The following materials were used as purchased without further purification and stored under Ar: PVdF (average Mw ≈ 534 000 by GPC, SOLVAY), lithium metal foil (0.2 mm thick, LiPF₆ in EC/DMC (1:1 v/v, Donghwa electrolyte). NCM811 (LG Chem), NCM622 (L&F), LFP, and Super P (>99%, Alfa Aesar) were dried at 120°C in a vacuum oven for 3 days. Separator was dried at 40°C in a vacuum oven for 3 days.

Active material, Super P (conductive material), and PVdF (binding material) in NMP were mixed in the respective weight ratio 90:5:5 (for NCM) and 8:1:1 (for LFP) using a Thinky mixer (AR-100) at 2000 rpm for 20 min and then coated onto Al current collectors. After drying at 120°C in a vacuum oven for 12 h, the electrode was punched into circular discs (ϕ = 12 mm). The mass loading of NCM was 7.5–8.5 mg cm⁻² and 2.5–3.5 mg cm⁻² for LFP. CR2032 coin-type cells were assembled using a separator (ϕ = 19 mm), a NCM electrode as the working electrode, and a lithium metal foil as the counter electrode (ϕ = 16 mm). 1 M LiPF₆ in EC/DMC (1:1 v/v) 200 μm electrolyte was used for each coin cell. For the fabrication of pouch cells, electrodes with area 5 cm² and 12 cm² were used.

All electrochemical methods were performed at 25°C using a potentiostat (WBCS3000, WBCS3000Ls, WonATech) for galvanostatic methods. The cells were treated after standing at the open-circuit voltage for 12 h. Galvanostatic intermittent titration technique (GITT) was conducted with the schedule of 10 minutes charge or discharge with 18 mA g⁻¹ and 50 minutes rest. Galvanostatic charge discharge result was obtained after 1 cycle of formation with charge cut-off potential 4.3 V, and discharge cut-off potential 2.7 V. Reverse voltage mode was conducted for 1 h 30 m with 5 C rate. Cut off voltage was set to −1 V, but the potential during RV mode did not reach −1 V and exhibited −1 V < x < 0 V.

Before the being transferred for XRD and SEM, electrodes recovered with disassembling of cells were sealed in an airtight container inside the glovebox. The XRD patterns were obtained by an X-ray diffractometer (Rigaku, SmartLab) with a 9 kW Cu target at a scan rate of 1° min⁻¹ within 2θ range of 10–80°. SEM measurements were performed using a JSM-7800F Prime (JEOL).

Analyses for comparison were conducted to identify the optimal discharge method for ensuring the safety of spent battery recycling process through electrical discharge. The first step was to understand the tendency of overpotential during the discharge process and its correlation with discharging rate. It was hypothesized that the overpotential at the end of the discharge process would directly affect to voltage rebound [22]. Therefore, GITT analysis was conducted at various discharging rates to establish strategies for minimizing voltage rebound (Fig. 2A, Supplementary Fig. 1). In all experiments, the discharging rates were set based on the capacity within the reversible range of LiNi_{0 .8}Co_{0 .1}Mn_{0 .1}O₂ (NCM811), from 2.7 to 4.3 V. Fig. 2A illustrates that even under same reactions, different overpotential is observed depending on the discharging rate. Comparing GITT results at 0.1 C and 1 C, a notable increase in overall overpotential is observed when driven at 1 C. Fig. 2B depicts overpotential concerning operating voltages at 0.1, 0.5, 1, and 3 C. Overpotential generally exhibit converging to each constant value for each C-rate condition. Consequently, voltage rebound depends on the discharging rate, which indicate that achieving a low voltage rebound with a short discharge time is challenging. However, after the 600 hours (0.1 C-rate, 60 hours for 1 C-rate) in GITT experiment as shown in blue region Fig. 2A, it was observed that the tendency of overpotential and voltage rebound drastically changed under the 0 V (vs Li/Li⁺). Region under the 0 V (vs Li/Li⁺), while higher C-rate result in larger overpotential, it is observed that during rest periods, the potential converges to 0 V independent of the C-rate. We identified such phenomenon and understood that achieving a full discharge method that minimizes rebound with short discharging is over-discharging of cell to the region of reverse voltage.

To assess the effectiveness of the RV mode as a full discharge method, we conducted experiment and compared the results of discharging time and voltage rebound of three different modes (Fig. 2C and D, Fig. S3–S7). Under the 1 C condition, the time required for discharge in each mode was 3 hours and 45 minutes for CC, 16 hours and 32 minutes for CC-CV, and 5 hours and 44 minutes for the RV mode, with corresponding voltage rebounds of 1.055 V, 0.116 V, and 0.011 V. Additionally, CC-CV mode, while expected to yield lower overpotential due to decreased current during the CV process, is confirmed to require longer discharge time as CV mode gradually decreases the current. The results of CC mode and CC-CV mode showed the difficulty of achieving both short discharge times and low voltage rebounds simultaneously, highlighting the importance of the RV mode in achieving this dual purpose. In other words, RV mode is the strategy that could offset the trade-off between discharging time and voltage rebound. Degree of voltage rebound was examined initially under various conditions (Supplementary Fig. 2). Discharging the cell with high current value (5 Crate) shows significant meaning of RV mode; break the trade-off relationship unlike conventional CC mode. Also, remarkable effect was shown with 1 hour 30 minutes discharging condition. As shown in Fig. 2E, both CC and CC-CV modes exhibited increasing voltage rebound with higher discharge rates, while the RV mode maintained a small voltage rebound even when conducting high-rate discharge. Fig. 2F, representing the deviation of results after three times experiments at various rates, shows minimal variation with three modes, and further confirms the reliability of those three modes. More detailed information about statistic is in supporting information.

To verify the effect of RV mode in actual recycling processes, it was necessary to confirm the results with different structures and compositions of cathode materials, and whether it could be applied without problems on a larger scale. Therefore, NCM622 (with lower nickel content) and LFP (olivine crystal structure) was applied to confirm the feasibility of this strategy. With NCM622 (Fig. 3A), RV mode exhibits lowest voltage rebound compared to CC mode and CC-CV mode, similar to NCM811. This phenomenon appears from the similarity of structural and reaction mechanism despite differences in transition metal contents. In the case of LFP (Fig. 3B), with an olivine structure and distinct electrochemical reactions, voltage rebound was minimized with short discharging time by applying the RV mode. The results obtained from three type of cathode material are plotted in Fig. 3C, which signify that RV mode can be applicated regardless of the type of cathode. To demonstrate the commercial application of RV mode, larger sized electrodes using pouch-type cell were adopted (Fig. 3D). Regardless of the size of electrodes, the voltage rebound with the RV mode was observed significantly smaller compared to the other two modes, similar with that observed with an area of 1.13 cm² (Fig. 3E, Fig. S8). These results imply the applicability of the RV mode even in large-scale cells for commercial recycling processes of spent batteries. A spider map is presented at Fig. 3F to quantitatively compare the characteristics of each mode for electrical discharging in this study. When comparing discharge time, CC mode has the shortest discharge time, while the CC-CV mode requires the longest discharging time. Similarly, assessing one of the crucial factors, the voltage rebound after discharging was observed the highest value with CC mode, while the least with the RV mode. Since voltage rebound is directly related to safety concerns, the introduction of the RV is essential for ensuring safety. When comparing the energy efficiency and thermal stability, it is shown that RV mode exhibits slightly higher values, whereas CC mode shows lower values. These comparative findings elucidate how the effect of RV mode aids in overcoming the trade-off relationship between voltage rebound and discharging time. The detailed explanation is on Supporting Information.

As illustrated in Fig. 4, the RV mode may be considered for the expansion in various directions. First, it has been verified that it could be utilized independent of cathode type. Additionally, by conducting studies at the various scales, as exemplified by coin-type cells and pouch-type cells, it has been shown that RV mode enables the full discharge of practical larger systems. Finally, this strategy aimed at mitigating the trade-off between discharging time and voltage rebound would contribute to laying the groundwork for exploring enhanced full discharging strategies tailored to diverse battery systems.