Optimizing Areal Capacities through Understanding the Limitations of Lithium-Ion Electrodes

The electrodes that composed the cell were fabricated by a commercial entity for the authors of the study. The positive electrode material is LiNi_0.6Mn_0.2Co_0.2O₂ (NMC622) and the negative electrode is primarily synthetic graphite. The positive electrodes consisted of 91.5% NMC622, 4.1% polyvinylidene difluoride (PVdF) binder, and 4.4% conductive carbon additive by weight on 15 μm aluminum foil. The negative electrode was 95.7% graphite, 3.3% styrene-butadiene-rubber/carboxymethyl-cellulose (SBR/CMC) binder, 1% conductive carbon additive by weight coated on 10 μm copper foil. An electrode porosity of 35 vol% was targeted in the calendering process that followed the coating and drying of the electrodes. The electrolyte used was 1.0 M LiPF₆ in a 3:7 by weight ethylene carbonate, ethylmethyl carbonate mixture with an additional 2 wt% vinylene carbonate (BASF). The manufacturer electrolyte specification sheet reports 8 ppm water content. The electrode pairs in full cell format are identified by the anticipated areal capacity measured in half cells for the positive electrode. The specific capacity of the NMC622 material was measured at 10 mA/g to be 195 mAh/g for 1^st charge to 4.3 V vs Li and 175 mAh/g for reversible capacity between 4.3 to 3 V vs Li. The specific capacity of the synthetic graphite material was measured at 35 mA/g to be 350 mAh/g for 1^st charge (lithiation) to 0 V vs Li and 335 mAh/g for reversible capacity between 0 to 1.5 V vs Li. Table I details the physical properties of the electrode pairs and the calculated negative-to-positive capacity ratios for first charge and reversible cycling.

Five negative-positive electrode pairs of targeted capacity loadings were fabricated with the same slurry composition as noted above and were coated to have a negative-positive capacity ratio between 1.05 and 1.20 for both 1^st charge and reversible discharge capacity at C/10. Each of the electrodes fabricated were characterized in half-cells that consisted of 2032 coin cells with the positive or negative electrode of interest, 2325 Celgard separator soaked in LiPF₆ based electrolyte, and lithium metal negative electrode. Four single layer pouch cells of 14.1 cm² active cathode area were constructed from each of the NMC622/Gr electrode pairs. Prior to cell assembly, the electrodes were vacuum baked overnight at 120°C. The Cell Analysis, Modeling, and Prototyping (CAMP) Facility at Argonne National Laboratory was used to assemble the single sided laminates into pouch cells in a dry room operating below –40°C dew point. More details on assembly have been reported elsewhere.¹⁹ An electrolyte volume to pore volume ratio of approximately 5 was used for each pouch cell. The pore volume used in this ratio is the sum of the calculated pore volumes of the negative and positive electrodes after calendering, and the pore volume of the whole separator piece used in the cell (not just the section between the electrodes). In most prototype cell systems, a ratio between 2.5 and 3 is adequate, but in this study the ratio was doubled to ensure consistent electrolyte availability throughout the life of the cell. Previous experimental results suggest that ∼0.1 mL is required for wetting of the cell hardware. This additional volume was added to the calculated amount of electrolyte for the 2.2 mAh/cm² cell. Thus, the electrolyte to pore volume ratios for the lowest to highest loading electrode pairs were: 6.2, 5.0, 5.0, 5.0, and 4.8 which corresponded to 0.50, 0.59, 0.75, 0.93, and 1.0 mL of electrolyte in each cell.

After filling, the cells were subjected to a tap charge and allowed to rest for 24 hours. Thereafter, the cells were formed by two consecutive, symmetric cycles at the C/20 rate between 3 and 4.2 V. Thereafter, a rate capability test commenced to examine the capacity utilized as a function of discharge rate. The rate capability test involved 3 consecutive cycles at a rate before increasing to a higher C-rate value: C/10, C/5, C/2, 1C, and 2C discharge rate between 3 and 4.2 V. The charge rates were symmetric to discharge for C/10 and C/5 and then set to the C/3 rate for the C/2, 1C, and 2C discharges. A voltage hold trickle charged the cell at 4.2 V until the measured current was less than that corresponding to C/20. A 96-channel MACCOR Series 4000 Test Unit was used for all of the electrochemical characterization and life cycle tests in this work. All coin cells and pouch cells were tested in forced ventilation ovens operating at 30 ± 1°C. The pouch cells were mounted in a test fixture with adjustable spring pressure to have a loading near 15 kPa on each cell.

The C-rate for the initial rate capability test was based on the available areal capacity measured in half cells of the positive electrodes. Thereafter, the rate capability results were interpolated to specify an appropriate current density for a desired time of discharge or charge. The measured full-cell pouch-cell capacities deviated from the available areal capacity as shown in Table II. The root cause of the deviations from half-cell capacity values are unknown, but believed to result from consumption of lithium with electrolyte and additive in the formation cycle, delamination of the graphite electrode, mismatch of the positioning of the layers, and/or poor wetting throughout the area of the electrode. The percentage of cathode available capacity utilized in the full-cell pouch-cell ranged from 92–97% for the mean cell group values. No monotonic trend with loading was observed, Table II.

The aging protocol for the single layer pouch cells involved a diagnostic series of tests followed by 50 cycles between 3 to 4.2 V at the C/3 discharge and C/3 charge with voltage hold at 4.2 V until the current decayed to less than C/20. The diagnostic series involves a discharge to 3 V until the current decays below C/20. Then the cell is charged to 4.2 V at the C/10 rate with a trickle charge until the current is less than C/20 followed by a C/10 discharge to 3.0 V. The cell is returned to 4.2 V at the C/3 rate utilizing the 4.2 V trickle charge with C/20 cutoff and then discharged at a 1C rate. Thereafter, the cell is charged to 4.2 V as before and allowed to rest for 1 hour before a series of pulse power measurements as a function of state-of-charge is commenced. The cell is discharged to decrease the SOC by 10% at the C/3 rate and allowed to rest for 1 hour before the next pair of current pulses was commenced. The pulse power experiments consisted of 2C discharge current pulses for 10 s followed by 40 s rest and a subsequent 10s 1.5C charge. This process was repeated at 10% SOC increments until the 3.0 V cutoff. The area-specific impedance (ASI) was calculated by dividing the polarization by the magnitude of the current pulse. The polarization is taken to be the difference between the voltage just prior to the current pulse and the final measured voltage before the current was interrupted. This series was then repeated until cycle 280, when the protocol was modified to accelerate degradation.

In the simulations, the NCA/Gr cell is initially fully charged and relaxed at 4.1 volts. Constant current discharges are simulated to a cell voltage of 3.0 volts. Preliminary simulations were conducted by setting the electrode thickness and the discharge time (e.g. 1 hour for C/1) then adjusting the current till the simulated discharge time was within 1% of the set discharge time. A series of simulations were then conducted at increasing electrode thicknesses till the cell discharge capacity began to level off, as seen in Fig. 5 inset for the indicated discharge rates. In Fig. 5 inset, the discharge capacity for the thin electrodes (i.e., <100 μm) is proportional to the electrode thickness according to the positive electrode's volumetric capacity. As the ionic current penetration depth of the positive electrode is approached further increases in thickness do not result in proportionally more capacity. Finally, decreasing the discharge time (i.e. increasing C-rate) also shortens the penetration depth.

Zoom In Zoom Out Reset image size

A number of different simulations were examined for the NCA/Gr cell chemistry, varying the discharge rate, salt concentration, positive electrode tortuosity and porosity. The results are given in Fig. 5 on a dimensionless semi-log plot of capacity vs. positive electrode thickness. Symbols represent normalized discharge capacity at a constant discharge time or C-rate for different electrode thicknesses that utilize a constant set of physical properties. Using the positive electrode penetration depth for normalization, all the cases fall along the same curve. The dashed line is the least-squares best-fit of the data. The accessible capacity begins to deviate strongly from the available capacity at an electrode thickness of 1.8 times the penetration depth in simulation. Thus, the simulation methodology suggests that the proportionality factor, γ (see Eq. 10), is 1.8 for these parameters values and physics included in the model.

The drop off in utilized areal capacity is a result of concentration gradients that form within the cell during discharge. The salt concentration in the electrolyte at the indicated discharge times is shown in Fig. 6 for a C/1 discharge of thick 245 μm electrodes. A large thickness is chosen to clearly show salt depletion. During discharge, lithium ions travel from the anode to the cathode, this causes a shift in salt concentration in the opposite direction. As the discharge proceeds, for this thick electrode the concentration on the current collector side of the cathode drops to zero. As can be seen in Fig. 7, there is a corresponding drop in the electrode current distribution. It is important to note here that the full cathode is accessible to the ionic current till the salt concentration drops to zero, which can take a significant fraction of the discharge time for a thick electrode (e.g. more than ⅓ of the discharge time in Fig. 5). The time for the electrolyte to reach a pseudo-steady-state increases roughly proportional to the square of the cell thickness and for thick electrode cells it can be quite long.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

An alternate simulation method is utilized to parallel the experimental protocols used in this study. In Fig. 8, the electrode thickness is fixed and the discharge current is increased until the cell capacity starts to fall off similar to rate capability tests common throughout the battery literature. The C-rate used in Fig 8. is based on the slow C/10 capacity as the cell specific rate dependence is only known after the study is complete. Four different electrode thicknesses are examined (i.e. 70, 105, 140, and 175 μm). We remind the reader that this simulated NCA/Gr cell construction composed negative and positive electrodes of equivalent thicknesses for a balanced loading. The simulated results are compared to the correlation (i.e. Eq. 14) for four different values of γ (i.e. 0.6, 0.9, 1.2, and 1.5). Most of the simulations are best fit by the value of 0.9 that is half the value estimated from the above simulations in Fig. 5. One reason for the lower γ value in Fig. 8 is the different definition of the C-rate used here compared to Fig. 5. This difference is removed when plotted as a function of current density rather than C-rate. However, the correlation deviates from the simulation to some extent with the tendency of simulated thinner electrodes to have lower values of γ. We believe this difference to result from the interplay between the time for the electrolyte to reach a pseudo-steady-state and the total discharge time. In any case, testing the cells in this manner should yield a relatively conservative value of γ.

Zoom In Zoom Out Reset image size

While all the parameters, except γ, in Eq. 14 are estimated from electrode construction or can be found in the literature, there is certainly some variability in their value. The electrolyte transport parameter determinations are often associated with significant uncertainty. The salt diffusion coefficient in the electrolyte and the cation transference number both are a function of salt concentration. Therefore, the use of any method that imposes concentration gradients to determine these values, which is commonly done, must take care to account for this dependency. For the electrode, the simulations use the Bruggeman equation with τ = ɛ^−0.5 for the electrode tortuosity.²⁵ More recent studies, specifically on lithium-ion electrodes, suggest a larger value (i.e. approximately a factor of 2 for the NCA electrode and larger for platelet-like graphite).^29–31 As shown in Fig. 8 with the lighter colored lines, the higher tortuosity would yield lower break-over currents. Many additional differences exist between how parameters are measured or estimated and what they might actually be in a functioning electrode. For example, the transport properties of the electrolyte are commonly determined by measurement of bulk properties or that when confined within the porosity of a separator. However, porous electrodes utilize a polymeric binder which swells in the presence of electrolyte. Thus lithium transport values will change if the medium is an electrolyte swollen binder matrix. Additionally, positive electrode active material is a collection of secondary particles (∼10 um) that are composed of primary particles (∼0.1 um). The internal porosity within the particle is likely different from the bulk electrode porosity and certainly smaller in length scale. In addition, the coating and drying of the electrode may result in nonuniform porosity and tortuosity throughout the depth of the electrode. All of these parameters would suggest γ would likely be lower in an experiment than in a porous electrode theory simulation using a traditionally derived parameter set. The converse might be obtained if the cell format is large and the internal resistance results in cell heating that improves transport parameters compared to room-temperature values. The prediction of limiting current values is a challenging exercise for any electrochemical system.

Like all dimensionless analysis or analytical approximations, insight is gained through examining a simple expression at the expense of precise representation of the complex physics involved. Combing the analytical expressions with the average transport property values allows one to estimate the onset of limitations in an electrode without running a numerical simulation.

As the salt concentration in the electrolyte goes to zero so does the conductivity and some combination of migration and diffusion hindrance shuts down the reaction at the back of the electrode. The electrode concentration penetration length is essentially an estimate of what fraction of the electrode has a significantly non-zero salt concentration in the electrolyte. The direct impact of migration is taken into account by the lithium ion transference number. For these porous electrodes, there is also a migration penetration length that takes into account the electrolyte conductivity and can also come into play if it is significantly shorter than the concentration penetration length, but it would not limit the fraction of the electrode accessible to current. It would only affect the uniformity of the reaction current in the electrode.

The pouch cell rate capability at beginning of life illustrates where limiting physics may onset for an electrode pair. The voltage profile for an exemplar cell with 3.3 mAh/cm² available areal capacity is shown in Fig. 9. This cell, which utilizes ∼76 μm thick positive electrodes, appears to be limited to currents at the 1C rate or lower to maintain >85% of rated capacity. The voltage profiles for discharge rates of C/5 to C/20 are difficult to visually distinguish from one another. Cell polarization on discharge is minimal until the C/2 rate.

Zoom In Zoom Out Reset image size

A comparison of the rate capability of the entire pouch cell set is shown in Fig. 10. The average of the four cells is plotted for each cell group as a function of C-rate (Fig. 10a) or current density (Fig. 10b). The capacity for each cell group is maintained until a limiting C-rate or current density is reached. The critical C-rate is defined here as the value where the measured capacity is less than 80% of reversible capacity at the C/10 rate. The definition of a percent of rated capacity is not founded on a theoretical basis; rather, 80% of rated capacity was taken as indication of the onset of the transport limitation. In Fig. 10a, the differing electrode loadings display different critical C-rates, with the highest loadings having the lowest critical C-rate. This observation suggests that transport within the intercalation materials is not the primary limitation of cell performance. A solid state diffusion transport limitation inside the particles should be independent of electrode thickness when compared at constant C-rate, where a single particle would see the same current density in each case. When compared as a function of current density, Fig. 10b, the capacity for each cell appears to reach limiting behaviors near a similar current density. This observation suggests that transport within the electrolyte is the limiting factor. As shown above in numerical simulations, concentration gradients form within the electrodes that lead to non-uniform current distributions and underutilization of the electrode. As the current density shifts toward the separator/electrode boundary, the flux into the particles in this region increases to carry the total applied current. As the lithium concentration builds up in particles near the separator/electrode boundary, the electrode potential is driven to more extreme values until it reaches the cell voltage cutoff threshold. Thus transport limitations in the electrolyte result in the underutilization of the capacity of the porous electrode.

Zoom In Zoom Out Reset image size

This study has focused on utilizing high electrode loadings (i.e. thick electrodes). Zheng et al. have reported a rate capability study for graphite negative electrodes coupled with LiNi_1/3Co_1/3Mn_1/3O₂ (NMC33) or LiFePO₄ (LFP) electrodes of various areal capacities.¹⁰ Their study focused largely on thinner electrodes and higher rates.¹⁰ We have transformed their results to be on a constant volumetric capacity basis (i.e. Q_A scaled with ) and plotted them on Fig. 10 for comparison. The values chosen from Zheng et al. are the capacities corresponding to the maximum areal capacity for a specific C-rate. In other words, the values are from a fitted line between points approximating a critical C-rate. We have also plotted lines of constant γ using Eq. 14. For the 1 M LiPF₆ electrolyte used in the experiments, the equations in Appendix A provide t₊⁰ = 0.46 and D = 1.6 × 10⁻⁶ cm²/s at 303 K. The capacity limitation at a critical C-rate is captured well by the physics in Eq. 14. Yu et al. previously conducted a half cell study examining electrode loading and rate capability for LFP electrodes that also supports the behavior captured in Eq. 14.⁶ They demonstrated lowering the electrolyte viscosity and increasing the electrolyte salt concentration both improved rate capability. They also reported a linear relationship between areal capacities with √t_d at the critical C-rate. Our results agree with Zheng et al. and Yu et al. that transport within the electrolyte is the primary limitation for discharging thick electrodes.

The pouch cells were subjected to cycle life testing at the C/3 rate with intermediate diagnostics after the initial rate capability measurement, Fig. 11. The cycling was interrupted on two occasions for lab maintenance. Anomalous capacities from these events were filtered out. All pouch cell groups exhibited modest decreases in capacity with cycling and increasing impedance. However, no electrode-pair cell-group cycling performance showed a dramatic loss in capacity. Very little capacity loss in each cell group is observed when the mean capacity measured at the C/10 rate from cycle 283 is compared to the mean capacity in cycle 5, Table II. All electrode cell groups retain mean capacity values higher than 96% of the cycle 5 C/10 capacity. The thickest electrode of 6.6 mAh/cm² is the only cell group whose 95% confidence intervals do not overlap between the 5^th and 283^rd cycle. Repeating the experiment with a larger cell population or more precise testing conditions would be necessary to make a more definitive statement about relative differences in rates of capacity fade.

Zoom In Zoom Out Reset image size

In the literature, differing reports of capacity retention as a function of electrode loading have been reported. Lu et al. showed only a minor effect of electrode loading on capacity retention at the C/2 rate; whereas, Zheng et al. presented dramatic capacity losses for higher electrode loadings at the C/1 rate.^8,10 Both of these studies utilized electrodes less than 110 μm in thickness. In Fig. 11, cell groups using positive electrodes ranging from 48 to 154 μm exhibit low levels of capacity fade at the C/3 rate. We suggest the difference between reported cycling performances is related to concentration gradients in the electrolyte as well as the lithium plating side reaction in the graphite electrode. Lithium plating may occur in the graphite electrode when the surface overpotential is < 0 V vs Li/Li⁺. Detecting lithium plating in situ is difficult, but various methods, often electrochemical, have been suggested. Irreversible capacity loss is the most commonly invoked technique.

To test for the onset of lithium plating, we increased the charge rate of each cell group to the C/1 rate (determined here by 3× the C/3 rate) starting on the 286^th cycle, while still maintaining the trickle charge at 4.2 V. The discharge rate and diagnostic steps were maintained as for the previous 285 cycles. The C/1 rate was selected based on the rate capability plot and Eq. 14 comparison in Fig. 10. We anticipated that this charging rate would accelerate degradation in the thickest electrodes, but not in the thinnest. The capacity retention during the C/1 charging is shown in Fig. 12. The diagnostic cycles are filtered out for clarity, but pertinent results are shown in Table II. The capacity measured at the C/10 rate for cycle 335 demonstrates a large irreversible capacity loss for electrode loadings of 4.4, 5.5, and 6.6 mAh/cm², Table II. The 2.2 mAh/cm² cell group shows no statistically significant evidence of accelerated degradation. The 3.3 mAh/cm² cell group includes two cells with no accelerated degradation, but two cells that displayed a relatively modest capacity fade. This would suggest the 3.3 mAh/cm² loading is near the maximum capacity for these materials and electrode construction for operation at C/1. This is further demonstrated by increasing the charge rate to 1.5C while maintain the C/3 discharge rate at cycle 549. The 3.3 mAh/cm² cells exhibit severe capacity loss while the 2.2 mAh/cm² cells show only a slight decrease in capacity. A closer inspection reveals that one of the 2.2 mAh/cm² cells has very slight capacity fade while the other two exhibit measureable fade. This would suggest that 2.2 mAh/cm² cells constructed with higher manufacturing precision should avoid capacity fade at this condition. It would also suggest that this current density is near the maximum before lithium plating occurs for this electrode loading. We note this electrode pair had the most aggressive negative to positive capacity ratio.

Zoom In Zoom Out Reset image size

The ASI measured for each cell group is shown in Fig. 13 for each 50 cycle increment. Overall, the ASIs increase as the cycle number increases. The ASI for the 5.5 mAh/cm² cell group appears to be anomalously high as it does not follow the expected trend for electrode thickness. The 5.5 mAh/cm² pouch cell group is offset >7 Ohm cm² higher than anticipated. Full-cell coin cells assembled and tested in a similar manner show an ASI value for the 5.5 mAh/cm² electrodes between the 4.4 and 6.6 mAh/cm² values as expected. In addition, the shape of the pouch cell polarization during a 10-s current pulse suggests the higher impedance has a small time constant. Thus, we hypothesize the offset in the pouch cell is due to a tab welding or lead connection issue consistent with high-frequency resistance. We do not believe this significantly alters the conclusions drawn in this work.

Zoom In Zoom Out Reset image size

Select pouch cells from each cell group were disassembled in a dry room after a 24 hour voltage hold at 3.75 V for visual inspection at the end of cycling. Positive and negative electrodes were washed with dimethyl carbonate. As expected, cells that exhibited large capacity fade showed evidence of large quantities of lithium plating on the surface of the negative electrode, examples shown in Fig. 14. The lithium was observed to be metallic-silver in color. Small pieces of the metallic-silver deposit removed from the electrode reacted much more vigorously (e.g. spark or flame) when put in contact with large quantities of water than pieces of graphite removed from the same electrode. The best performing cell of the 2.2 mAh/cm² group showed some small quantity of lithium plating when disassembled and visualized, Fig. 14. This suggests the current density was near the tipping point for the onset of lithium plating across the entire area. Small differences in coating weight and electrolyte wetting may initiate lithium plating in some areas before others. A thickness gauge was used to measure the thickness of the electrodes before and after testing. In general, the positive electrode showed little change in thickness. Conversely, the graphite electrode showed increased thickness of 5–10 μm in regions without visual defects compared to the pristine electrodes and dramatically increased thicknesses in regions associated with lithium plating. The graphite electrode of the best performing cell of the 2.2 mAh/cm² group, Fig. 14, measured an additional 7 μm greater in thickness in lithium plating regions than those without. In cells that exhibited substantial capacity fade (cell groups 3.3, 4.4, 5.5, and 6.6 mAh/cm²), an increase of thickness of 70 μm was common between regions of the cycled graphite electrode with and without obvious lithium plating. In other words, the deposit associated with lithium plating dramatically increased the negative electrode thickness. Higher loading graphite electrodes are omitted from Fig. 14 as they appear similar to the 4.4 mAh/cm² electrode. Fully discharging (or charging) the 4.4 mAh/cm² cells did not remove the lithium deposits. This indicates that the lithium deposits are not electrically or electrochemically connected to the negative electrode. In addition to lithium plating, graphite was observed to flake off during disassembly around the edges of the punched layer. The severity of the flaking increased as the loading increased. It is unknown if this region was active during the cycling and only disconnected during disassembly. For the 4.4 mAh/cm² electrode, gold colored graphite, believed to be LiC₆, is found in the corners of the layer and is believed to have become electrically isolated at some point during the testing.

Zoom In Zoom Out Reset image size

Much more work is needed to understand the transition from stable cycling to the onset of transport limitations and side reactions such as lithium plating. Figure 15 is proposed as an early performance map that might be used for electrode design. The critical C-rate for the NMC662/Gr electrodes, γ = 0.6, is plotted along with a more conservative value of γ = 0.3. The available areal capacity for charging the cell may be approximated if one assumes that the parameters in Eq. 14 used for the positive electrode are a reasonable approximation of the negative electrode. The volumetric capacities of the positive and negative electrodes considered in this work are similar in value. We note the γ value may differ for other graphite types such as platelet shaped, common for natural graphite, which exhibits a highly anisotropic electrode tortuosity. In addition, series that represent constant polarization of the graphite electrode from the first (120 mV) and second (90 mV) plateaus are shown. An electrode ASI is required to calculate this current-voltage relationship for the graphite electrode. The ASI of the graphite electrode was estimated from half-cell measurements of 2-sec charge (lithiation) pulses, as described in Appendix B. Experimental observations of stable cycling (green symbols), onset of loss (yellow symbols) and rapid loss from assumed lithium plating (red/orange symbols) are plotted against lines of mass transport limitation and overpotential limitations for charging. Square symbols are from observations reported in this work while diamonds are from Burns et al. using high precision coulometry measurements.¹⁸ The 2.2 mAh/cm² data from Burns et al. was a NMC333/Gr pouch cell that utilized the same electrolyte as the experiments in this paper. The 4.4 mAh/cm² data from Burns et al. originated from a nominal 3.4 Ah Panasonic 18650 cell based on NCA/Gr chemistry with unknown electrolyte composition.

Zoom In Zoom Out Reset image size

A guideline may be considered that graphite cells should avoid current densities near or above 4 mA/cm² on charge unless additional precautions have been made to avoid the side reaction. The line representing a constant overpotential of 90 mV at the negative electrode correlates with the separation between stable cycling and rapid capacity loss for high areal capacities. Electrolyte transport limitations are likely important in this region as the combination of overpotentials lead to lithium plating. For low electrode areal capacities or thicknesses, this line begins to bend toward lower current densities on account of the increasing electrode ASI from diminishing interfacial active area.³² Chandrasekaran reported for a 2 mAh/cm² cell that simulations indicated conditions thermodynamically favorable to lithium plating occurred at the 2C rate (i.e., 4 mA/cm²).¹⁶ There-in, both electrolyte concentration overpotential and interfacial kinetic overpotentials contributed to the favorable conditions for lithium plating. This agrees well with our observations as well as Burns et al.¹⁸ Modifications may be made to a graphite electrode to allow for higher rate continuous or pulse charging. These might include a higher negative to positive capacity ratio (N:P ratio), including hard carbon as a physical blend, or selecting a graphite with lower interfacial impedance. Cell designers may also choose to de-rate the charging power capability of the cell when it is on the lower voltage plateau of the graphite electrode. The higher voltage plateau has comparatively lower interfacial impedance and greater overpotential before lithium plating may ensue. Understanding the onset of lithium plating in a graphite electrode requires accurate treatment of transport within the electrolyte and active phases, interfacial kinetics, and real world use of the cell. We note lower temperature operation may drive the design requirement as electrolyte transport properties will decrease in value and the interfacial kinetics for impedance have been shown to be problematic. Future work would be well served to examine the current density design space where interfacial kinetics and electrolyte transport limitations coincide as a function of temperature of operation.