Controllable Cell Structure for LPF Fast Charging.

The key idea of LPF fast charging is to charge a cell always above a temperature that can prevent lithium plating, hereafter referred to as LPF temperature (T_LPF). As illustrated in Fig. 1 C–E, a rapid internal heating step (Fig. 1D) is added before the charging step (Fig. 1E) to make sure that the cell is charged at a temperature above T_LPF.

Rapid heating is essential for LPF fast charging, as the overall charge time including heating is limited to 10 min to 15 min. Conventional battery heating methods using external heating devices or thermal management systems are restricted by the intrinsic conflict between heating speed and uniformity (i.e., high heating rate leads to nonuniform temperature and localized overheating near the cell surface), as detailed in ref. 25; thus their heating speed is limited to ∼1 °C/min (26), meaning that heating from −20 °C to 20 °C would already take >40 min. Adding the time for charging, it is no longer in the category of fast charging. In this work, we use the self-heating LiB structure (27) which has thin nickel (Ni) foils embedded inside a cell that can create immense and uniform heating, as sketched in SI Appendix, Fig. S4. The Ni foil is an inherent component of a single cell along with electrodes and electrolyte. It serves as an internal heating element, as well as an internal temperature sensor as its electrical resistance varies linearly with temperature (SI Appendix, Fig. S5). Further, the introduction of Ni foils only adds 0.5% weight and 0.04% cost to a conventional LiB single cell.

A control strategy based on the self-heating battery structure is developed in this work, as illustrated in Fig. 1 C–E. The key to this strategy is an intelligent partition of the input current between the Ni foils (heating) and electrode materials (charging) based on cell temperature (T_cell). If T_cell < T_LPF (Fig. 1D), a constant voltage close to cell open-circuit voltage (OCV) is applied along with closing the switch between positive and activation terminals. As cell voltage ≈ OCV, all of the current from the charging source is guided to the Ni foils to generate immense internal heat without going into anode materials (no lithium plating). Once T_cell > T_LPF (Fig. 1E), the switch is opened to transit from the heating mode to charging mode, with the current fed to electrode materials without any risk of lithium plating.

We select 9.5-Ah graphite/NMC622 pouch cells for a demonstration of the LPF fast charging. The cells have an areal capacity of 1.85 mAh/cm² and a cell-level energy density of 170 Wh/kg. Selection of charge rate and T_LPF are based on the simulation results of Li deposition potential (LDP) in SI Appendix, Fig. S3 using a calibrated LiB model. In general, T_LPF should be the minimum temperature that can avoid lithium plating at the given charge rate. Although higher temperature is always favorable for eliminating lithium plating, it can also accelerate growth of solid electrolyte interphase (SEI). In this work, a charge rate of 3.5 C and T_LPF of ∼25 °C are selected based on SI Appendix, Fig. S3C.

Fig. 2 shows the overall LPF fast charging process of the 9.5-Ah cell from an extreme temperature of −40 °C. Before the test, the fully discharged cell was soaked in an environmental chamber at −40 °C for >12 h. To ensure that the cell was not charged (no lithium plating) in the heating step, a voltage of 3.15 V, slightly lower than OCV (∼3.2 V), was applied along with closing the switch (referring to Fig. 1D). As such, all of the input current went through the Ni foils (Fig. 2E) automatically, without going into battery materials. As the cell voltage was set to be 50 mV lower than OCV, the cell was slightly discharging in the heating step, which gradually increased to ∼0.2 C toward the end when the cell became warmed up (Fig. 2F). Nonetheless, the total discharge capacity during the heating step only amounts to 6.85 × 10⁻³ Ah or 0.072% of cell capacity and hence is negligibly inconsequential. With the huge current flowing into the Ni foils, the cell was heated up rapidly (Fig. 2C). Once surface temperature reached 20 °C, the switch was opened to complete the heating step, and the cell then rested 10 s for relaxation of the internal temperature gradient. As shown in Fig. 2G, the Ni foil temperature, the highest temperature inside the cell, was <45 °C during heating and quickly dropped and met with surface temperature at ∼27 °C after the 10-s rest period, indicating that there is no safety concern regarding the rapid heating. Thereafter, the cell switched to charge mode using constant current constant voltage (CCCV) protocol at a current of 3.5 C limited by a cutoff voltage of 4.2 V until reaching 80% SOC. The whole process took 894.8 s (14.9 min), including 61.6 s of heating and 10 s of thermal relaxation.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 2.

The 15-min fast charging at −40 °C. (A–D) Evolutions of (A) cell voltage, (B) current partition between nickel (Ni) foils and cell, (C) surface temperature, and (D) SOC. The cell was initially at 0% SOC and −40 °C, with an OCV of ∼3.2 V. The whole charging process was divided into a rapid internal heating step, followed by a 10-s rest and then CCCV charging (3.5 C, 4.2 V) until reaching 80% SOC. (E and F) Intelligent control of current partition between Ni foils and electrode materials in the heating process. (E) All input current goes into the Ni foils and (F) negligible current goes into the anode materials (no plating) in the heating step. (G) Evolutions of surface and Ni foil temperatures during the heating and relaxation steps.

For comparison, an identical baseline cell was charged without the rapid heating step using the same CCCV protocol at −40 °C (SI Appendix, Fig. S6). Due to the extremely sluggish electrochemical kinetics and electrolyte transport and hence high internal resistance, cell voltage touched the 4.2-V limit immediately upon charging (SI Appendix, Fig. S6A), and the starting current was only ∼0.2 C (SI Appendix, Fig. S6B). The charging current recovered slowly with the sluggish rise of temperature (SI Appendix, Fig. S6C) due to the limited heat generation rate. The maximum charging current was only 0.85 C, and it took 115 min to reach 80% SOC, which is 7.7× of the LPF cell.

In general, at very low temperatures, it may be possible to develop a battery to discharge a reasonable percent of capacity; however, it is virtually impossible to charge the battery at a reasonable rate. This stems from the asymmetric electrochemical kinetics of charging versus discharging prevalent in electrochemistry. On the other hand, applications usually demand a higher rate of charging for time saving. The method of heating−charging via the self-heating battery structure presented herein is able to decouple the charge and discharge processes by rapid modulation of internal temperature; thus it is able to overcome poorer electrochemical kinetics of charging than discharging for a wide array of electrochemical energy storage cells.

Unified Charging Curve Independent of Ambient Temperature.

Fig. 3 compares charging of the 9.5-Ah LPF cell at different ambient temperatures (−50 °C, −40 °C, −20 °C, and 0 °C). The testing protocol was the same in all cases: (i) fully discharging at 25 °C and then cooling down to the testing temperature; (ii) rapidly heating by applying a constant voltage of 3.15 V until surface temperature is >20 °C; (iii) 10-s relaxation; and (iv) CCCV charging (3.5 C, 4.2 V) to 80% SOC. It can be seen that the voltage curves are almost the same in all cases despite the huge difference in ambient temperature (Fig. 3A). It took 69 s to heat the cell from −50 °C to 20 °C (∼1 °C/s) and 30.2 s from 0 °C to 20 °C (0.66 °C/s). The faster heating at lower ambient temperature benefited from the reduction of Ni foil resistance with temperature (SI Appendix, Fig. S5), leading to higher heating current at lower temperature (Fig. 3C). Even in the case of −50 °C, the heating step only accounted for 7.6% of the time of the entire process. The total time to charge the cell to 80% SOC was similar in all four cases (Fig. 3B, 905.7 s at −50 °C and 863.2 s at 0 °C, ∼5% difference). Thereby, the strong restrictions of ambient temperature on charging time, as in all of today’s EVs, is totally removed with the LPF cell.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 3.

Unified charging practice independent of ambient temperature. (A) Voltage curves of the LPF cell at different ambient temperatures. In all of the tests, the cell went through a rapid heating step at 3.15 V until reaching surface temperature of >20 °C, rested for 10 s, and then was charged at a constant current of 3.5 C followed by constant voltage of 4.2 V until reaching 80% SOC. (B) Summary of heating time and total time, demonstrating that restrictions of ambient temperature on charge time are eliminated. (C and D) Evolutions of (C) current through nickel foils and (D) cell surface temperature in the rapid heating step.

Both surface and Ni foil temperatures reached ∼27 °C after the 10-s thermal relaxation (SI Appendix, Fig. S7) in all four cases, indicating that the starting point of charging is similar. Thus, the voltage curves in the subsequent CCCV charging were quite similar (SI Appendix, Fig. S8A). The slightly higher voltage at lower ambient temperature was ascribed to the larger temperature drop during charging (SI Appendix, Fig. S8B), due to the strong cooling in the environmental chamber. With better thermal insulation and management, it is reasonable to expect that the charging curve can become unified and independent of ambient temperature. A unified charge curve could greatly simplify the battery management system and enhance the accuracy of battery state estimation (SOC, state of health, etc.) and hence is extremely useful for EVs.

It should be noted that today’s EVs, in principle, can also be heated to >T_LPF before charging, using thermal management systems outside single cells; however, the intrinsically low speed of external heating (<1 °C/min) makes it impossible to tackle the problem of fast charging. Further, as automotive cells are becoming bigger and thicker to reduce the manufacturing cost, the speed of external heating should be further reduced to avoid localized overheating on the cell surface (25). Our method of inserting Ni foils enables rapid and uniform internal heating no matter how large the cell is (heating uniformity can be guaranteed by adding multiple Ni foils). The method can also be applied to other cell geometries. For instance, Ni foil can form a sleeve wrapped around the first half of a cylindrical jelly roll before the second half is wound, thus placing it right in the middle of a jelly roll for a cylindrical cell. Several examples of Ni foil designs for various types and form factors of cells can be found in ref. 28. Furthermore, the flow of current inside the cell, between the heating element and battery materials, is actively controllable, enabling seamless switching between the rapid heating mode and charging mode based on cell temperature. Even in the extreme case of −50 °C where the electrolyte already ceases to work, the LPF cell is still charged to 80% SOC in 15 min, just as at room temperature, further demonstrating its potential to make EVs truly region- and weather-independent.

Remarkable Cycle Life by Elimination of Lithium Plating.

We further demonstrate the elimination of lithium plating in the LPF cell. Charging of the LPF cell at 0 °C is compared with two baseline conventional cells with identical materials and electrodes, which were charged with the same CCCV (3.5 C, 4.2 V) protocol to 80% SOC without the heating step. One baseline cell was tested at 0 °C, and the other was tested at 25 °C. As shown in Fig. 4A, the voltage curve of the LPF cell at 0 °C after the rapid heating step almost overlapped with that of the baseline cell at 25 °C, with a very slight difference due to the difference in temperature (Fig. 4B). The baseline cell at 0 °C, however, has much higher voltage than the other two cells due to its high internal resistance. All three cells were left at open circuit after charging to 80% SOC, and the voltage curves during relaxation are compared in Fig. 4C. A clear voltage plateau is observed in the relaxation curve of the baseline cell at 0 °C, leading to a local peak in the differential voltage curve (Fig. 4D). The voltage plateau and peak of differential voltage indicate the occurrence of Li metal stripping, and thus are clear evidence that lithium plating occurred in the 3.5-C charging of the baseline cell at 0 °C. In the other two cases, cell voltage drops rapidly to a relatively stable value, indicating that no lithium plating occurred during charging.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 4.

Remarkable cycle life of the LPF cell. Comparison of baseline cells at 0 °C and 25 °C with the LPF cell at 0 °C in terms of (A) voltage and (B) surface temperature during charging and (C) voltage and (D) time derivative of voltage during cell relaxation after charging. All cells were charged with 3.5-C current limited by 4.2 V until they reached 80% SOC. The voltage plateau in C and the local peak of differential voltage in D of the baseline cell at 0 °C indicate lithium metal stripping. (E) Capacity retention vs. cycle number for an LPF cell and a baseline cell cycling with 3.5-C charging at 0 °C ambient temperature.

Eliminating lithium plating significantly boosted cycle life at low temperatures. Cycling tests were performed with 3.5-C charging to 4.2 V followed by a 2-min rest and then 1-C discharge to 2.7 V. For the LPF cell, a rapid heating step at a constant voltage of 3.4 V was performed in the beginning of each cycle and completed at T_cell > 20 °C, followed by 10-s relaxation. Cells were cooled down fully to 0 °C after the discharge step before starting the next cycle. Evolutions of voltage and temperature during cycling are given in SI Appendix, Fig. S9 (one cycle) and SI Appendix, Fig. S10 (10 cycles). Discharge capacity of each cycle is given in SI Appendix, Fig. S11. Cycling tests were paused periodically for calibration of cell capacity with reference performance test (RPT) at 25 °C (SI Appendix, Fig. S12). The measured C/3 discharge capacity in RPT was plotted against cycle number in Fig. 4E for both the baseline and LPF cells. The baseline cell lost 20% capacity in only 50 cycles, whereas the LPF cell sustained 4,500 cycles at the same capacity retention, which is a 90× boost in cycle life. Even if EV drivers perform fast charging once a day, 4,500 cycles mean 12.5 y of operation. Converting to EFCs (i.e., total capacity discharged during cycling divided by nominal capacity 9.5 Ah), 2,806 EFCs were achieved at 80% capacity retention, which is a 122× boost compared with the baseline cell (23 EFCs). Assuming 100-mile driving range per EFC (e.g., BMW i3), 2,806 EFCs indicate >280,000 miles of lifetime, far beyond the warranty of today’s ICEVs.

The above two cells in Fig. 4E are further compared with additional baseline cells, one cycled at 10 °C and one cycled at 22 °C. These two baseline cells were initially at 20% SOC and charged and discharged by a fixed amount of capacity that equals 60% SOC of fresh cell in each cycle, with CCCV (3 C, 4.2 V) charge and 1-C discharge. As cycling protocols are somewhat different, capacity retentions of these cells are plotted against EFC in Fig. 5A. We note that the cell with 3-C charging at 10 °C only lasted 317 EFCs at 80% capacity retention. Moreover, the LPF cell at 0 °C has an even longer cycle life than the baseline cell at 22 °C. The reason is twofold. First, as lithium plating is eliminated, the dominant aging mechanism is SEI growth, which depends primarily on temperature. As shown in SI Appendix, Fig. S10B, the discharge and cooling portions of the LPF cell were below 22 °C. The average temperature of the LPF cell in the 10 cycles shown in SI Appendix, Fig. S10B is 11.6 °C, far lower than the average temperature of the baseline cell (∼28 °C). As such, the SEI growth in the LPF cell was, overall, slower than that in the baseline cell. Second, the baseline cell was charged by a fixed amount of capacity in each cycle, which equaled 60% SOC of the fresh cell but became larger than 60% SOC as the cell degraded. Thus, the baseline cell was charged to higher SOC than the LPF cell (charged to 4.2 V, no constant voltage step) in the late stage of cycling. The higher SOC would also lead to faster SEI growth.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 5.

Paradigm shift of ambient temperature effects on cell aging. (A) Cycle life comparison of the LPF cell with 3.5-C charging at 0 °C with same baseline cells at different temperatures. (B) Aging rate vs. reciprocal temperature of the four cells in A. Aging rate is defined as the ratio of capacity loss (in percent) to EFC at the end of life, and is plotted in log scale. (C) Aging rate of next-generation HE cells (with thick electrode) in the literature. The optimal charging temperature of HE EV cells shifts from ∼25 °C for current PHEV cells to ∼40 °C to 50 °C.

Fig. 5B further compares the aging rate of the above four cases, which is defined as the ratio of capacity loss (in percent) to EFC at the end of life, and plotted in log scale against reciprocal temperature. For baseline cells, the logarithm of aging rate vs. 1/T can be fitted with a linear line, confirming that the aging rate of conventional LiBs follows Arrhenius law (12). The activation energy is estimated to be −1.37 eV, which is within the range reported in the literature (29). We note that the aging rate of the LPF cell at 0 °C was brought down by two orders of magnitude compared with the baseline conventional cell, and became close to that of the baseline cell at room temperature, indicating a paradigm shift of the relationship between aging rate and ambient temperature.

LPF Fast Charging of High-Energy Cells by Elevated Temperature.

For future long-range EVs, a system-level energy density of at least 225 Wh/kg is demanded, which requires cell-level energy density to be >300 Wh/kg (30). A typical approach to raise cell-level energy density is to increase the areal capacity (and thickness) of electrodes. Cells with thicker anodes, however, are more prone to lithium plating due to larger electrolyte transport resistance. Recent work (30) showed that a graphite/NMC622 pouch cell with an areal loading of 3.3 mAh/cm², ∼1.8× of the PHEV cell in this work, lost 22.5% capacity in 52 cycles of 1.5-C charge at 30 °C. A large amount of lithium metal was observed after dismantling the aged cell, indicating that lithium plating can be a serious issue in the high-energy (HE) cells even around room temperature.

A feasible approach to eliminate lithium plating in the HE cells is to further increase charging temperature. As shown in SI Appendix, Fig. S2, increasing from 25 °C to 45 °C boosts lithium intercalation kinetics by 5.6×, lithium diffusivity in graphite by 2.4×, and electrolyte conductivity by 1.4×, and hence can help alleviate lithium plating. SI Appendix, Fig. S13 shows the model-predicted LDP of an HE cell having 1.65× areal capacity and thickness of the PHEV cell in this work. We note that the maximum charge current at 25 °C without lithium plating drops from 4 C for the PHEV cell (SI Appendix, Fig. S3C) to ∼1.5 C for the HE cell (SI Appendix, Fig. S13A) due to the increased electrode thickness. If charging the cell at 45 °C, the maximum charge rate of the HE cell can be raised to 3 C. Indeed, recent studies have shown that cells with thick electrodes have a longer cycle life at 40 °C to 45 °C than at room temperature. Jossen’s group (31) reported that a graphite/LiCoO₂ cell with a 77-μm-thick anode (1.6× of present work) lost 30% capacity in 250 cycles with 1-C charge at 25 °C, but only lost 5% capacity after 400 cycles at 40 °C. Similarly, Winter’s group (32) found that the cycle life of a graphite/NMC532 cell with a 77-μm-thick anode increased from 400 cycles at 20 °C to 1,100 cycles in 45 °C at 70% capacity retention. Most recently, researchers from Samsung (20) developed an HE cell with 5-C charging capability at 60 °C.

Fig. 5C compares the aging rate of the above HE cells with the PHEV cells in this work. The aging rate of the PHEV cell at 45 °C is also added. As reported in the literature (33), cell aging is a combined effect of SEI growth and lithium plating. For the PHEV cell, 25 °C is high enough to prevent lithium plating at the charge rate of 3.5 C (SI Appendix, Fig. S3). Further increasing to 45 °C reduced the cycle life to ∼613 EFCs at 80% capacity retention due to faster SEI growth. For the HE cells, however, it is beneficial to operate at ∼40 °C to 45 °C due to the alleviation of lithium plating, which outpaces the negative impacts of faster SEI growth. Hence, operating at higher temperatures can be a promising approach to enhance the life of HE cells. In this regard, heating would be an essential step for the charging of HE cells. Given the inherently low speed of external heating, the present LPF cell holds great promise for next-generation EVs, as it can modulate cell internal temperature almost instantaneously on demand.

In broad perspectives, the scientific merit of the LPF cell described herein is that it offers a general solution to decouple charge kinetics from discharge in battery science, and to accelerate battery charging without demanding new material or chemistry. It also offers a platform for material scientists to develop more advanced battery materials without temperature concern. On the application side, the present work permanently removes the longstanding restrictions of ambient temperature on battery charging, enabling a vast array of new electronics and devices such as all-weather smartphones, outdoor robots, drones and microsatellites working at high altitudes, as well as new applications such as rescuing vehicles stranded in snow, and explorations in the space and Arctic.