especially for common high-nickel cathodes [29], [30], [31]. Electrochemical methods have also been used to achieve selective leaching of Li⁺. For example, when cathode materials are used as anodes and NaCl solutions are used as electrolytes in electrolytic cells, Li⁺ can be selectively extracted from spent cathodes at 2.0 V (DC voltage) [32]. Hydrothermal treatment results in excellent lithium extraction results. In this process, only H⁺ is used to extract the lithium from waste at low temperatures with few additives and no by-products [33]. However, this technology has poor applicability for high-cobalt cathodes.

An inherent concept, which has been identified in many previous investigations, is that mechanical activation enables total leaching of valuable metal ions [34,35]. Moreover, leaching of transition metals is difficult to be limit in acidic systems. However, selective extraction of lithium was achieved in acidic systems with mechanical assistance in this study. We found that the insert of H⁺ in the intermediate phase of the leaching process after mechanical activation provided a two-stage leaching process leading to selective leaching of Li⁺ from spent cathodes. Moreover, it was also proved to be widely applied for all typical cathode materials of spent lithium-ion batteries except for LiFePO₄ at low extraction temperature, and maintained high extraction efficiencies. Because of its broad applicability, this approach also reduce the difficulty of front-end classification and preprocessing.

2. Materials and methods

2.1. Materials

The LiCoO₂ cathode scraps were supplied by a spent LIB recycling company (Huayou Cobalt Co., Ltd, China) in China. Before extraction, the spent cathodes were calcined at 500 °C for 6 h to decompose the organic binder, and then the cathode powders were separated from the Al foil. LiNi_0.33Co_0.33Mn_0.33O₂ (NCM111), LiNi_0.5Co_0.2Mn_0.3O₂ (NCM523), LiNi_0.6Co_0.2Mn_0.2O₂ (NCM622), LiNi_0.8Co_0.1Mn_0.1O₂ (NCM811), and LiMn₂O₄ (LMO) were unused cathode materials and were purchased from MTI Corporation. Some of these materials (powders) were described in our previous article [33]. The sulfuric acid solution was obtained from concentrated sulfuric acid (95∼98%; Beijing Chemical Works). The aqua regia solution was prepared by mixing concentrated nitric acid and concentrated hydrochloric acid at a volume ratio of 1:3. The Li standard solution used for characterization was purchased from Central Iron & Steel Research. The standard solutions of Ni, Co, Mn, Al, Fe and Cu used for characterization were purchased from Guobiao (Beijing) Testing & Certification Co., Ltd. K₂Cr₂O₇ was a working reference reagent purchased from Sinopharm Chemical Reagent Co., Ltd (Tianjing). Na₂S₂O₃, KI, and Na₂CO₃ were of analytical purity and purchased from Xilong Science Co., Ltd. NaOH was of analytical purity and was purchased from Sinopharm Chemical Reagent Co., Ltd.

2.2. Extraction methods

The waste was first activated in a ball crusher (Pulverisette7, FRITSCH, Germany). After that, the waste powder was transferred to a hydrothermal reactor (100 ml, PTFE), which was heated in a homogeneous reactor manufactured by Yantai Keli Chemical Equipment Co., Ltd. The heating time of the equipment was fixed at 60 min. Then, separation of the extraction liquid and residues was conducted with a vacuum suction filter manufactured by Zhengzhou Great Wall Scientific Industrial and Trade Co., Ltd. Unless otherwise specified, the solid-liquid ratio of the hydrothermal leaching process was 100 g/L, the leaching time was 4 h, the molar ratio of H⁺ to Li⁺ was 1:1; rotation was carried out at 600 rpm for 20 min, then paused for 10 min, and this was continued for 6 cycles with ball/waste mass ratios of 10:1. Other leaching experiments, which were conducted under atmospheric pressure, involved heating the samples in a water bath (Collector type constant temperature heating magnetic stirrer, DF-101 K, Beijing Century Yuhua Instrument Co., Ltd). The calculations of the leaching efficiencies, leaching selectivity and utilization efficiencies of H⁺ are also shown in Supporting Information.

2.3. Syntheses of anode materials

Leaching residues with few impurities are synthesized as anode materials for the LIBs. The leaching residues prepared at different rotation speeds were marked as L-“rotation speed”, in L-800. The leaching residues could be directly used as anode materials, but were also calcined to obtain better electrochemical performance. The anode materials produced at different calcination temperatures, such as R-800–500 (800 rpm, 500 °C), were denoted as R-“rotation speed”-“calcination temperature”. Calcination was conducted in a muffle furnace under air with heating from room temperature to the target temperature at a heating rate of 5 °C min⁻¹.

2.4. Characterization methods

The contents of the materials used in this study were analysed via ICP (ICP-OES 6300, ICAP 6300Radial, Thermo Scientific). The materials were first dissolved in aqua regia, and then the contents of metals in the solution and the leachate were quantified by ICP-OES. The structures of the materials or residues were characterized by X-ray diffraction (XRD, X'pert PRO, PANalytical) with a Cu Kα radiation source. The data were collected over the 2θ range 5–90°. Highscore software was used to fit the XRD patterns and obtain relative data. The morphologies of the samples were characterized with a scanning electron microscope (SEM, JEOL JSM-7610F) equipped with an energy dispersive spectroscope (EDS, OXFORD INSTRUMENTS). TEM equipped with SAED was carried out with a JEOL JEM-F200 field-emission electron microscope. The Raman spectra were determined with a Raman spectrometer (inVia, Renishaw). The oxygen vacancy data were obtained with electron paramagnetic resonance spectrometry (EPR, MS-5000 Freiberg Instruments). The valance states were determined with X-ray photoelectron spectroscopy (XPS, ESCALAB 250Xi). The particle sizes were analysed with a laser particle size system (Mastersizer 2000). The thermal decomposition processes of the materials were determined with a high temperature thermogravimetric analyser (TG-DSC-MS, LABSYS EVOT). The valence state of the transition metals in the waste or residues were also determined with the iodimetry method, which is described in detail in the supporting information.

2.5. Open circuit potential-time (OCPT) tests

First, 0.04 g of sample was mixed with 100 µL of a Nafion solution, 400 µL of ethanol and 500 µL of deionized water. After dispersing the sample evenly through ultrasound, the mixed solution was dipped on a glassy carbon electrode (GCE) and placed under an infrared lamp. Then, the OCP of the GCE with samples was monitored by the OCPT method with a CH760E workstation using a SCE reference electrode and a 1 cm * 1 cm Pt electrode as the counter electrode.

2.6. Computational method

The Vienna Ab Initio Package (VASP) was used to perform all the density functional theory (DFT) calculations within the generalized gradient approximation (GGA) using the PBE formulation [36], [37], [38]. The projected augmented wave (PAW) potentials was used to describe the ionic cores and take valence electrons into account using a plane wave basis set with a kinetic energy cutoff of 400 eV [39,40]. The Gaussian smearing method was used to describe the partial occupancies of the Kohn−Sham orbitals with a width of 0.05 eV. The electronic energy was regarded as self-consistent when the change of energy was lower than 10−5 eV. A geometry optimization was considered convergent when the change of force was lower than 0.02 eV/Å. Grimme's DFT-D3 approach was used to describe the dispersion interactions [41].

2.7. Electrode preparation and cell testing

The electrode slurry was made by mixing the active materials, Super P and PVDF with a weight ratio of 7:2:1, and dispersing them in N-methyl-2-pyrrolidone solution (NMP). Then, the solution was coated on the Cu foil with a doctor-blade. It was dried in an oven at 60 °C for 2 h and then dried in a vacuum oven at 110 °C overnight. Then, the foil was cut into Φ14 disks for coin-cell assembly. A polypropylene membrane (Φ19, Celgard 2400) was used as the separator. The electrolyte contained 1 M LiPF₆ dissolved in a 1:1:1 EC:DEC:MEC mixed solvent. Li metal was used as the counter electrode. The CV curves were obtained with a CH760E system over the voltage window 0.2 to 3.0 V versus Li⁺/Li with a 0.0001 mV/s scanning rate. The other tests of electrochemical performance were conducted with a CT2001A tester system (Land, China).

3. Results and discussion

3.1. Metal extraction from cathode scraps

Based on Reaction (1), the leaching efficiencies of Li⁺ and Co²⁺ maintained a ratio of 2:1³³. The introduction of a reductant facilitated leaching of Li⁺, but also increased the leaching efficiency of Co²⁺ [42]. Therefore, selective extraction of Li⁺ was so difficult that even at high reaction temperatures (80 ºC or 160 ºC), Li⁺ was not effectively extracted, as shown in Fig. 1a. Only 61.04% of the Li⁺ was leached from the waste with a 26.19% leaching selectivity. As shown in Fig. 1b, leaching of Li⁺ and Co²⁺ was rapid and synchronous, suggesting the difficulty limiting the leaching of Co²⁺. Therefore, the difficulty of selective extraction turns into how to make the Co²⁺ ions back to the solid. It is a typical comproportionation reaction between Co²⁺ and Li_1–2xCo_1-xO_2–2x (Reactions 1–3) or hydrolysis reaction of Co²⁺. Thus, a two-stage reaction process for selective extraction of Li⁺ is proposed, as shown in Fig. 1c. The results shown in Fig. S1 indicate that the intermediate products from stage 1 had different activities with Co²⁺. The electrode potential of the intermediate was monitored by OCPT. The results of OCPT indicated the oxidizabilities of the electrode materials [43,44]. The OCPT results revealed that more lithium was extracted from LiCoO₂ leading to a higher oxidizability of the intermediate, which resulted in higher reactivity for leaching of Li⁺ in stage 2 (Fig. S2). Therefore, the key was to increase the reactivity of the intermediates obtained from stage 1 [45].

$$\text{LiCo}{\mathrm{O}}_2+4\mathrm{x}{\mathrm{H}}^{+}=2\mathrm{x}\mathrm{L}{\mathrm{i}}^{+}+\mathrm{x}\mathrm{C}{\mathrm{o}}^{2+}+\mathrm{L}{\mathrm{i}}_{1-2\mathrm{x}}\mathrm{C}{\mathrm{o}}_{1-\mathrm{x}}{\mathrm{O}}_{2-2\mathrm{x}}+2\mathrm{x}{\mathrm{H}}_2\mathrm{O}$$ (1)

$$\begin{array}{ccc}& & \mathrm{L}{\mathrm{i}}_{1-2\mathrm{x}}\mathrm{C}{\mathrm{o}}_{1-\mathrm{x}}{\mathrm{O}}_{2-2\mathrm{x}}+\mathrm{x}\mathrm{C}{\mathrm{o}}^{2+}+2\mathrm{x}{\mathrm{H}}_2\mathrm{O}\hfill \\ & & =(1-2x)\mathrm{L}{\mathrm{i}}^{+}+(4\mathrm{x}-1){\mathrm{H}}^{+}+\text{HCo}{\mathrm{O}}_2\hfill \end{array}$$ (2)

$$\begin{array}{ccc}& & \mathrm{L}{\mathrm{i}}_{1-2\mathrm{x}}\mathrm{C}{\mathrm{o}}_{1-\mathrm{x}}{\mathrm{O}}_{2-2\mathrm{x}}+(1-2\mathrm{x})/2\mathrm{C}{\mathrm{o}}^{2+}+(3/4-\mathrm{x}){\mathrm{H}}_2\mathrm{O}\hfill \\ & & =(1-2x)\mathrm{L}{\mathrm{i}}^{+}+(3/2-2\mathrm{x})\text{HCo}{\mathrm{O}}_2+\left(\frac{\mathrm{x}}{2}-\frac{1}{8}\right){\mathrm{O}}_2\uparrow \hfill \end{array}$$ (3)

Fig. 1.

(a) Effect of temperature on leaching of valuable metals from raw waste materials with 0.5 M H₂SO₄; (b) effect of reaction time on leaching of valuable metals from raw waste materials and activated waste with 1.5 M H₂SO₄ at 60 °C; (c) and (d) schemes for selective extraction of Li⁺; effects of (e) rotation speed, (f) rotation time and (g) ball/waste mass ratio on selective extraction of Li⁺.

We utilized mechanical activation to increase the energy input and change the phases of the intermediate products, which facilitated the reactions occurring in stage 2 (Fig. 1d). Although many additives were commonly utilized in previous mechanical-chemistry activation processes, additives would be unnecessary if the two-stage reaction could be enhanced by changing the reactivities of the intermediate products. Figs. 1e-g show effects of mechanical activation on the leaching efficiencies of metals from the cathode scrap. The damage to the waste particle sizes and crystal structures had an excellent effect on the subsequent extraction of Li⁺. The leaching efficiencies of Li were positively correlated with the rotation speed, rotation time and ball/waste mass ratio (B/M), as shown in Figs. 1e-g. The extraction efficiency and selectivity of Li⁺ increased from 57% and 26% to 95% and 97%, respectively. The utilization efficiency of H⁺ also increased from 60% to 99%. This meant that the leaching reagent had a high utilization efficiency and led to fewer secondary pollutants in the later process. The effect of the leaching time on the extraction of Li⁺ is shown in Fig. S3. Some Co²⁺ was leached from the solid waste into the solution in the first stage but returned to the solid phase in the second stage. The leaching efficiency of Co²⁺ decreased from 12% to 3% during the leaching process. Moreover, it also decreased with increasing energy input into the activation process. These results indicate that higher input energy led to higher reaction activities of the intermediate products with Co²⁺ in stage 2 (Fig. 1e). Additionally, higher reaction temperatures led to more selective extractions (Fig. S4) and activation program had little effect on the leaching of Li⁺ (Fig. S5). All of the results indicate that higher input energy led to better extraction effects. Therefore, the effects of the input energy on the phase, morphology and lattice structure were analysed to reveal the mechanisms for activation and selective extraction.

3.2. Mechanisms for the activation process

After mechanical activation, the changes in the morphology, crystal structure and defects of the cathode scrap were obvious. The raw waste particles were ellipsoids (Fig. 2a). They were first crushed into smaller ellipsoids or fragments at low rotational speeds, and then reunited at high rotation speeds to form large spheres with very rough surfaces. Although there were still large particles that were not damaged, overall, the particle sizes were greatly reduced (Figs. 2b-e). Fig. 2f was obtained at a larger magnification to show the local morphologies of the particles in Fig. 2e. Moreover, the large particle under high rotation speed (800 rpm) were agglomerated by many small particles. Decreases in the particle sizes may increase the leaching rates. Figs. 2g-l show TEM images of the particles of the raw waste and activated waste. Obvious cracks were seen in the particles after activation. The SAED image in Fig. 2h shows that some particles in the raw waste were large and had single dominant crystal planes, so the SAED images were similar to those of monocrystalline particles. However, after activation, the SAED images in Fig. 2k showed diffraction rings, which indicated smaller particles and different crystal planes. The HRTEM images in Figs. 2i and 2f show changes in the diffraction patterns of the raw waste and activated waste. For the original particles, a continuous diffraction pattern was displayed. Although stacking of the particles resulted in intersecting lattice fringes, they were continuous, parallel and not bent. Some small areas had no obvious lattice fringes, which was attributed to residual organics or other impurities. However, in the activated particles, more amorphous areas and different planes were found in small areas. The diffraction fringes were also not continuous, as shown by the red line in Fig. 2l. The yellow lines show bending of the lattice fringe, which represented crystal defects. Fig. S6 shows crystal defects in the activated waste materials. The defects increased the internal energy of the crystal, which increased the reaction activity of the waste, and facilitated the reaction in the second stage.

Fig. 2.

SEM images of (a) the raw waste and waste activated at (b) 200 rpm, (c) 400 rpm, (d) 600 rpm, and (e) 800 rpm; (f) local enlarged SEM image of the waste activated at 800 rpm; (g) TEM images, (h) relevant SAED images, and (i) HRTEM images of the waste; (j) TEM images, (k) relevant SAED images, and (l) HRTEM images of the activated waste from the 600 rpm system.

The particle sizes produced at different rotation speeds and reaction times were analysed with a laser granularity analyser (Fig. 3a,b and Fig. S7a-b). The results indicated that rotation speed of 200 rpm reduced the d(0.8) of the waste from 23.538 to 7.613 µm. However, overmilling lead to agglomeration of particles, which increased the particle sizes. Moreover, only for the raw waste did the cumulative distribution of particle sizes reach 100% (Fig. 3b). This suggested that the sizes of the particles were rarely smaller than 100 nm in the raw waste, but the amount increased to approximately 4% after activation. Similar effects on the waste were found with different rotation time. The particle sizes of the waste first increased and then decreased, as shown in Figs. S7a-b. Noticeably, the trend for the leaching results differed from that of the particle sizes, which means that the particle sizes were not the main influencing factor. The surface area increased from 0.98 m²/g for the raw waste to 44.541 m²/g for the waste activated at 800 rpm (Fig. 3c). This meant that agglomeration of the particles did not determine the increase in surface area. The N₂ adsorption/desorption curves suggested that few pores were created during the activation process. The hysteresis line was caused by stacking of flaky particles, which resulted in some pores [46]. Overall, the leaching results were consistent with the changes in the surface areas.

Fig. 3.

(a) Particle size distributions for the waste treated at different rotation speeds; (b) Cumulative distributions of the particle size distribution; (c) N₂ adsorption-desorption curves for the waste treated at different rotation speeds; (d) XRD patterns for the products from the activation processes with different rotation speeds (the characterization angles ranged from 5 to 90° in 9.01 min); (e) UDM plot for different activated wastes; (f) relationship between the micro-strain (Ɛ) and rotation speed; (g) OCP results for the stage 1 intermediate products of the waste obtained after activation at different rotational speeds; (h) TG-MS results for the intermediate product from 800 rpm; (i) DFT results for ΔH of the different reactions.

A larger surface area led to a higher leaching rate, but it is difficult to change the thermodynamics. Therefore, more characterization methods were used to analyse the effects of the mechanical activation process. Fig. 3d and Fig. S8 show the XRD patterns of the waste materials prepared with different rotational speeds and times, respectively. Although the structure of the waste was continuously destroyed in the activation process, the damage still had an order. The I(003)/I(104) ratio showed that the (104) peak was destroyed first in the process, suggesting that waste particles was torn from the corresponding face. After that, the particles with the (003) plane are continuously destroyed. The XRD patterns were fit with “Profile Fit” in Highscore software to analyse the parameters. These results suggested that the model described the intrinsic strain of the waste from the activation process well (Fig. 3e). The intrinsic strain of the waste, which represented crystal defects, increased with increasing rotational speed, which led to increased reaction activity (Fig. 3f). A greater energy input led to greater damage and more defects in the structure. In the energy calculation, the energy involved was proportional to the third power of the rotation speed. Therefore, the rotation speed had a dramatic effect on the structure of the waste. Fig. S9 shows the effects of the energy input. Extension of the activation time at 200 rpm from 2 h to 6 h also led to a great decrease in the intensity of the (003) peak. This meant that the damage to the structure depended on the energy input but not only on the rotation speed.

Therefore, the OCPT values of the different intermediate products were shown in Fig. 3g. However, the oxidizability of the intermediate products after activation were lower than those of intermediate products obtained from raw cathode scrap (Fig. 3g). This was not consistent with the results in Fig. S2. The TG-MS results showed that more H⁺ was inserted into the intermediate products with increasing input energy. With the insert of H⁺, increases in the valence states of transition metal ions caused by Li⁺ extraction were suppressed, leading to lower oxidizability of the intermediate products. Fig. 3h and Fig. S10a-d show that the content of H₂O in the decomposition products from the intermediate products increased with increasing rotation speed. The mass losses in the TG curves also indicated that more H⁺ inserting in the intermediate products with increasing rotation speed. Due to the difficulty in quantifying the H content, we chose two possible intermediate products for a thermodynamic comparison of reactions in stage 2 (Reactions (4) and (5)). The difference between them was that more H⁺ was inserted into the intermediate products instead of Li⁺. The ΔH values of the reactions were calculated from the H values of the reactants and products, which were obtained via DFT calculations. The results indicated that with the insert of H⁺, ΔH decreased by 6.97 eV, which meant that the reaction in stage 2 was easier to conduct.

$$10{\mathrm{H}}_2\mathrm{O}+10\mathrm{L}{\mathrm{i}}_{0.3}{\mathrm{H}}_{0.2}\text{Co}{\mathrm{O}}_2+5\text{CoS}{\mathrm{O}}_4\to 1.5\mathrm{L}{\mathrm{i}}_2\mathrm{S}{\mathrm{O}}_4+15\text{CoOOH}+3.5{\mathrm{H}}_2\mathrm{S}{\mathrm{O}}_4$$ (4)

$$10{\mathrm{H}}_2\mathrm{O}+10\mathrm{L}{\mathrm{i}}_{0.5}\text{Co}{\mathrm{O}}_2+5\text{CoS}{\mathrm{O}}_4\to 2.5\mathrm{L}{\mathrm{i}}_2\mathrm{S}{\mathrm{O}}_4+15\text{CoOOH}+2.5{\mathrm{H}}_2\mathrm{S}{\mathrm{O}}_4$$ (5)

Furthermore, the Raman spectra for the waste from different rotation speeds indicated no obvious shifts, shown in Fig. S11a, which meant that the crystal of waste is still in short-range order. The three modes were assigned to two phases; the peaks at 482 cm⁻¹ and 593 cm⁻¹ were from LiCoO₂ and the peak at 693 cm⁻¹ was from Co₃O₄ [47,48]. The appearance of Co₃O₄ could have been caused by reduction during the pretreatment process. A light shift in the peak at 693 cm⁻¹ was found after activation, which suggested distortion of the spinel Co₃O₄ lattice during the activation process. The Raman bands at 593 cm⁻¹ and 693 cm⁻¹ were blueshifted and redshifted after H⁺ leaching, respectively (Figs. S11a-b). The strong shifts of the main Raman bands at 593 cm⁻¹ and 693 cm⁻¹ toward higher wavenumbers for the cathode scrap were attributed to irreversible cobalt migration into the interlayer lithium sites, which led to spinel-type cation ordering [49]. In contrast, the A_1g mode of LiCoO₂ show a blue shift after activated by mechanical force, as shown in Fig. 4a. This may have occurred because the cathode scrap was activated, which resulted in the formation of a new CoOOH layered structure during the acid leaching process. The A1g mode of CoOOH was at 572 cm⁻¹ [50].

Fig. 4.

(a) Raman spectra for the residues of the first stage; (b) valence state for cobalt in the spent LCO, activated spent LCO, and leaching residues (HCoO₂); (c) EPR results for the waste formed with different activation time; XPS results for the (d) C 1s, (e) O 1s, and (f) Co 2p XPS data for the raw waste; (g) C 1s, (h) O 1s, and (i) Co 2p XPS data for the activated waste.

The effect of the valence state on the activation process was also investigated. The iodometry results indicated that reduction mainly occurred on the surface of the waste, and had little effect on the valence state of Co³⁺ in the bulk spent LCO (Fig. 4b). The valence of Co in the raw waste and activated waste was nearly 2.99, but was 3.09 for that in the leaching residue. The lack of an obvious change in the valence state of cobalt after activation excluded its influence on the extraction results. During the extraction process, the valence of Co first increased and then decreased, which was consistent with the dissolution of cobalt ions (Fig. S12). This also proved the route shown in Fig. 1c. The peak at g=2.16 was attributed to the single electron in Co²⁺ (Fig. 4c). Co²⁺ was oxidized to Co³⁺ during the activation process, leading to a decrease in the intensity of the peak. Furthermore, surface analyses of the waste and activated waste were also conducted with XPS (Figs. 4d-i). Figs. 4d and 4g show the C 1s spectra of the waste and activated waste, respectively. An obvious peak for carbonate C was found on the surface of the activated material. Similar results have also been reported in previous investigations [34]. A portion of the organic C or super P was oxidized by O from the spent LCO in the activation process. Accordingly, the peak for C=O in the activated materials decreased more than that in the raw waste (Figs. 4e and h). The XPS results for cobalt showed little change in the activation process (Fig. 4f and i).

The phase of the residue from the second stage was identified with different characterization methods to complete the analysis of the process. The XRD patterns (Fig. 5a) for the activated waste and leaching residues showed that the phase transformed from LiCoO₂ to HCoO₂ (CoO(OH)). Although Co(OH)₂ has an XRD pattern similar to that of HCoO₂, the residues did not include Co(OH)₂, as indicated by TG-DSC-MS (Fig. 5b), which also showed that O₂ was released during the decomposition process. The thermal decomposition processes of these two materials were different. For the Co³⁺ existed in the HCoO₂ (CoO(OH)), during decomposition, O₂ was released into the environment under an argon atmosphere or other inert atmospheres (Reaction 6). For Co(OH)₂, no O₂ should be detected (Reaction 7). The TG results showed that the tested residues began to decompose rapidly at approximately 200 °C, and ultimately lost approximately 11.7% of their weight. The theoretical loss was 12.69% for Reaction 6 and 19.37% for Reaction 7. Obviously, the actual loss was close to that shown in Reaction 6. Moreover, MS revealed the release of O₂ and H₂O at similar times when the residues began to decompose (Fig. 5b). As shown in Fig. 5c, except for the small particles, many flocculent products were found wrapped around the particles. These flocculent products were consistent with the microscopic characteristics of hydroxide.

$$6\text{HCo}{\mathrm{O}}_2\genfrac{}{}{0pt}{}{\Delta }{=}2\mathrm{C}{\mathrm{o}}_3{\mathrm{O}}_4+1/2{\mathrm{O}}_2+3{\mathrm{H}}_2\mathrm{O}$$ (6)

$$\text{Co}{\left(\text{OH}\right)}_2\genfrac{}{}{0pt}{}{\Delta }{=}\text{CoO}+{\mathrm{H}}_2\mathrm{O}$$ (7)

Fig. 5.

(a) XRD patterns for the activated waste and leaching residue; (b) TG-MS results for the leaching residue (HCoO₂); (c) SEM image of the HCoO₂; (d) schematic diagram of reaction mechanism; (e) leaching results for the raw materials and activated waste including LCO, NCM111, NCM523, NCM622, NCM811 and LMO (data for activated spent LCO and LMO were obtained at 800 rpm; LCO is from the waste and other cathodes are commercial materials).

The route for selective extraction is clear, as shown in Fig. 1c. The process can be summarized as Reaction 8. The transitions among different Co species were achieved in one leaching process with two different reaction stages. In this process, selective leaching of Li⁺ with only H⁺ was achieved with an ion-exchange-like reaction.

$$\text{LiCo}{\mathrm{O}}_2+{\mathrm{H}}^{+}=\mathrm{L}{\mathrm{i}}^{+}+\text{HCo}{\mathrm{O}}_2$$ (8)

The applicability of the method was also investigated. Leaching of NCM111, NCM523, NCM622, NCM811 and LMO were tested with this method. Fig. 5 shows that after activation the hydrothermal treatment was applied to both high cobalt materials, and high nickel materials. An excellent increase on the selective leaching efficiencies of Li⁺ for different cathode materials have been obtained. Improved adaptability will provide technologies for extracting valuable metals from various waste LIBs systems in the future. Fig. S13a-d show the XRD patterns for the activated cathode materials and relative leaching residues. The typical peaks for the different cathodes were seen for the activated cathode materials, indicating incomplete damage to the crystal structure. In addition to LMO, the other residues showed similar peaks, but was difficult to assign the phases, which should be investigated in the future. A preliminary analysis suggested that the residues were mixtures of the MO₂ and M(OH)₂, which meant that the process involved re-oxidation of different metal species. The mechanical activation process facilitated the relevant reaction, similar to the effect on the waste LCO system.

Furthermore, the residues from the selective leaching process also had high contents of other valuable metals that required recycling. Considering the phases, the residues were recycled as anode materials for LIBs. For poorly crystalline residues from the hydrothermal treatment, calcination was used to enhance the electrochemical performance. Fig. S14 compares the XRD patterns of the anode materials synthesized under different conditions. The results clearly showed that a high temperature led to a peak with a high intensity, indicating higher crystallinity. Moreover, according to TG-MS, the material decomposed into Co₃O₄ at high temperatures. For this reason, only R-800–120 still retained the CoOOH phase. The phases of the other anode materials were Co₃O₄.

The CV profiles in Figs. S15a-d demonstrated the stable redox behaviours of the various anode materials. One redox couple was observed for the different materials, with peaks at 0.75–1.0 V (vs. Li/Li⁺) for the cathodic process and 2.0–2.3 V for the anodic process. The cathodic peak in the first cycle occurred at 0.5–0.75 V for the different anode materials. With a scan rate of 0.0005 V s⁻¹, there were no obvious changes in the initial cycles, indicating a reversible and constant reaction. The first charge/discharge voltage profile for the 1st cycle at a current density of 100 mA g⁻¹ showed a longer plateau than those for subsequent cycles (Fig. 6a). This meant that the capacity from the irreversible electrolyte decomposition reaction was present during the first discharge. After that, the curves showed little change. The CV curve of R-800–120 (Fig. S15d) was different from those of the other anode materials. There were two obvious cathodic peaks, but three anodic peaks. This meant that R-800–120 contained a CoOOH phase and some amorphous phases, which were difficult to be found via XRD. Fig. 6b shows that at 25°C, R-800–300 delivered discharge capacities of 1059.5, 1081.1, 866.4, 600.1, 286.5 and 44.2 mA h g⁻¹ from 100 to 5000 mA g⁻¹, respectively. All of these regenerated materials failed to achieve sufficient rates because the discharge capacities were lower than 300 mAh g⁻¹ at 5 A g⁻¹. As shown in Fig. 6c, the cycling performance of R-600–500 was stable, with a coulombic efficiency > 97% over 100 cycles at a current density of 200 mA g⁻¹. The capacities of R-800–120 in the initial cycles were higher than those of the other materials. It was concluded that the CoOOH phase had a higher theoretical capacity than the Co₃O₄ phase. The residues containing valuable components were directly recycled for use as anode materials. Moreover, they showed similar electrochemical performance to those in other published laboratory studies [51].

Fig. 6.

(a) Charge/discharge voltage profiles for scans from 0.01 to 3.0 V (vs. Li⁺/Li) at a current density of 200 mA g⁻¹; (b) rate capacity curves of materials between 0.01 and 3.0 V (vs. Li⁺/Li); (c) capacity versus cycle number for discharge curves generated at 0.2 A g⁻¹ between 0.01 and 3.0 V (vs. Li⁺/Li); the line with balls and the line with squares represented the discharge capacity and coulombic efficiency of the materials, respectively.

4. Conclusion

In conclusion, we devised a sustainable recycling strategy for spent LIBs by harnessing mechanical forces to obtain a novel intermediate phase with high reactivity, which ensured high utilization efficiency of the leaching reagent. The leaching process was bifurcated into two stages, with the second stage typically presenting challenges due to the low reaction activity of the intermediate from the initial stage. However, the incorporation of mechanical force was transformative, significantly activating the intermediates, thereby enhancing the presence of defects and H⁺, and subsequently lowering the barrier in the second stage. With extensive characterization, including XRD, TEM, Raman, and XPS, we identified defects in the activated waste. The TG-MS results revealed the insertion of H⁺ in the intermediate. Our findings demonstrated that increased mechanical input energy resulted in more defects in the activated waste, leading to more H⁺ insertion in the activated intermediate phases. DFT results indicated that the insertion of H⁺ facilitated the second reaction. This technology, even when applied to different spent cathodes, yielded leaching efficiency of Li⁺ exceeding 90% with high utilization efficiency of H⁺ (> 97%) at 160 °C. Furthermore, the cobalt-containing phase, which was regenerated as the anodes of LIBs, demonstrated electrochemical performance comparable to those of commercial Co₃O₄ materials. Our approach leveraged an inexpensive inorganic acid in the leaching process and outperformed existing disposal methods due to the lower chemical consumption, milder reaction environment, and superior atom economy. This strategy contributes to sustainable recycling of spent LIBs.

Declaration of competing interest

The authors declare that they have no conflicts of interest in this work.

Biographies

Zhi Sun is a technical senior expert for critical raw materials recycling and cleaner production. He is corresponding author &co-author in 150 research papers, 5 books, 7 book chapters. He has lead more than 40 R&D funded projects, with 10 scientific and technological awards. He holds around 50 patents and has been working closely with industrial partners with a number of technologies being used. He is the vice director of Spent Battery Recycling Standardization Working Group under SAC/TC297/SC4, as well as both members of China Solid State Lighting Alliance Standardization Committee and China Standardization Committee (TC294). He is currently serving as associate Editor for journal of sustainable metallurgy. His research interests include: WEEE and spent battery recycling, criticality analyses of metal supply and cleaner production of critical raw materials.

Weiguang Lv graduated from the Institute of Process Engineering, Chinese Academy of Sciences. He received his B.Sc. degree in metallurgical engineering (Central South University). His current research focuses on efficient recovery of valuable metal from spent LIBs. He has published 11peer-reviewed articles and 2 patents.

Xiaohong Zheng is a doctoral student at the University of Geosciences, Beijing. He has received his B.Sc. degree in metallurgical engineering in 2012 from Anhui University of Technology and M.Sc. in Chemical engineering in 2017 from CAS. He has co-authored more than 15 publications and 5 patents. His research focuses on sustainable processes for metal recovery from urban mines.