The benefits of overcharge in Li-O 2 batteries. Figure 1b depicts the proposal of cathode regenerating by overcharge. A new Li-O2 battery consists of fresh Li anode and carbon cathode. During battery cycling, the parasitic reactions and poor rechargeability will result in side product accumulating on the cathode. When the cathode is passivated to some extent, the discharge voltage will plummet to lower than 2.0 V, a sign of battery failure4,26. The failed battery is considered useless in most previous investigations. However, at this stage, the anode is not fully utilized because of its excess amount. Disposing the failed battery will lead to safety problems owing to the active nature of lithium and result in resource waste. Therefore, the failed batteries should be recycled to meet the demand for a sustainable and green society. Nevertheless, the traditional procedures for battery recycling are complicated, time-consuming, and costly (Fig. 1a), which will increase the price of the renewed battery. In contrast, the battery regenerated by a simple overcharge process can prolong the battery life and maintain a low cost simultaneously. As can be seen from Fig. 1B, after overcharge, the passivation layer on the cathode can be removed, thus providing free space for the deposition of following discharge product and battery cycling. This overcharge strategy can be used many times until the exhaustion of active Li. This brings another benefit. If there remain residue Li in the battery, risks of firing and explosion exist during reservation and transportation because of its active nature. However, LiOH is much safer and easily recycled without considering the Li processing (Fig. 1c). At this time, the battery components realize system-level efficient use, and the failed battery is safer and easy to be collected and recycled.
Cathode passivation and overcharge to remove passivation. To confirm the assumptions above, we first verified that the cathode passivation resulted in the initial battery failure, rather than the Li anode. As shown in Fig. 2a, the newly assembled Li-O2 battery can run for 100 cycles at a current density of 44.2 µA cm2 and a fixed capacity of 44.2 µAh cm2. After the battery failed (discharge voltage dropped to 2.0 V), the cathode and anode were decoupled to pair with a fresh anode and cathode, respectively, to assemble new batteries to check their viability. From Fig. 2b we can see that the battery with the new cathode can run normally without discharge voltage drop for 10 cycles, indicating good cooperation of the new cathode and the remnant of active Li in the old anode. However, even at the 1st cycle, the battery with the used cathode and fresh Li plate can hardly work with a steep discharge voltage drop to 1.0 V (Fig. 2c). As expected, overcharging the battery at 17.7 µA cm2 for 10 h enables it to deliver a higher discharge voltage in the subsequent cycles even though the voltage decrease gradually. This performance recovery should be attributed to the overcharge induced cathode side product decomposition. Furthermore, the status of the cathode and anode in the initially failed battery was checked. The X-ray diffraction (XRD) patterns of the Li anode show strong peaks of LiOH, but Li peaks still exist (Fig. 2d). The relatively weak signals for Li are due to the upper thick LiOH shielding layer (~ 131 µm, Supplementary Fig. 1). This means that only 30% active Li is consumed after first battery failure and the residue Li could be a safety threat if not being handled properly during battery recycling procedures. For the cathode, the pristine CNTs with clean surface are totally covered by thick side products (Supplementary Fig. 2), and these side products are in amorphous states since there are no characteristic peaks from Li2O2, Li2CO3, or other species (Fig. 2e). These results unambiguously identify that the reason for the initial battery failure is from the cathode passivation, supporting the assumption of reviving the failed battery by following the procedures in Fig. 1b.
So the question is what causes the side product accumulation on the cathode. To check the reason, the cathode morphologies after discharge and charge for 0.1, 0.2, 0.4, and 0.8 mAh (Supplementary Fig. 3) at 0.05 mA were examined. When the fixed capacity is small (0.1 or 0.2 mAh), the cathode structure can be recovered and most of the products can be removed after charge (Supplementary Fig. 3a,b). Even though the undecomposed product is negligible in these small capacities, it can be a tough problem in high-capacity cycling. When the capacity is increased to 0.4 mAh, a passivation layer can be observed on the cathode after charge (Supplementary Fig. 3c). Further enlarging the capacity to 0.8 mAh, except the passivated film, particle agglomeration emerges (Supplementary Fig. 3d), indicating that the amount of side product increases with the enlargement of discharge capacity. The poor charge efficiency of Li-O2 batteries can also be reflected on the discharge and charge profiles. In Supplementary Fig. 4a, when the charge capacity reached the discharge value (0.39 mAh), the voltage just experiences a slight increase and follows a quick rise at 0.45 mAh due to the decomposition of the electrolyte. However, for the battery with a much higher discharge capacity of 2.74 mAh, there is almost no voltage change when being charged to the same capacity (Supplementary Fig. 4b). Even charged to 3.22 mAh, only a weak voltage increase emerges, implying that the reactions between 2.74–3.22 mAh are identical to those before 2.74 mAh. This means that, if cycling the battery at the capacity-limited mode and high discharge depth, more undecomposed products will accumulate on the cathode, and the poor charge efficiency will be amplified. As cycling continues, the constant accumulation of the undecomposed product will completely passivate the cathode and cause the death of the battery.
The morphology and composition evolution of the undecomposed product after different cycles were then characterized. After the 10th discharge, the discharge products cover the CNT cathode (Fig. 3a) and their composition has been confirmed by the Fourier transform infrared (FTIR) spectra with the existence of Li2O2 (500–600 cm− 1) and Li2CO3 (880 and 1400–1520 cm− 1) (Supplementary Fig. 5). Although most of the discharge products decompose and a clear CNT structure is observed during the following charge process (Fig. 3b), there still remain some undecomposed Li2O2 and Li2CO3 connecting the CNTs. After 50 cycles, more products deposited on the discharged cathode (Fig. 3c), and they almost exhibit no decomposition after the charge process (Fig. 3d), accompanied with strong peaks from Li2CO3, Li2O2, and HCOOLi (Supplementary Fig. 5), revealing the continuous accumulation of these products on the cathode. After confirming the composition of the undecomposed products, we subsequently checked the reactions happening during the overcharge process by monitoring the pressure of the battery to quantify the change of gas. Supplementary Fig. 6a gives the cycling performance of a Li-O2 battery at 0.1 mA with a fixed capacity of 0.4 mAh. At the 17th cycle, the discharge voltage drops to 2.0 V and then the battery has been charged for 12 h at 0.1 mA. The pressure change of the battery at this cycle is shown in Supplementary Fig. 6b. The discharge process shows a 2.09 e/O2 reaction, suggesting Li2O2 is the main discharge product. However, during overcharge, the coefficient is much higher than 2, reaching 4.01 e/O2, indicating severe side reactions happen, including the decomposition of previous accumulated products and the high voltage (> 4.5 V) induced electrolyte degradation. Besides, the coefficient is a constant value during the whole overcharge process, suggesting similar electrochemical reactions continuously happen. After overcharge, the battery can work normally with a discharge plateau around 2.6 V (Supplementary Fig. 6b). Since we have demonstrated that the initial battery failure is caused by the cathode passivation, the revival of the battery can be attributed to the removal of the accumulated products on the cathode by overcharge, which is visualized in Supplementary Fig. 7. After 15 cycles, the cathode is covered by a passivation layer (Supplementary Fig. 7a). It is clear that overcharging the battery for 4 hours can remove most of the previously deposited products, leaving only minor aggregates (Supplementary Fig. 7b). When the overcharge time is extended to 8 h, the deposited products can be completely removed (Supplementary Fig. 7c), recovering the cathode to the fresh state with exposed active sites and ample free space. Therefore, the impedance of battery decreases with longer overcharge time (Supplementary Fig. 8).
The above discussions have verified that the gradual increase of the passivation product with prolonging the cycling number is due to the negative impact of the poor charge efficiency of Li-O2 batteries. To quantify the amount of the accumulated products after different cycles, acid-base titration measurement was conducted. As shown in Fig. 3e, the Li-containing products increase as cycling goes on even though the increasing trend is not linear. This has confirmed the residue products accumulation on cathodes during cycling. The EIS results in Fig. 3f also show a higher impedance at the 10th cycle than the 1st cycle. When the battery failed at the 85th cycle, the impedance further increases and is much higher than the initial value owing to the significant amount of undecomposed products accumulated on the cathode. However, after overcharging for 10 hours, the impedance recovers to a normal value and the battery can be further cycled for another 14 cycles. It should be noted that the strategy of reviving the battery by overcharging can be done many times until the active Li is used up.
Overcharge realizing long-life Li-O 2 batteries. Now that we have demonstrated that overcharge can indeed remove undecomposed products on the cathode and revive the battery for further cycling, the long cycling performance of Li-O2 battery with overcharge deserves testing to fully take its advantage. As shown in Fig. 4a,b, the Li-O2 battery fails after 180 cycles, and a 4-hour overcharge enables the battery to work normally for extra 10 cycles before the next voltage dropping to 2 V at 387 h (Fig. 4b). Every time the battery fails, overcharge is conducted and thus repeating many times. It is necessary to mention that, for Li-O2 battery, the electrolyte is constantly consumed during cycling because of side reactions and volatilization. Adding electrolyte to the battery can no doubt help extend the battery life as well, and this is possible due to the semi-open structure of Li-O2 battery and can be easily achieved without the requirement of battery disassembly like LIBs. Therefore, when overcharge cannot revive the battery, about 150 µL electrolyte was added (Fig. 4c). After electrolyte wetting, the battery can run again and the discharge voltage plateau recovers. In the following cycles, 4-hour-overcharge was adopted when the battery failed. After 624 cycles (1320 h), the anode was exhausted (Fig. 4f) and a new anode was coupled with the original cathode to assemble a battery. Surprisingly, this battery could keep running with a stable discharge plateau (Fig. 4d). The overcharge, electrolyte addition, together with refreshment of the Li anode finally enable the Li-O2 battery to deliver a long life of 1316 cycles (Fig. 4e). Seeing the cathode can be continuously regenerated by overcharge and the easy addition of electrolyte to the battery, the life of the Li-O2 battery is limited by the durability of the Li anode (Fig. 4a). Hence, adopting Li protection strategies can further improve the battery performance. In our previous work, we verified that CO2 could stabilize the Li-O2 batteries by forming a protective Li2CO3 layer on the Li anode and capturing superoxide radical (O2−•)27. Furthermore, a Li-O2/CO2 battery was constructed to take full advantage of the cathode and anode and further extend the battery life (Supplementary Fig. 9a). The Li-O2/CO2 battery first fails after 92 cycles, followed by a 4-hour overcharge, and then it works normally again (Supplementary Fig. 9b). As expected, the integration of overcharge and electrolyte addition eventually makes the Li-O2/CO2 battery realize a record-high lifetime, reaching 2714 cycles (Supplementary Fig. 9 and Fig. 5a, ~ 270 days). The stabilization effect on Li anode can be proved by the clean surface after 626 cycles (Supplementary Fig. 10a). However, the ultimate failure of the Li-O2/CO2 battery is resulted from the Li anode pulverization and exhaustion because of this ultra-long cycling (Supplementary Fig. 10b), thus we believe the battery could live longer if a better anode protection strategy can be developed in the future.
The cumulative capacities of the Li-O2 battery and Li-O2/CO2 battery are 9.5 and 4.6 mAh respectively before the first failure. With overcharge, the cumulative capacities are enlarged to 65.85 mAh for Li-O2 battery and 135.7 mAh for Li-O2/CO2 battery (Fig. 5b). This great improvement should be ascribed to the effective revival of the degraded cathode by overcharge. Furthermore, we conducted battery cycling tests at full discharge state (discharge to 2.0 V) and checked the capacity retention by overcharging the battery with additional 1/10, 1/5, and 1/3 discharge capacity (Supplementary Fig. 11). The battery without overcharge displays the poorest capacity retention, only ~ 20% of the initial capacity can be delivered at the 6th cycle (Fig. 5c). For the overcharged batteries, the two with 1/10 and 1/5 overcharge have comparable capacity retention rate, about 40% after 20 cycles, while only 14.56% and 6.35% remain for the batteries without overcharge and with 1/3 overcharge. The total capacities of the four batteries delivered in the first 20 cycles are 4.54, 11.60, 11.19, 6.99 times their first discharge capacities (Supplementary Fig. 11e), respectively, meaning that the degree of overcharge is not the higher, the better, but should be rationally controlled.
We further checked whether overcharge could prolong the life of Li-O2 batteries with a redox mediator. A Li-O2 battery with 50 mM LiI in the electrolyte was tested at 200 mA/g with a capacity of 2000 mAh/g. As shown in Supplementary Fig. 12, just after 15 cycles, the battery fails. After overcharge, the battery can run another 5 cycles. Then overcharging again, the battery runs for more than 56 cycles. The final life of the battery is 88 cycles, ~ 5 times longer than the initial 15 cycles.
Overcharge in Li-O 2 batteries with high mass loading. We acknowledge that the development of Li-air batteries is still far from extensive application because of many unsolved problems, including the small active material loading on cathode. In the future, mass loading should be increased from submilligram to milligram level to boost capacity and cycling life. To prove the potential of this strategy in real-world application, Li-O2 batteries with 5 mg Ru/CNT on carbon paper (1 cm2) as cathodes were tested. After battery failure, the surface of Ru/CNT was covered by residue products (Fig. 6a). After overcharge, the residue was decomposed, leaving uncovered CNT structures clearly, which proves that the overcharge can remove the excessive products even in cathodes with such a high mass loading. Then the battery cycling performance was checked at different conditions. Figure 6c exhibits the performance at 0.5 mA cm− 2 with fixed capacity of 0.2 mAh cm− 2. After 47 cycles, the discharge voltage dropped to below 2.0 V, indicating the first failure. However, the battery revived by charging to 0.4 mAh cm− 2, doubling the normal charge capacity. By doing so, the battery can be revived more than 60 times, and the battery life was extended to 907 cycles, far exceeding the initial 47 cycles. We further lifting the current density to 1 mA cm− 2 with the same capacity. Similar to the battery in Fig. 6c, this battery failed after a short time (24 cycles). Then the battery was overcharged repeatedly to prolong its life. Surprisingly, 900 cycles ware achieved at such a high current density (Fig. 6d). In addition, batteries with higher capacities (0.5 mAh cm− 2 and 1 mAh cm− 2) are checked in Supplementary Fig. 13. The batteries cycled at 0.5 mA cm− 2 and 0.5 mAh cm− 2, 0.2 mA cm− 2 and 1 mAh cm− 2 achieved 325 cycles and 91 cycles respectively, far exceeding the original cycles. Given that high loading is less used in Li-O2 batteries, a list has been made to compare the battery performances (Table S1).
We note that no extra electrolytes were added during the long cycling and even the cathode loading increased to 5 mg, the amount of electrolyte added during battery assembly was kept at 130 µL despite the thicker cathode (419 µm). One abnormal phenomenon has attracted our attention, at the later stage of the cycling, each overcharge can enable more cycles (25000–45000 min in Fig. 6c and 13500–22000 min in Fig. 6d). In addition to the solvent evaporation, the decompositions of electrolyte and Li2O2 are competing reactions and proceed simultaneously during charge. When the batteries were disassembled, the electrolyte was totally drained. We anticipate that electrolyte has dried up before the final failure and the decomposed electrolyte has transformed to solid-state electrolyte (SSE) after decomposition to sustain the battery life. The formed SSE was more stable than the liquid, thus the decomposition of Li2O2 prevailed. Then the accumulation of product was alleviated and the battery life extended naturally
General applicability of overcharge in metal-O 2 batteries. The above batteries were all based on CNTs cathodes. To check whether overcharge is also effective in ordinary carbons to improve the performance, Super P cathode was used as an example to configure a Li-O2 battery. The battery only delivers a life of 31 cycles in traditional evaluation criterion and the life can be extended to 98 cycles with the help of two times of overcharge (Supplementary Fig. 14), proving that overcharge is a general method to regenerate the carbon-based cathodes. We anticipate carbon-free cathodes can further improve the performance due to the decrease of side reactions.
Moreover, overcharge is not only effective in Li-O2 batteries but also applicable in Na- and K-O2 batteries. For the Na-O2 battery, the first failure happens after 73 cycles, and overcharging twice enables the battery to work for 60 more cycles (Supplementary Fig. 15a). The thick solid electrolyte interphase (SEI) formed on the Na anode may be the reason of the large overpotential (Supplementary Fig. 15b). The K-O2 battery is very unstable with discharge voltage dropping to 2.0 V only after 8 cycles. Since there are limited undecomposed products deposited on the cathode with this short life, the battery failure is presumably caused by the anode side. After overcharge, the battery revives with much longer life and higher overpotential, probably due to the overcharge induced electrolyte decomposition that enables the formation of a stable SEI on the K anode. In the following battery failure, the function of overcharge is to decompose the accumulated products on cathode. Finally, the battery runs for 123 cycles, much longer than the initial 9 cycles without overcharge (Supplementary Fig. 15c). After disassembling the battery, we find that there is even thick SEI on the backside of the K anode (Supplementary Fig. 15d), resulting in a higher overpotential.