Using green deep eutectic solvents for targeted regeneration to improve the cycle life of spent lithium iron phosphate batteries

Abstract

The rapid growth of lithium iron phosphate (LFP) batteries used in electric vehicles and energy storage systems has necessitated the development of sustainable and efficient recycling strategies to address impending end-of-life management challenges. Conventional pyrometallurgical and hydrometallurgical methods often require high energy consumption, cause environmental pollution, and suffer from economic inefficiency. In this study, a green deep eutectic solvent (DES) system, composed of lithium chloride, urea, and ascorbic acid, was developed for the direct regeneration of degraded LFP cathodes. Through integrated experimental and theoretical approaches, the regeneration mechanism was elucidated, wherein a reductive environment was provided for the conversion of Fe³⁺ to Fe²⁺, and highly efficient Li⁺ replenishment into vacant sites was achieved. The regenerated cathode material was shown to exhibit a high specific capacity of 163.5 mA h g⁻¹ at 0.1C and excellent cycling stability, with 92.6% capacity retention after 300 cycles at 1C. With significantly reduced energy consumption and economic cost compared to conventional hydrometallurgical and pyrometallurgical routes, this DES-based regeneration method presents a highly promising and sustainable pathway for resource recovery from spent lithium-ion batteries.

---

Green foundation

1. In this study, a green deep eutectic solvent (DES) was rationally designed through density functional theory calculations using lithium chloride, urea, and ascorbic acid with low-cost, biodegradable, and environmentally benign precursors. This DES system was employed for the hydrothermal regeneration of degraded lithium iron phosphate (LFP) cathodes under mild conditions, successfully restoring the electrochemical performance of the spent material without generating secondary pollution.

2. The regeneration process was systematically optimized and the mechanism was elucidated through combined experimental and theoretical approaches, highlighting the role of the DES in lithium replenishment and defect correction. Furthermore, the DES exhibited excellent reusability over multiple cycles, significantly reducing chemical waste and operational cost.

3. Economic and environmental benefit analysis demonstrates that this DES-based direct regeneration method markedly reduces energy consumption and greenhouse gas emissions, compared to conventional hydrometallurgical and pyrometallurgical routes. This work provides a sustainable strategy for the valorization of end-of-life lithium-ion batteries, aligning with the principles of green chemistry and contributing to circular economy objectives in energy storage materials.

1 Introduction

With the rapid development of the global new energy vehicle industry, the demand for lithium-ion batteries, which serve as the primary power source, has increased significantly. Among these, lithium iron phosphate (LFP) batteries have emerged as a mainstream choice for electric vehicles and energy storage systems due to their high safety, long cycle life, low cost, and environmental friendliness.¹ As the first generation of large-scale applied LFP batteries approaches end-of-life, the proper handling of spent LFP (SLFP) has become an urgent priority. It is estimated that by 2030, the global recycling market for electric vehicle and stationary energy storage batteries will exceed 1 terawatt-hour, with the recycling rate of LFP exceeding 58%.² Therefore, developing efficient and environmental-friendly recycling methods for SLFP is a critical challenge for achieving resource circularity and supporting the sustainable development of the battery industry.

The primary recycling technologies for SLFP include pyrometallurgy, hydrometallurgy and direct regeneration.^3,4 In contrast to traditional methods, direct cathode regeneration has recently gained widespread attention, which not only simplifies the recycling process but also decreases energy consumption, minimizes environmental pollution, and produces high-value regenerated products.⁵ For example, Yang et al. successfully regenerated degraded LFP cathode materials using DL-malic acid as a reductant via a combination of hydrothermal and annealing processes.⁶ Wang et al. used sodium sulfite as a reducing agent and achieved effective lithium replenishment for SLFP through hydrothermal treatment.⁷ By preserving the original structure of the cathode material, direct regeneration offers a promising strategy for sustainable recycling of SLFP.

Recently, deep eutectic solvents (DESs) have attracted growing interest in the field of LFP recycling owing to their environmental friendliness, strong designability, and low cost. Zhang et al. proposed a DES composed of chloroacetic acid and ethanol for the synergistic leaching of LFP with oxygen as the oxidant, achieving nearly 100% Li⁺ dissolution while enabling the recycling of DES.⁸ Similarly, He et al. used a DES based on choline chloride and phosphorus acid, achieving nearly complete leaching of lithium and cobalt within 20 minutes.⁹ However, most existing DES-based approaches focus on the extraction of metal elements from spent cathodes. In fact, due to their designable physical and chemical properties, DESs also hold the potential for the green and economical regeneration of degraded cathode materials. For instance, Wang et al. developed a DES using lithium chloride (LiCl) and urea that successfully restored the electrochemical performance of lithium cobalt oxide (LCO) cathodes.⁸ Lin et al. systematically investigated a DES using LiCl and ethylene glycol, demonstrating effective structural regeneration of spent cathodes at room temperature.¹⁰ Despite these promising advances in DES-assisted recycling of lithium-ion batteries, particularly for cobalt- and nickel-containing cathodes, the industrial-scale destructive recycling of LFP remains economically challenging due to its low content of high-value metals. A reasonable direct regeneration scheme for SLFP could avoid excessive cost and create economic and environmental benefits. Nevertheless, research on the direct and efficient regeneration of LiFePO₄ cathodes is still in its early stage, and there is a lack of systematic and in-depth understanding of critical issues such as how DESs interact with the SLFP crystal structure and the influence of key process parameters (e.g., composition, temperature, duration, and additives) on the electrochemical performance of regenerated materials.

In this study, a green DES system composed of LiCl, urea, and ascorbic acid was developed for the direct regeneration of spent SLFP, combining theoretical calculations and economic considerations. LiCl served as the HBA to provide the lithium source, urea functioned as the HBD and main ion-transport channel, while Vc acted as a co-HBD and reducing agent supplement, enabling concurrent lithium replenishment, structural restoration, and defect reduction. Key regeneration parameters, including temperature, duration, and the concentration of the reducing agent, were systematically optimized to enhance the electrochemical performance of SLFP. The regeneration effectiveness was comprehensively evaluated through multi-method characterization at the structural, morphological, and electrochemical levels, while the recyclability of the DES was also examined to assess its potential for sustainable application through simple ascorbic acid replenishment. The underlying regeneration mechanism was further elucidated through the integration of experimental and theoretical approaches, and the economic and environmental advantages of the DES-based direct regeneration method were quantitatively compared with those of conventional pyrometallurgical and hydrometallurgical processes, highlighting its potential as a sustainable pathway for battery recycling.

2 Experimental section

2.1 Materials and reagents

Spent batteries were provided by Hefei Guoxuan High-Tech Co., Ltd (Anhui, China). Sodium chloride (NaCl, ≥99.0%), lithium chloride (LiCl, ≥99.0%), urea, ascorbic acid (Vc, ≥99.7%), N-methyl-2-pyrrolidone (NMP), polyvinylidene fluoride (PVDF) and conductive carbon black (≥99.5%) were purchased from Shanghai Aladdin Bio-Chem Technology Co., Ltd (Shanghai, China). Celgard 2500 membrane was obtained from Zhengcheng Scientific Research Experiment Platform. The commercial electrolyte (1 M LiPF₆ in EC : EMC : DMC = 1 : 1 : 1 by volume) was supplied by Kelude Co., Ltd (Shenzhen, China).

2.2 Pretreatment of spent LFP batteries

The spent LFP cells were first fully discharged in a 5 wt% NaCl solution for 24 h to ensure safety during handling, and then transferred into an argon-filled glove box (VGB-1S, H₂O < 0.1 ppm, O₂ < 0.1 ppm) for disassembly. The positive electrodes were carefully separated from the current collectors. To remove the PVDF binder and facilitate detachment of the active material, the electrode sheets were immersed in boiling deionized water at 100 °C for 10 min. The aluminum foil was manually peeled off, and the active material was collected and dried at 80 °C for 12 h in a vacuum oven. The dried material was ground into fine powder using an agate mortar. To eliminate residual electrolytes, conductive additives, and the solid electrolyte interphase (SEI), the powder was heated in a tube furnace (OTF-1200X-5L, Hefei) under a nitrogen atmosphere at a heating rate of 5 °C min⁻¹ to 450 °C and held for 2 h. The resulting black powder, denoted as SLFP, was stored in a glove box for further use.

2.3 Preparation of a deep eutectic solvent (DES)

LiCl (hydrogen bond acceptor, HBA), urea, and ascorbic acid (hydrogen bond donors, HBDs) were weighed at molar ratios of 1 : 3 : x (x = 0, 0.1, 0.2) and transferred into a round-bottomed flask. The mixture was magnetically stirred and heated at 100 °C until a homogeneous, colourless, and transparent liquid formed. The resulting DES was stored in a sealed container at room temperature for further use.

2.4 Direct regeneration process

The as-prepared DES and SLFP powder were combined at a preset ratio in a 100 mL Teflon-lined hydrothermal reactor. The mixture was stirred thoroughly to ensure uniform contact and then subjected to hydrothermal treatment in a blast drying oven at preset temperatures (80–120 °C) for different durations (3–24 h). After the reaction, the mixture was cooled to room temperature, washed repeatedly with distilled water and ethanol to remove the residual DES and impurities, and dried at 100 °C for 10 h to obtain the regenerated precursor. The precursor was then annealed in a tube furnace under a nitrogen atmosphere. The temperature was raised to 650 °C at a heating rate of 5 °C min⁻¹ and held for 2 h to promote crystallization and carbon layer formation. The final product was collected as regenerated LFP (designated RLFP).

2.5 Electrochemical performance testing

The electrochemical performances of SLFP and RLFP were tested by assembling CR2032 coin cells, with commercial LFP (LFP) as a reference. The cathode slurry was prepared by mixing the active material, PVDF binder, and conductive carbon black at a mass ratio of 8 : 1 : 1 in NMP solvent. The mixture was stirred magnetically for 4 h to form a homogeneous slurry. The slurry was then coated onto aluminium foil using a doctor blade, dried at 80 °C for 4 hours in a blast drying oven, and then transferred to a vacuum drying oven for further drying at 120 °C for 8 hours to completely remove the residual solvent. The dried electrodes were calendared to a thickness of 50 μm and punched into 15 mm diameter discs. Coin cells were assembled in an argon-filled glovebox (H₂O < 0.1 ppm, O₂ < 0.1 ppm). Each cell consisted of a cathode disk of 15 mm-diameter, a Celgard 2500 separator (18 mm diameter), a lithium metal foil (15.8 mm diameter) as the counter/reference electrode, and an electrolyte (1 M LiPF₆ in EC : EMC : DMC = 1 : 1 : 1 by volume). The assembled cells were sealed and rested for 12 h before electrochemical testing to ensure complete electrolyte wetting.

Galvanostatic charge–discharge tests were performed using a LAND battery testing system (CT2001A, Wuhan LAND Electronics Co., Ltd) within a voltage window of 2.5–4.0 V (vs. Li⁺/Li). The rate capability was evaluated by cycling the cells at current densities of 0.1, 0.2, 0.5, 1, 2, and 3C, and again at 0.2C (where 1C = 170 mA g⁻¹), with 5 cycles at each rate. Long-term cycling stability was tested at 1C for 300 cycles. All tests were conducted at room temperature.

Cyclic voltammetry (CV) and electrochemical impedance spectroscopy (EIS) measurements were performed using a CHI660E electrochemical workstation (Chenhua Instrument, Shanghai, China). For CV tests, the potential was scanned between 2.5 and 4.2 V (vs. Li⁺/Li) at a scan rate of mV s⁻¹. EIS tests were conducted under different State of Charge (SOC) conditions within a voltage range of 2.5–4.2 V, with a frequency range of 0.01–100 kHz. All measurements were carried out at room temperature.

2.6 Characterization

The structure of cathode materials and DES properties were characterized using Fourier transform infrared spectroscopy (FTIR, Nicolet iS50). Inductively coupled plasma optical emission spectrometry (ICP-OES, Arcos II MV) was used to determine the compositions of the cathode materials. X-ray diffraction (XRD, D8 Advance 250) with Cu Kα radiation characterized the crystal structure of the powders, with 2θ scans from 10° to 90°. X-ray photoelectron spectroscopy (XPS) with a Thermo Fisher ESCALAB Xi and Al Kα radiation (hv = 1486.6 eV) was performed to analyse elemental distributions, with binding energies calibrated to the standard C 1s peak at 284.8 eV. Scanning electron microscopy (SEM, SU8010) and transmission electron microscopy (TEM, FEI Tecnai G2 F30) were used to examine the sample morphology.

2.7 Computational methods

2.7.1 Density functional theory (DFT) calculations. DFT calculations were carried out to screen potential hydrogen bond acceptors (HBA) and hydrogen bond donors (HBD). The initial molecular structures of the target compounds were constructed using Gauss View 6.0. Geometry optimization and energy calculations were performed using the B3LYP functional and the 6-31G(d) basis set within the Opt + Freq module. The Opt + Freq module was employed to obtain the optimized molecular configurations and key electronic properties, including HOMO energy, LUMO energy, and the total energy of each system, denoted as E_(A). The energy of a single lithium ion, E_Li⁺, was also computed under the same theoretical framework. The lithium ion was then introduced into the pre-optimized HBA structure, and the entire Li⁺–A complex was re-optimized to determine its total energy, E_(Li⁺–A). The adsorption energy between the lithium ion and the HBA molecule was calculated using eqn (1). A more negative E_β value indicated a stronger interaction and a greater tendency for lithium-ion adsorption.

E_β = E_(Li⁺–A) − E_(Li) − E_(A) (1)

2.7.2 Molecular dynamics (MD) simulations. The (010) crystal surface of LiFePO₄ was selected for interfacial behaviour analysis. The model was constructed using the Build Layer and Amorphous Cell modules in Materials Studio 2020 and integrated into a simulation box. MD simulations were conducted using the Forcite module with the COMPSSIII force field. The NVT ensemble was utilized for system optimization with a velocity assignment set to “current”, the temperature set to 300 K using the Nose thermostat, a step size of 1 fs, and a total simulation time of 500 ps to simulate the behaviour of materials under constant volume and temperature. van der Waals interactions and Ewald electrostatic interactions were treated under a cutoff distance of 12.5 Å. Components involved in the simulation were processed using the DMol3 module. Generalized gradient approximation (GGA) with the PBE function was conducted to solve the exchange–correlation functional with a polarization (DND) basis set. A smearing of 0.005 Ha of the orbital occupation was applied to accelerate electronic convergence. Geometry optimizations were obtained until the maximum displacement was below 0.005 Å, energy tolerance below 1 × 10⁻⁵ Ha, and the maximum forces lower than 0.004 Ha Å. The TS correction method was applied for DFT-D dispersion interactions to enhance the description of the adsorbed system.

3 Results and discussion

3.1 Design and synthesis of DES for S-LFP regeneration

DESs are eutectic mixtures formed by mixing HBAs, such as quaternary ammonium salts and metal salts, with HBDs, such as urea, carboxylic acids, polyols, or sugars, in specific molar ratios.¹¹ These components interact through a strong hydrogen-bonding network, resulting in liquids with advantages including simple preparation, low cost, readily available constituent, biodegradability, and high tunability.¹² Due to their design flexibility, DESs are now widely used in the treatment of spent lithium-ion battery materials, mainly through two strategies, namely leaching and recovery of valuable metals utilizing their inherent acidity, and direct regeneration of cathode materials. For the regeneration of spent materials like LFP, it is essential to clarify their failure mechanism first. As stated previously, during long-term cycling, the thickening of the SEI film leads to the irreversible loss of the active lithium.^13,14 Lithium deficiency leads to the formation of lithium vacancies (Li_v), triggering the oxidation of Fe(II) to Fe(III) and resulting in the generation of Li–Fe anti-site defects. These defects hinder lithium-ion diffusion pathways, restrict the kinetics of electrochemical reactions, and consequently deteriorate the electrochemical performance of the cathode.¹⁵ Moreover, repetitive cycling-induced mechanical stress can damage the carbon coating layer, further aggravating the degradation of LFP.¹⁶ Therefore, an effective DES for regenerating spent LPF must fulfil several critical functions, including precise lithium replenishment at lithium-deficient sites, provision of a reductive environment to revert the Fe(III) to Fe(II) phase, and correction of anti-site defects.¹⁵ Meanwhile, the DES should support sufficient mobility of Li⁺ to guarantee adequate driving force for migration into the vacant sites. Based on these criteria, the adsorption energy between Li⁺ and candidate HBDs was calculated to evaluate the diffusion trend of Li⁺ in the solvent system. In addition, the highest occupied molecular orbital (HOMO) energy levels of different HBDs were assessed to quantify their electron-donating capabilities and thus their reductive potential.

Water, ethanol, ethylene glycol, glycerol, and urea were considered as potential HBDs, as shown in Fig. 1a. Although water exhibited high lithium-ion mobility, its reducing ability was relatively low, rendering it unsuitable for regeneration. Ethanol, ethylene glycol, and glycerol possessed similar HOMO energy levels, measured at −7.58, −7.55 and −7.55 eV, respectively, while urea exhibited a higher HOMO energy level of −7.31 eV, indicating a stronger reducing ability. In addition, urea had the most negative adsorption energy for lithium ions (−2.54 eV) among the candidates, indicating an enhanced diffusion ability within its matrix, which is beneficial for the Li⁺ supplementation process required during regeneration. Based on these results, urea was selected as the HBD for subsequent studies (Fig. 1b).

Fig. 1 Design concept and synthesis of the DES. (a) Adsorption energy of lithium ions for different HBDs; (b) HOMO energy levels of different candidate HBDs; (c) synthesis route diagram of the DES; (d) FTIR spectra of the synthesized DES and its individual components.

To satisfy the requirement of lithium supplementation, HBAs should provide easily dissociable Li⁺. Thus, common lithium salts, such as LiCl, lithium sulfate (Li₂SO₄), and lithium bromide (LiBr), were considered. Li₂SO₄ was excluded due to its high melting point and poor solubility, which prevent the formation of a stable eutectic system. The adsorption energies and costs of LiBr, LiF, and LiCl with urea were further compared, as illustrated in Table S2. The differences in the adsorption energies among these lithium salts were negligible, while LiCl presented a cost advantage. Finally, LiCl was selected as the HBA.

Based on the above analysis, LiCl and urea were selected as the HBA and HBD, respectively, to form the DES. By optimizing the molar ratio of urea to LiCl and the synthesis temperature, a homogeneous transparent DES solution was successfully obtained (Fig. 1c). To verify the formation of the DES, FTIR spectroscopy was employed to detect the structural changes in the components. As shown in Fig. 1d, a comparison between the synthesized DES and its original components revealed that the spectrum of the synthesized DES solution was not merely a superposition of those of pure LiCl and urea. Characteristic vibrational peaks exhibit a slight red shift at 1150, 1630, and 3450 cm⁻¹, which was attributed to the enhanced intramolecular and intermolecular hydrogen bonding between the two components.¹⁷ Meanwhile, the peak spectrum of the eutectic mixture retained the original characteristic peaks of the two components. These indicated that the DES was formed through hydrogen-bond interactions without significant chemical reactions, confirming the successful preparation of the LiCl–urea DES.

3.2 Electrochemical performance of regenerated LFP

To systematically optimize the regeneration process, hydrothermal time, temperature, and the amount of Vc were selected as experimental variables. Meanwhile, the rate capabilities of the regenerated samples were evaluated at current rates of 0.1, 0.2, 0.5, 1, 2, and 3C, and back to 0.2C to determine the optimal regeneration conditions. The initial discharge capacity of SLFP was only 70.6 mA h g⁻¹. Under the fixed hydrothermal time and Vc addition (12 h and 0.1 mol/mol Vc), the initial specific capacities achieved at five different temperatures of 80, 90, 100, 110, and 120 °C were 148.1, 145.6, 163.5, 153.5, and 143.2 mA h g⁻¹, respectively (Fig. 2a). These results showed significant performance recovery across all temperatures, with the optimal effect observed at 100 °C. Beyond this temperature, the capacity decreased notably, likely due to the partial thermal decomposition of DES components, which affected the regeneration efficiency. A similar trend was observed under varying hydrothermal durations (Fig. 2b). Specifically, the initial specific capacities after 3, 6, 12, and 24 h of treatment were 110.7, 111.8, 163.5, and 151.4 mA h g⁻¹, respectively. A longer hydrothermal time promoted more sufficient lithium diffusion and a better regeneration effect to a point, beyond which prolonged treatment might induce side reactions or structural degradation. In addition, to evaluate the necessity of Vc, the rate performance of a control sample without Vc was compared simultaneously. As shown in Fig. S1, the first charge–discharge capacity of the Vc-free sample was only 117.3 mA h g⁻¹, indicating that the reducing capacity provided by urea alone was inadequate for effective regeneration. The optimal regeneration conditions were determined as 0.1 mol/mol Vc addition, 12 h hydrothermal time, and 100 °C temperature, with an initial specific capacity of 163.5 mA h g⁻¹ and a high specific capacity of 113.1 mA h g⁻¹ even at a high rate of 3C. These results confirmed that the DES-based hydrothermal system can effectively restore the structural and electrochemical properties of degraded SLFP.

Fig. 2 Electrochemical performance comparison between SLFP and RLFP. (a) Rate capability at different regeneration temperatures; (b) rate capability under different hydrothermal durations; (c) CV curves at 0.1 mV s⁻¹; (d) CV curves at different scan rates, (e) linear fitting of peak current versus square root of scan rate; (f) galvanostatic intermittent titration technique profiles; (g) initial charge–discharge curve at 0.1C; (h) electrochemical impedance spectroscopy Nyquist plots at different state-of-charge (SOC) levels; (i) long-term cycling performance at 1C.

Subsequent electrochemical tests were conducted under the optimized regeneration conditions. The regenerated lithium iron phosphate (RLFP) battery and SLFP were systematically subjected to various electrochemical tests to intuitively characterize the regeneration effect. Fig. 2c shows the cyclic voltammetry (CV) curves of RLFP and SLFP measured at a scan rate of 0.1 mV s⁻¹. Both samples exhibited distinct oxidation and reduction peaks near 3.6 V and 3.3 V, corresponding to the Fe(II)/Fe(III) redox couple.¹⁸ The potential difference between the anodic and cathodic peaks for the RLFP sample was 0.279 V, which was smaller than that of the SLFP sample (0.302 V), indicating significantly improved electrochemical reversibility after regeneration. To further investigate the kinetic properties of the regenerated material, CV measurements of RLFP were performed at scan rates from 0.1 to 1.0 mV s⁻¹ (Fig. 2d). The relationship between peak current and scan rate was analysed using the Randles–Sevcik equation.

I_p = (2.69 × 10⁵)·n^3/2·A·D^1/2·C·υ^1/2 (2)

where I_p is the peak current, n is the number of electrons exchanged per molecule or ion during the redox process, A is the electrode surface area, D is the diffusion coefficient of lithium, C is the concentration of lithium in the redox process, and υ is the scan rate. Assuming parameters such as n, A, and C remained constant for a given electrode system, the diffusion coefficient D is proportional to the slope of I_pversus υ^1/2.¹⁹ As shown in Fig. 2e, the fitted slopes of the anodic and cathodic peak currents (K_a = 1.932, K_b = 2.611) for RLFP are significantly larger than those of SLFP (K_a = 0.212, K_b = 0.295), indicating a substantial enhancement in lithium-ion diffusion kinetics after regeneration. This finding was further supported by galvanostatic intermittent titration technique (GITT) tests, as presented in Fig. 2f. Compared with SLFP, RLFP exhibited a notably extender charge–discharge plateau, reflecting its enhanced capacity and reaction kinetics. The overpotential required for lithium extraction/insertion was markedly reduced in RLFP, suggesting decreased polarization and more rapid ion transport. Additionally, the calculated lithium-ion diffusion coefficients derived from the GITT tests (Fig. S2) confirmed higher ion mobility in RLFP throughout the phase transition process. SLFP demonstrated higher resistance to lithium-ion diffusion, which could be attributed to the presence of Li–Fe anti-site defects that blocked the one-dimensional diffusion channels in the olivine structure, thereby increasing the energy barrier for lithium-ion diffusion. Fig. 2g shows the galvanostatic charge–discharge (GCD) profiles of SLFP and RLFP at 1C. The voltage difference of SLFP was 142 mV, while that of the RLFP was reduced to only 62 mV, providing further evidence of improved reaction kinetics and electrochemical reversibility following regeneration.

EIS and corrected Zview fitting were conducted to analyse the interfacial and charge transfer behaviour as shown in Fig. 2h and Tables S3 and S4. Eleven state-of-charge (SOC) levels, namely charge-2.5 (C-2.5), C-3.0, C-3.4, C-3.5, C-3.6, C-4.2, discharge-4.0 (DC-4.0), DC-3.4, DC-3.3, DC-3.2, and DC-3.0, were selected for in-depth analysis of the batteries. The intercept of the curve with the horizontal axis represented the ohmic resistance (R_s), which included contributions from the electrolyte and electrical contacts. The semicircle in the high-frequency range corresponded to the impedance of the SEI film (R_SEI), while the semicircle in the medium-frequency range was attributed to the charge transfer impedance (R_ct). The linear segment in the low-frequency region reflected the Warburg impedance (Z_w),²⁰ which was associated with solid-state lithium-ion diffusion within the electrode material. Tables S3 and S4 show that the RLFP cathode had a slightly higher inherent impedance (R_s) due to the difference in coin cell assembly technologies or other experimental variations. Comparatively, the more critical parameter for evaluating electrode kinetics R_ct exhibited consistent and significant improvement in all RLFP samples compared to SLFP. During the charging stage, R_ct decreased gradually and reached a minimum in the fully charged state. Upon discharging, R_ct showed a slight increase due to the occurrence of lithium intercalation, and then stabilized at a certain value.²¹ Specifically, during the charging phase, the R_ct value of SLFP gradually decreases from 583.9 Ω to 303.4 Ω. Subsequently, it increased again in the discharge phase and reached 545.5 Ω at the DC-3.0. Although RLFP exhibited a similar resistance trend to SLFP throughout the cycling process, its internal impedance was significantly lower at all states of charging.

To evaluate the practical electrochemical performance of the batteries, long-term cycling tests were carried out at 1C rate. As shown in Fig. 2i, RLFP still retained a high specific capacity of 119.5 mA h g⁻¹ after 300 cycles, with a capacity retention rate of 92.6%. In contrast, SLFP had markedly poor stability during long-term charge–discharge processes, with more significant performance degradation and a remaining capacity retention rate of 83.6% after the same number of cycles. These results collectively demonstrated that the DES-based regeneration process effectively enhanced the electrochemical properties of degraded LFP materials, as reflected in the reduced internal impedance, improved reaction kinetics, and superior cycling stability.

Furthermore, to validate the recyclability of the DES, the filtrate obtained after vacuum filtration was distilled and reused for DES synthesis, which was subsequently applied to regenerate the additional spent LFP material. As observed from the initial charge–discharge curves (Fig. S3), the regenerated samples maintained capacities comparable to those treated with fresh solvent, retaining 159.6 mAh g⁻¹ even after the fifth regeneration cycle. This excellent retention of the regenerative performance confirmed the outstanding reusability of the designed DES system, highlighting its potential to significantly reduce reagent consumption and operational costs in sustainable battery recycling processes.

3.3 Microstructural characterization of LFP before and after regeneration using the DES

To evaluate the regeneration effect and compare the structural differences between SLFP and RLFP, scanning electron microscopy (SEM) was used to characterize their morphologies. As shown in Fig. 3a and f, both materials exhibited similar particle sizes and aggregation states, indicating that the regeneration process did not cause physical damage or significant morphological alteration. To gain a deep insight into the phase and structure of the materials before and after hydrothermal treatment, TEM was employed to capture the images of different regions (Fig. 3d and i), wherein three representative regions were selected. Region I (Fig. 3d1), taken from the core of SLFP, shows well-defined lattice fringes with an average spacing of 0.4268 nm from 10 diffraction peaks, matching the (1 0 1) crystal plane of LiFePO₄. Region II (Fig. 3d2) displays distorted and blurred fringes, and its fast Fourier transform (FFT) analysis showed a disordered diffraction pattern, reflecting a mixed phase of degraded FePO₄ and residual LiFePO₄. In region III (Fig. 3d3), located at the particle edge, narrow and irregular lattice fringes are observed. The average lattice spacing measured from 10 diffraction peaks yielded a lattice spacing of 0.4321 nm, corresponding to the (110) crystal plane of FePO₄ (Fig. 3e), indicating lithium loss and phase degradation. These findings suggest that Li loss mainly occurred on the particle surfaces and near-surface regions. In contrast, the TEM image of the DES-treated RLFP (Fig. 3i) shows uniform and continuous lattice fringes across the particle. Measurements from 10 diffraction peaks revealed a consistent lattice spacing of 0.349 nm, indexed to the (111) crystal plane of LiFePO₄ (Fig. 3j). The corresponding FFT pattern of RLFP exhibited ordered and regular diffraction spots, demonstrating the recovery of a homogeneous crystalline LiFePO₄ structure after DES treatment. This phenomenon was attributed to the selective diffusion of lithium from the DES to the vacant sites in SLFP, which drove the conversion of the FePO₄ and disordered mixed phases back into a homogeneous well-crystallized LiFePO₄ structure.

Fig. 3 Microstructural characterization of SLFP and RLFP. (a and f) SEM images; (b and g) TEM images at 50 nm scale; (c and h) TEM images at 20 nm scale; (d and i) HRTEM images; (e and j) analysis and measurement of lattice spacing; (k and l) HAADF-STEM images and the corresponding EDS elemental mappings.

Energy dispersive X-ray spectroscopy (EDS) mapping was performed to analyse the elemental distribution of SLFP and RLFP (Fig. 3k and l). Both samples showed the homogeneous distributions of Fe, P, and O, indicating a stable and regular crystal structure of LiFePO₄. Additionally, the carbon mapping revealed a uniformly continuous carbon layer on the RLFP surface, whereas SLFP exhibited a fragmented coating structure. This disruption in SLFP was likely due to accumulated mechanical stress and side reactions during long-term charge–discharge cycles. Benefiting from the subsequent annealing treatment, the uniformity and integrity of the carbon coating in RLFP were enhanced, which significantly improved its electrical conductivity. Furthermore, nitrogen EDS mapping showed an enhanced nitrogen signal in RLFP compared to SLFP, suggesting the incorporation of nitrogen which was originated from urea in the DES into the carbon layer. The nitrogen-doped carbon coating contributed to the improved electrical conductivity by providing more charge carriers and facilitating electron transfer; meanwhile, due to the higher electronegativity of nitrogen compared to carbon, local negatively charged regions are formed on its surface. These regions reduce the tendency of carbon particle agglomeration through electrostatic repulsion, enabling the carbon coating to form a more uniform and porous structure, which helped reduce impedance, mitigate polarization, and facilitated faster ion transport.^22–24

To intuitively demonstrate the lithium supplementation effect, ICP-OES is used to quantify the iron, lithium, and phosphorus contents in the material. As shown in Fig. 4a and Table S5, the molar ratios of Li/Fe and Li/P in SLFP are only 0.739 and 0.8, respectively, confirming a non-stoichiometric Li-deficient state. After hydrothermal treatment with the DES, these molar ratios in RLFP increased to 0.957 and 1.03, respectively, approaching the theoretical value of stoichiometric LiFePO₄, which suggested effective lithium replenishment. XPS was performed to further elucidate the valence and bonding state of each element in the materials for both SLFP and RLFP. In the Fe 2p (Fig. 4b) peak, the spectrum of SLFP exhibited two fitted peaks at 709.4 eV and 710.04 eV, corresponding to Fe(II) and Fe(III) in the 2p_3/2 energy level, respectively,^25,26 indicating severe surface iron oxidation. Comparatively, the Fe 2p spectrum of RLFP (Fig. 4c) was dominated by a main peak at 710.5 eV, which corresponds to Fe(II).²⁷

Fig. 4 Analysis of the composition change and phase transition of lithium iron phosphate. (a) ICP-OES analysis results of SLFP and RLFP, (b–e) XPS results of SLFP and RLFP, (f) FTIR results of SLFP, RLFP, and pristine LFP, (g) XRD analysis results of SLFP and RLFP, and (h and i) XRD Rietveld refinement profiles of XRD patterns.

The above result demonstrates the highly effective reduction of Fe³⁺ to Fe²⁺ by the reductive DES environment, consistent with the recovery of electrochemical activity. As shown in Fig. S8, in the O 1s peak spectrum, consistent phase diagrams were observed for both SLFP and RLFP, confirming that the stable olivine crystal structure framework of lithium iron phosphate remained undamaged during the repair process. The C 1s XPS spectra provided additional insight into the surface modification. In SLFP (Fig. 4d), the spectrum showed a dominant peak at 290.88 eV, assigned to the residual polyvinylidene fluoride (PVDF) binder,²⁷ and a notably stronger peak at 288.68 eV, corresponding to C=O bonds. The pronounced C=O signal likely originated from residual electrolyte salts, their decomposition products, and oxidative degradation of the conductive carbon black during long-term cycling,^28,29 reflecting the degraded surface state of the spent material. In RLPF (Fig. 4e), a distinct fitted peak appeared at 285.6 eV, attributed to the C–N bond,³⁰ corroborating the DES mapping results and indicating the formation of a nitrogen-doped carbon layer during regeneration.

In the Fourier Transform Infrared (FTIR) spectrum (Fig. 4f), a blue shift was observed in the vibration band corresponding to the PO₄ group after regeneration. This shift indicated that DES treatment modified the electronic density distribution around the phosphate group, resulting in an increase in P–O bond energy. This enhance was beneficial for the rigid framework of the olivine structure and improved the structural stability of the regenerated LFP.³¹ XRD patterns of the two materials are also compared in Fig. 4g. It can be found that RLFP has sharper and more intense diffraction peaks than SLFP, suggesting improved crystallinity and a more ordered microstructure after regeneration that was consistent with the previous TEM results. Although no diffraction peaks corresponding to the degraded FePO₄ phase were observed in SLFP, this absence might be due to its low proportion and incomplete structural conversion, leaving the material in a mixed two-phase state. After magnifying the characteristic peaks, it can be observed that compared with the peak spectrum of SLFP, the (200) diffraction peak of RLFP shifts slightly to a lower angle. According to Bragg's law:

2d sin θ = nλ (3)

where n denotes the diffraction order and λ represents the X-ray wavelength. A decrease in the diffraction angle θ implies an increase in the lattice spacing d. The expanded lattice spacing in RLFP likely facilitates lithium-ion diffusion by providing wider transport pathways.

Further structural analysis was carried out using Rietveld refinement of the XRD pattern (Fig. 4h, i and Table S6). Both SLFP and RLFP were confirmed to crystallize in the same orthorhombic space group Pnma (no. 62). The refinement results clearly demonstrate that the reductive environment provided by the DES treatment significantly lowered the energy barrier for the migration of mislocated Fe ions, enabling their movement from Li sites back to the original Fe sites. Consequently, the proportion of Fe–Li anti-site defect decreased from 6.6% in SLFP to 2.4% in RLFP, demonstrating the effective remediation of crystal defects through the DES regeneration process.

3.4 Mechanism analysis of the direct regeneration process of SLFP by DES

MD simulation was employed to gain deeper insights into the coupling interactions between the DES system and the degraded SLFP (Li_0.739FePO₄). The adsorption energies of Li⁺ on both the SLFP crystal surface and within the urea-based carrier medium were systematically evaluated. As shown in Fig. 5a, the interaction energy between Li⁺ and the host lattice in SLFP was calculated to be −254.72 kJ mol⁻¹, while that between Li⁺ and urea reached −274.02 kJ mol⁻¹. The more negative value indicates a stronger affinity, suggesting that within the established model, Li⁺ exhibited a higher tendency to diffuse through the urea medium rather than remaining bound to the defective lattice. In contrast, the intermolecular interaction between urea and Li⁺ already embedded in SLFP was significantly weaker with only −132.18 kJ mol⁻¹ (Fig. S5). This energy difference implied that under appropriate experimental conditions, Li⁺ could be efficiently transported via urea and subsequently incorporated into lithium-deficient regions of SLFP without being strongly recaptured by the solvent.

Fig. 5 Molecular dynamics simulation analysis of the DES regeneration method for SLFP. (a) Intermolecular energy of Li⁺ within the system; (b) RDF of Li⁺ relative to oxygen atoms on the Li_0.75FePO₄ surface; (c) density distribution of Li⁺; (d) mean square displacement curves for adsorbed and free Li⁺ on the Li_0.75FePO₄; and (e–i) schematic illustration of the adsorption and regeneration process, showing progressive Li⁺ incorporation and structural recovery.

To further investigate the Li⁺ replenishment process, RDF was used to analyse the proximity between oxygen atoms on the SLFP and adjacent Li⁺ in the DES, determining whether Li⁺ in the DES could effectively diffuse into the supplementary sites. As depicted in Fig. 5b, a pronounced peak appeared at approximately 1.7 nm, confirming the adsorption of Li⁺ ions onto the SLFP surface, providing a critical preliminary step for effective Li⁺ supplementation into vacant sites. Moreover, the coordination state of Li⁺ was examined through spatial localization analysis, as illustrated in Fig. 5c, which showed the accumulation of Li⁺ at the material–solvent interface. The diffusion coefficients of free Li⁺ were also quantitatively assessed as shown in Fig. 5d, in which the diffusion coefficients of free Li⁺ in the DES medium and of Li⁺ adsorbed on the S-LFP surface were compared. The adsorbed Li ions exhibited a significantly reduced diffusion mobility, indicating that the adsorption process was followed by a stabilized, directed incorporation into the crystal lattice. This restricted diffusion significantly lowered the probability of adsorbed Li⁺ returning to the solvent, thereby enhancing the efficiency of lithium replenishment.

Based on the comprehensive characterization presented in previous sections (e.g., XRD, XPS, EIS, TEM) combined with the DFT and MD simulation results outlined above, the regeneration mechanism of S-LFP via the DES-mediated process can be summarized as shown in Fig. 5e–i. After mixing and stirring, the liquid DES thoroughly encapsulated the SLFP particles. The synergistic reducing environment provided by urea and ascorbic acid facilitated the reduction of Fe³⁺ to Fe²⁺, thereby mitigating cation disorder and reopening obstructed lithium diffusion channels. Meanwhile, the molten urea acted as a transport path for the free Li⁺ dissociated from LiCl, enabling efficient diffusion to lithium-deficient sites. Li⁺ was then incorporated into the crystal structure, restoring the stoichiometric composition of LiFePO₄. In addition, urea, as a precursor of carbon and nitrogen sources, provided a nitrogen-containing coating for the surface coating of the cathode powder. The subsequent annealing process further promotes carbonization, which not only enhances the lattice structure but also decomposes and facilitates the formation of a nitrogen-doped carbon coating on the regenerated particles. This carbonization process enhances the crystallinity of the material and produces a uniform, conductive carbon layer that improved the electronic conductivity, reduced interfacial impedance, and further stabilized the structure during cycling. As a result, the regenerated cathode demonstrated significantly improved lithium transport kinetics, structural stability, and overall electrochemical performance (Fig. 6).

Fig. 6 Mechanism of the DES-mediated regeneration process for SLFP.

To sum up, this multi-step mechanism, encompassing reduction, lithiation, and carbon coating, highlighted the effectiveness of the DES-based approach in directly regenerating SLFP in a single integrated process, offering a sustainable and efficient alternative to conventional recycling methods.

3.5 Analysis of economic and environmental benefits

Fig. 7a schematically compares the basic processes of three main recycling methods for SLFP batteries, including pyrometallurgical recycling, hydrometallurgical recycling, and direct regeneration. Among these, the direct regeneration method allowed for the restoration of the degraded LFP cathode material to a state suitable for direct reuse in new battery production. In contrast, pyrometallurgical and hydrometallurgical processes typically yielded lower-value intermediate products such as metal alloys or chemical precursors, which require further processing for reutilization in battery production. Fig. 7b and c present a comparative analysis of the energy consumption and greenhouse gas (GHG) emissions associated with these different recycling technologies, based on the data from the EverBatt 2023 database. The direct regeneration method demonstrated markedly superior environmental performance, with a total energy consumption of 11.67 MJ kg⁻¹ of processed material and GHG emissions of 0.62 kg kg⁻¹ cell. In comparison, pyrometallurgical and hydrometallurgical recycling processes were considerably more energy- and emission-intensive, consuming 34.336 MJ kg⁻¹, 55.721 MJ kg⁻¹, and 1.96 kg kg⁻¹ cell, 1.493 kg kg⁻¹ cell, respectively. These results suggested the significant advantages of direct regeneration in energy conservation and emission reduction, as well as notable environmental benefits.

Fig. 7 Economic analysis of recycling processes. (a) Schematic comparison of process flows; (b) total energy consumption among different processes; (c) greenhouse gas emissions among different processes; (d) benefits versus for hydrometallurgy and direct regeneration; (e) detailed cost breakdown for hydrometallurgy and direct regeneration; (f) proportion of each component in battery manufacturing.

In addition, an economic assessment was conducted to evaluate the costs and benefits of recycling one ton of spent LFP using hydrometallurgy and direct regeneration (Fig. 7d). The direct regeneration method incurred a cost of 14 019 CNY per ton and yields a benefit of 18 306 CNY, resulting in a favorable net profit of 4287 CNY. Hydrometallurgical recycling, although achieving a benefit of 16 065 CNY per ton, required a higher processing cost of 15 039 CNY, leading to a marginal net economic return of only 1206 CNY. The superior profitability of direct regeneration was attributed to its simplified process and higher-value output. Furthermore, a detailed breakdown of cost structures for both methods is provided in Fig. 7e. The hydrometallurgical process relies heavily on chemical reagents and large volumes of water and involves complex operational stages, such as leaching, purification, and chemical resynthesis. These factors contribute to its higher operational and capital costs. In comparison, the direct regeneration method simplified the process flow by avoiding destructive leaching and reprecipitation steps, thereby significantly reducing the consumption of chemicals, water, and energy, as well as lowering equipment and operational expenses.

As a whole, this techno-economic analysis systematically compared direct regeneration with conventional hydrometallurgical recycling, on the basis of idealized process models excluding factors such as transportation, collection, and sorting. The results affirmed that the direct regeneration method offered not only substantial environmental benefits due to its low energy consumption and minimal emissions, but also enhanced the economic viability through streamlined operations and higher-value output. Furthermore, the comparative analysis was expanded to incorporate various direct regeneration strategies reported in recent years, as summarized in Table S15. The developed DES system was demonstrated to achieve a competitive profit of $1.799 per kg, comparable to the highest-performing methods. This economic viability was attributed to the excellent recyclability of the system and the thermal stability of its components, which allowed effective recovery and reuse through straightforward distillation operations. This dual advantage positions direct regeneration as a promising sustainable pathway for the circular management of spent LFP batteries.

4 Conclusions

In summary, this study successfully achieved non-destructive structural regeneration of spent LFP using a green and low-cost DES, where LiCl served as both a lithium source and HBA, urea functioned as a HBD and ion transport medium, and Vc acted as a co-HBD and reducing agent supplement. Through a mild hydrothermal process followed by short annealing, the DES effectively facilitated structural regeneration and lithium replenishment, while simultaneously forming a nitrogen-doped carbon coating on the regenerated material. DFT calculations confirmed an efficient one-dimensional directional lithium supplementation process, supported by molecular dynamics simulations showing preferential lithium diffusion through the EDS medium. The results of electrochemical tests indicated that the regenerated LFP electrode exhibited outstanding electrochemical performance, delivering a high initial charge/discharge capacity of 163.5 mA h g⁻¹ at 0.1C and maintaining excellent rate capability with a capacity of 113.1 mA h g⁻¹ even at 3C. After 300 cycles at 1C, the material retained 92.6% of its capacity, demonstrating superior cycling stability. Comparative techno-economic analysis showed that the direct regeneration method offered inherent advantages in efficiency, simplicity, and cost-effectiveness, reducing energy consumption by approximately 66%–79% and greenhouse gas emissions by 58%–68% compared to conventional hydrometallurgical and pyrometallurgical processes. The DES-based regeneration strategy not only addresses the environmental challenges associated with spent battery disposal, but also shows considerable potential for large-scale application, thereby supporting the transition toward a circular economy in the energy storage industry.

Author contributions

Jin Wu: writing – original draft, formal analysis, data curation, and investigation. Lin Chen: writing – review & editing, funding acquisition, conceptualization, and resources. Ruichao Zhu: investigation. Yixuan Zhou: investigation. Chuqing Cao: investigation and funding acquisition. Liang Zhu: funding acquisition and resources. Jun Zhang: writing – review & editing.

Conflicts of interest

The authors declare that there is no competing interest or personal relationship of any kind that might have influenced the work reported in this paper.

Data availability

The data included in this article are available within the article and its supplementary information (SI). Supplementary information is available. See DOI: https://doi.org/10.1039/d5gc04838a.

These data are also available from the corresponding authors upon request.

Acknowledgements

This study was financially supported by the Central Guided Local S&T Development Fund, China (grant number: 202407a12020011) and the Open Research Fund of Anhui Province Key Laboratory of Machine Vision Inspection (KLMVI-2023-HIT-09).

References

1. Y. J. Ding, J. Y. Fu, S. G. Zhang, X. F. He, B. H. Zhao, J. Ren, J. H. Zhong and Z. J. Liu, Sep. Purif. Technol., 2024, 338.
2. Y. Yang, E. G. Okonkwo, G. Y. Huang, S. M. Xu, W. Sun and Y. H. He, Energy Storage Mater., 2021, 36, 186–212.
3. A. Battistel, M. S. Palagonia, D. Brogioli, F. La Mantia and R. Trócoli, Adv. Mater., 2020, 32.
4. J. Neumann, M. Petranikova, M. Meeus, J. D. Gamarra, R. Younesi, M. Winter and S. Nowak, Adv. Energy Mater., 2022, 12.
5. T. Yang, D. Luo, A. Yu and Z. Chen, Adv. Mater., 2023, 35.
6. X. F. Liang, W. L. Zhou, R. Yang, D. D. Zhang, H. Wang, Q. Z. Li, Z. Y. Qi, Y. Z. Li and W. Lin, J. Environ. Manage., 2023, 348.
7. X. Y. Huang, Y. Yang, J. L. Zhao, Y. Huang and X. H. Wang, J. Alloys Compd., 2023, 947.
8. Y. Z. Zhang, B. R. Wang, F. Wang, Y. S. Dai, S. H. Ren, Y. C. Hou and W. Z. Wu, Sep. Purif. Technol., 2025, 354.
9. X. He, K. Gu, T. Kang, L. Li and Q. Li, Chin. J. Nonferrous Met., 2024, 34, 4079–4089.
10. Y. X. Lin, T. S. Wang, C. C. Gao, X. X. Zhang, W. Yu, M. Wang, C. Yang and J. H. Zhang, Adv. Sci., 2025, 12 DOI:10.1002/advs.202504683.
11. T. El Achkar, H. Greige-Gerges and S. Fourmentin, Environ. Chem. Lett., 2021, 19, 3397–3408.
12. K. A. Omar and R. Sadeghi, J. Mol. Liq., 2022, 360.
13. H. P. Gao, D. Tran and Z. Chen, Curr. Opin. Electrochem., 2022, 31.
14. X. J. Qiu, C. Y. Wang, L. L. Xie, L. M. Zhu, X. Y. Cao and X. B. Ji, J. Power Sources, 2024, 602.
15. C. F. Guo, Y. F. Zhang, M. Y. Yao, Y. N. Cao, Q. Feng and Y. Wang, ACS Appl. Mater. Interfaces, 2022, 12,  DOI:10.1021/acsami.2c08366.
16. Y. H. Chen, Y. D. Lei, J. Wensvoort, S. Gourlie and F. Wania, Environ. Sci. Technol., 2022, 8,  DOI:10.1021/acs.est.2c00310.
17. X. S. Huo, J. L. Wang, Z. H. Cong, C. Liu, C. J. Cai, Y. X. Wang, X. Q. Zhang, C. N. Li, S. Lan and J. Niu, Adv. Funct. Mater., 2025, 35.
18. H. Z. Ju, X. T. Zhao, W. Liu, Y. H. Song, L. Liu, J. Ma, Y. Li, J. X. Wu and Z. D. Zhang, ACS Appl. Mater. Interfaces, 2021, 9,  DOI:10.1021/acsami.1c17653.
19. G. F. Wang, J. Xu, L. Y. Ran, R. L. Zhu, B. W. Ling, X. L. Liang, S. C. Kang, Y. Y. Wang, J. M. Wei, L. Y. Ma, Y. F. Zhuang, J. X. Zhu and H. P. He, Nat. Sustainability, 2023, 6, 81.
20. B. M. Wieliczka, J. A. Márquez, A. M. Bothwell, Q. Zhao, T. Moot, K. T. VanSant, A. J. Ferguson, T. Unold, D. Kuciauskas and J. M. Luther, ACS Nano, 2021 DOI:10.1021/acsnano.1c05642.
21. S. Oswald, D. Pritzl, M. Wetjen and H. A. Gasteiger, J. Electrochem. Soc., 2020, 167.
22. D. H. Long, M. G. Jeong, Y. S. Lee, W. Choi, J. K. Lee, I. H. Oh and H. G. Jung, ACS Appl. Mater. Interfaces, 2015, 7, 10250–10257.
23. Q. L. Chen, H. L. Liu, J. X. Hu, L. Wang, Y. Li and Y. C. Yao, Ionics, 2023, 29, 4537–4545.
24. J. L. Zhang, N. Nie, Y. Y. Liu, J. Wang, F. Yu, J. J. Gu and W. Li, ACS Appl. Mater. Interfaces, 2015, 7, 20134–20143.
25. Y. Ke, J. S. Guo, D. C. Kong, J. M. Wang, G. Kusch, C. Lin, D. W. Liu, Z. Y. Kuang, D. M. Qian, F. Y. Zhou, G. B. Zhang, M. L. Niu, Y. Cao, R. A. Oliver, D. X. Dai, Y. Z. Jin, N. N. Wang, W. Huang and J. P. Wang, Adv. Mater., 2024, 36.
26. R. M. Zhu, W. K. Han, Y. Liu, J. X. Fu, J. D. Feng, J. F. Zhang, H. Pang and Z. G. Gu, Chem. Eng. J., 2024, 483.
27. W. X. Guan, C. X. Lei, Y. H. Guo, W. Shi and G. H. Yu, Adv. Mater., 2024, 36.
28. R. Xu, S. Y. Lei, T. Y. Wang, C. X. Yi, W. Sun and Y. Yang, Waste Manage., 2023, 167, 135–140.
29. H. T. Qiao, B. W. Zhao, X. D. Suo, X. M. Xie, L. F. Dang, J. Yang and B. W. Zhang, Global Challenges, 2022, 6.
30. Z. R. Wang, Y. Zhao, Y. X. Hou, G. H. Tang, R. F. Zhang, Y. L. Yang, X. Y. Yan and K. L. Fan, Adv. Mater., 2024, 36.
31. H. B. Shu, X. Y. Wang, Q. Wu, B. N. Hu, X. K. Yang, Q. L. Wei, Q. Q. Liang, Y. S. Bai, M. Zhou, C. Wu, M. F. Chen, A. W. Wang and L. L. Jiang, J. Power Sources, 2013, 237, 149–155.

---