Identifying rate limitation and a guide to design of fast-charging Li-ion battery

3.1 Determination of cathode-to-anode capacity ratio

Areal capacity and cathode-to-anode (C/A) capacity ratio were first determined by cycling at 0.1 mA cm⁻² between 3.0 V and 4.2 V for a Li/NCA half-cell and between 0.005 V and 1.0 V for a Li/graphite half-cell, respectively. Voltage curves of the second cycle for the above two cells are indicated in Figure 1. It is determined that the NCA cathode and graphite anode have a 1.38 mAh cm⁻² and a 1.47 mAh cm⁻² areal capacity, respectively, leading to a 0.94 of C/A capacity ratio (corresponding to a 1.06 of negative-to-positive, N/P, capacity ratio as described elsewhere). Based on the above results, a current density of 1.30 mA cm⁻² was defined as the 1C rate for further cycling tests.

3.2 Impedance characteristic of Li-ion cells

A two-electrode Li-ion cell was examined by AC-impedance spectroscopy. A typical impedance spectrum of Li-ion cells shows two partially overlapped semicircles, followed by a slope straight line as depicted in Figure 2A. According to timescale of the impedance response to AC-signals, the intercept in the high-frequency end is ascribed to bulk resistance (R_b), the diameter of the first semicircle in high-frequency region to surface layer resistance (R_sl), the diameter of the second semicircle in medium frequency region to charge-transfer resistance (R_ct), and the slope straight line in the low-frequency end to Warburg impedance (W). In the above electric elements, the R_b measures overall resistance of the cell's hardware and electrolyte/separator, the R_sl corresponds to Ohmic resistance of the surface layer on the electrode particles or the grain boundary contact between the electrode particles, the R_ct to Faradic resistance reflecting the electrodes’ reaction kinetics, and the W relates to diffusion process of the Li⁺ ions on the electrolyte-electrode interface.20, 21 Impedances of a Li-ion cell at various SOCs were measured, and the values of R_b, R_sl, and R_ct were fitted using the equivalent circuit shown in Figure 2A and are plotted as a function of the SOC in Figure 2B. It is indicated that the R_b and R_sl are almost SOC-independent, whereas the R_ct changes vastly with the SOC. It is important to observe that the Li-ion cell's resistance is contributed predominantly by the R_ct, and that the R_ct gets much higher approaching the charging and discharging ends (ie, in the regions near 100% and 0% SOCs) as compared with those in the middle SOC region. We previously found that the R_sl in the lithium half-cell changes greatly with the SOC, and those in the Li/cathode and Li/graphite half-cells are changed inversely with respect to the SOC.20 Therefore, the apparent R_sl-SOC independence observed in Figure 2B can be attributed to a combined result of the R_sl in the NCA cathode and graphite anode. Since the R_ct reflects the electrode's reaction kinetics, there is an excellent corresponding relationship between the R_ct and the cell's differential capacity. That is, the change of R_ct with the SOC well follows the pattern of differential capacity, and a low R_ct corresponds to a high differential capacity.20, 21 This explains why the R_ct in Figure 2B gets much higher near the charging and discharging ends, where the cell's differential capacity approaches zero.

Since the R_sl and R_ct of the cathode and anode change with the SOC, impedances of the Li-ion cell, NCA cathode, and graphite anode are respectively measured in three representative SOCs of 0%, 50%, and 100%, and the results are compared in Figure 3. Regardless of the SOC, the following three phenomena are observed: (a) the Li-ion cell's impedance equals to the sum of the cathode and anode's impedances, (b) the Li-ion cell's impedance is contributed predominantly by the NCA cathode, and (c) the NCA cathode's impedance is dominated by the R_ct. At 100% SOC (Figure 3E,F), for example, the total resistance (ie, R_b + R_sl + R_ct) of the Li-ion cell, NCA cathode, and graphite anode are calculated to be 68, 62, and 5 Ω, respectively, in which the NCA cathode contributes to 91% of the cell's overall impedance. In particular, in the NCA cathode the R_ct is fitted to be 49.5 Ω, occupying ~80% of the cathode's impedance. The above results are attributed to the much sluggish Li⁺ ion mobility in the NCA host compared with that in the graphite. It was reported that Li⁺ ion diffusivity in the graphene planar direction of graphite reaches as high as 10⁻⁷-10⁻⁶ cm² s⁻¹,22 while that in the NCA is only ~10⁻¹⁰ cm² s⁻¹.23, 24 Therefore, it may be concluded that the NCA cathode is the rate limitation to fast-charging of the graphite/NCA Li-ion cell.

3.3 Fast-charging performance

After two forming cycles at 0.1 C, the Li-ion cell was tested by charging at various rates and discharging at 0.5 C between 3.0 V and 4.2 V. The effects of charging rate on capacity (or called “charge acceptance”) and polarization are depicted in Figure 4. It is estimated from Figure 4A that the Li-ion cell was able to retain a 1.05 mAh cm⁻² capacity (equating to 84% of the capacity, 1.25 mAh cm⁻², at 0.5 C) and a 99.7% coulombic efficiency even when the charging rate was increased to 10 C. More importantly, the capacity well recovered after the charging rate was returned from 10 C to 0.5 C. Figure 4B compares polarization and charge acceptance of the cell at various charging rates. It is observed that with an increase in the charging rate, the cell's polarization is dramatically increased, which consequently leads to a decrease in both the charge acceptance and the constant-current to constant-voltage (CC/CV) charging capacity ratio.

A three-electrode Li-ion cell was cycled to monitor the cell's voltage and two electrodes’ potential by charging at various rates and discharging at 0.5C with a 20 minutes rest between the charging and discharging steps. The cell's voltage and two electrodes’ potential are recorded using three independent testing channels, and the results are presented in Figure 5A. In a three-electrode cell, there is an excellent correlation as described by Equation (1) between the cell's voltage (U_cell) and the electrodes’ potential (E₊ and E₋ for the cathode and anode, respectively) regardless of the SOC and cycling conditions.14

(1)

The E₊ and E₋ are affected by polarization, which can be further written as Equations (2) and (3), respectively.

(2)

(3)

where E^°₊ and E^°₋ are the potentials at zero current, and η₊ and η₋ are the polarizations suffered in each electrode. It is surprised to find in Figure 5A that the E₋ was never declined to lower than 0 V vs Li/Li⁺ even if the charging rate was increased to 10 C. More detailed views of the U_cell, E₊ and E₋ at 5 C, 8 C, and 10 C are displayed in Figures 5B-D, from which the followings are observed: (a) the CC/CV capacity ratio dramatically decreases with an increase in the charging rate, (b) the E₋ has never declined to lower than 0 V vs Li/Li⁺, instead, it gradually goes up in the CV-charging period due to a decrease in the polarization (η₋) although the Li⁺ ions are consecutively intercalated into graphite. It should be noted that the direction of the E₋ change with time in the CV-charging period is determined by the interaction between the Li⁺ ion intercalation into graphite and the decrease in the η₋ value. The E₋ goes up only when the decrease in the η₋ value is more than a decrease in the potential caused by the Li⁺ ion intercalation. The above observations suggest that fast-charging Li-ion cell without Li plating on the graphite anode is possible when the NCA cathode and graphite anode are optimistically engineered in terms of the formulation, active material loading, C/A capacity ratio, porosity, and so on.25

Figure 6 compares polarizations of the cathode and anode at various CC-charging rates with the E₊ and E₋ being plotted on the same scale. It is observed that the polarization of the NCA cathode is much larger than that of the graphite anode, which is in good agreement with the impedance results as shown in Figure 3. In the state of 0.40 mAh/cm² (corresponding to a 30.8% SOC, as marked by the dot line in Figure 6), for example, the E₊ increases from 3.865 V to 4.115 V, leading to a 0.250 V increase, whereas the E₋ declines from 0.078 V to 0.035 V, leading to only a 0.043 V decrease when the charging rate was increased from 5 C to 10 C. Much higher polarization of the cathode than the anode enables safe fast-charging of the Li-ion batteries without Li plating on the graphite anode.

3.4 A guide to design of fast-charging Li-ion battery

While Li plating on the graphite anode during fast-charging was frequently reported and the graphite anode was considered to be the rate-limiting electrode,2, 6, 11, 12, 14 the exceptionally good fast-charging performance obtained in the present work can be attributed to the much larger impedance of the NCA cathode than the graphite anode. Other factors affecting Li plating are electrolyte (with separator) and temperature. When ionic conductivity of the electrolyte/separator is low, especially at low temperature, the large η₋ can drive the E₋ to lower than 0 V vs Li/Li⁺ and hence trigger Li plating. In view of increasing the electrolyte’ ionic conductivity and electrode's reaction kinetics, the Joule heat generated by high current in fast-charging is favorable for avoiding Li plating. This is true, especially for large-size Li-ion batteries in which self-heating effect is significant. Theoretically, the Li metal observed in the post-mortem analysis must be the dead Li or that in the area where the electrode was dried so that no micro-galvanic cells could be formed between the Li and the adjacent graphite. Nearly all Li plating phenomena previously reported were observed from either severely aged or failed Li-ion cells,2, 6-8, 10-13 in which the impedance of the graphite anode was comparable or even much higher than that of the cathode. As such, the E₋ falls to lower than 0 V vs Li/Li⁺, triggering Li plating, when the Li-ion cell is charged rapidly. However, in such a case, the cathode must be under-charged as predicted by Equation (1). That is, the E₊ is lower than the U_cell so that capacity of the cathode cannot be fully accessed if the charging process is still ended by the normal upper cutoff voltage. In addition, dissolution of transition metal (TM) ions is a common problem of the Li-ion cathode materials, and the TM dissolution aggravates with an increase in the charging cutoff voltage.26-29 In the aged or failed Li-ion cells, more or less TM ions were dissolved into the electrolyte and migrated to the graphite anode where the TM ions were either electrochemically reduced to metal or incorporated to the solid electrolyte interphase (SEI), resulting in a dramatic increase in the SEI impedance.30, 31 When this happens, the order of the impedances of the cathode and anode would be inverted, and the E₋ could easily decline to lower than 0 V vs Li/Li⁺. It was reported18 that for a graphite/LiMn₂O₄ Li-ion cell, the impedance of the graphite anode was much smaller than that of the cathode in the initial stage, however, it became much larger than that of the cathode after only 25 cycles at 55°C as a result of the Mn²⁺ dissolution and resultant incorporation to the graphite's SEI. As such, the graphite anode turned to be the rate-limiting electrode for fast-charging.

Based on the above analyses, fast-charging without Li plating on the graphite anode is possible only for health Li-ion cells, in which the cathode has much higher impedance than the graphite anode. Li plating on the graphite anode would occur only when the polarizations of the cathode and anode are comparable or when the polarization of anode is larger than that of the cathode. Unfortunately, this is often the case when the Li-ion cells get aged or approach failure. Li plating not only raises safety risk by dendritic deposition, but also depletes Li⁺ ions from the cathode by low coulombic efficiency. The latter lifts the cathode's working potential region so that the cathode can be overcharged even if the Li-ion cell is charged within the normal voltage range.12, 32 On the other hand, the TM dissolution dramatically increases the graphite anode's impedance, and it aggravates with an increase in the E₊ or temperature.26-29 Therefore, overcharging must be avoided for safe fast-charging and long cycle life. The results of this work indicate that low overall impedance with a high C/A impedance ratio is the key to enabling safe fast-charging and that overcharging is the origin for performance degradation of the cathode and Li plating on the graphite anode.