Hot Formation for Improved Low Temperature Cycling of Anode-Free Lithium Metal Batteries

Dry (no electrolyte) and sealed Li[Ni_0.5Mn_0.3Co_0.2]O₂ (NMC 532) || Cu pouch cells were obtained from Li-FUN Technology (Xinma Industry Zone, Golden Dragon Road, Tianyuan District, Zhuzhou City, Hunan Province, China, 412000). The NMC used in these pouch cells was single crystal NMC with a titanium based coating.²² The single side positive electrode loading was 16.1 mg cm⁻² (50 μm) and the electrode consisted of 94% active material. Positive electrode area is 97.5 cm². Assuming the positive electrode active material has a capacity of 200 mAh/g when charged to 4.5 V, these cells have a theoretical capacity of 290 mAh at 4.5 V. The measured C/2 discharge capacity from 4.5 V to 3.6 V on the first cycle after formation is 245 mAh at 40°C or 210–220 mAh at 20°C. Cells were transferred into an argon-filled glove box, opened, and dried at 100°C for 14 hours under vacuum.

Reagents used for electrolytes include fluoroethylene carbonate (FEC, BASF, purity 99.4%), diethyl carbonate (DEC, BASF, purity >99%), lithium difluoro(oxalato)borate (LiDFOB, Capchem, 99%), and lithium tetrafluoroborate (LiBF₄, BASF, 99%). The electrolyte used for all cells tested was 0.6M LiDFOB 0.6M LiBF4 in FEC:DEC 1:2 v:v. unless otherwise specificed. Cells were filled with 0.5 mL of electrolyte (approximately 2g Ah⁻¹), resealed, and removed from the glove box. The cells were held at 1.5 V for 24 hours at room temperature to ensure complete wetting and then transferred to a temperature controlled chamber for cycling.

Anode free pouch cells were cycled between 3.6 and 4.5 V with a C/5 charge rate and C/2 discharge rate in a temperature controlled environment. The "hot formation" protocol consisted of 2 initial cycles between 3.6 and 4.5 V with a C/10 charge rate and C/2 discharge rate at 40°C. The cells cycled at low applied pressure were clamped with rubber blocks inside a plastic enclosure (Figure S1a) which maintained a uniaxial pressure of approximately 75 kPa. Cells cycled under the high pressure condition were constrained in an aluminum enclosure (Figure S1b) with a uniaxial pressure of approximately 1200 kPa.

Samples were analyzed on a Hitachi S-4700 Field Emission Scanning Electron Microscope in secondary electron imaging mode with an electron accelerating voltage of 3.0kV and working distance of approximately 12 mm. Lithium samples were rinsed with DMC and dried before being affixed to the SEM sample stub via carbon tape. The samples were transferred to the SEM in an argon-filled container and were briefly (< 10s) exposed to air during the final transfer into the SEM chamber.

Figure 1a summarizes the capacity retention of identical NMC 532 || Cu pouch cells with LiDFOB/LiBF₄ dual salt electrolyte cycling at varying temperatures. All cells were cycled with a C/5 charge rate and a C/2 discharge rate and constrained with low pressure (75 kPa) in a plastic enclosure. It's clear that the performance of these cells is a strong function of the cycling temperature. Replicates or brother cells run for each condition are represented by open and filled symbols of the same color. After 50 cycles at 40°C the anode-free cell has shown minimal fade and is still well above 80% capacity retention. However, when the temperature is reduced to 30°C this same cell falls below 80% retention at just over 30 cycles. If the temperature is reduced to 20°C, fade is further accelerated as the cell only achieves 18 cycles to 80% retention and is almost completely dead after just 50 cycles. Conventional lithium-ion cells generally demonstrate the opposite trend within this temperature range; higher temperatures lead to accelerated fade due to the increased rate of parasitic reactions that decrease coulombic efficiency. Why then does the performance of these anode free cells deteriorate so rapidly when the temperature is reduced? There are a number of factors to consider. Due to its low melting point the mechanical properties of lithium metal such as its yield stress and creep stress are strong functions of temperature. At higher temperatures lithium is softer and easier to deform so at 40°C the low pressure applied to the cells in Figure 1a may be enough to adequately constrain the lithium and delay the onset of mossy or dendritic growth. Additionally, there may be beneficial "cross-talk" reactions occurring within the cell producing "additives" that aid in lithium metal cycling; at lower temperatures the rates of these reactions would be reduced.

Zoom In Zoom Out Reset image size

To first explore the role of the mechanical properties of lithium, the effect of mechanical pressure on anode-free cell performance was evaluated at high and low temperature. Figure 1b shows capacity versus cycle number for anode-free cells cycled at 20°C and 40°C under low pressure (75 kPa) along with identical cells cycled at the same temperatures but constrained at much higher pressure (1200 kPa). At 20°C, pressure has a significant impact on performance, the number of cycles to 80% retention increases from only 18 in the low pressure case to 50 under high pressure. The longer term fade rate is much reduced at high pressure as the 1200 kPa cell retains 60% capacity after 100 cycles compared to only 10% at 75 kPa. At 40°C, however, the higher pressure has a smaller impact with the 1200 kPa cell showing only a modest improvement over the one cycled at 75 kPa. This aligns with the concept of lithium being softer at higher temperatures and therefore needing higher pressures at lower temperature in order to be adequately constrained. However, the capacity retention of the cell under high pressure at 20°C is still worse than the cell cycled at 40°C with very little pressure. Normalized capacity is plotted here for clarity, but due to the lower cell impedance at higher temperature, the 40°C cell also has a larger capacity (cycling more lithium) and still shows better capacity retention than the 20°C cell under pressure. This suggests that pressure, while important, is not the only factor influencing the low temperature performance of these cells and additional measures need to be taken to improve the 20°C cycling behavior.

Initial formation cycles have been used in lithium metal cells previously, but these formation cycles occurred at the same temperature as the subsequent cycles.²³ Due to the superior performance of our anode-free cells at higher temperature we wanted to investigate the impact of high temperature formation cycles on subsequent lower temperature cycling. Figure 2a shows the cycling behavior of dual salt anode-free cells which underwent two formation cycles at 40°C consisting of a C/10 charge and C/2 discharge followed by cycling (C/5 charge C/2 discharge) at 20°C, under low pressure. The cycling behavior of identical cells with no formation, simply C/5 charge and C/2 discharge cycling at both 20°C and 40°C under low pressure are included for comparison. Surprisingly, this simple two cycle hot formation procedure has an enormous effect on the performance of the 20°C anode free cells. The cells which underwent hot formation show a much smaller fade rate, achieving 60 cycles to 80% retention compared to only 18 cycles for the 20°C control case. The magnitude of the impact of hot formation is comparable to that of applying 1200 kPa pressure. In terms of cycles to 80% retention, the hot formation 20°C cells are still slightly worse than those cycled at 40°C, however with hot formation the cells show better longer term retention after 100 cycles. When high pressure is applied to the hot formation cells, the performance is further enhanced (Figure 2a). These high pressure hot formation cells cycled at 20°C show 85% capacity retention at 100 cycles, superior to both the low and high pressure cells cycled at 40°C. The anode free cells cycled 40°C show an obvious "knee" in the capacity retention curves around cycle 80–90 where the fade rate increases. We have previously demonstrated that this knee and subsequent failure are caused by significant salt consumption from the electrolyte during cycling.²¹ For the hot formation cells cycled at 20°C, the long term fade rate is much more stable and there are no sharp slope changes in the capacity retention curves. At room temperature the parasitic reactions consuming electrolyte salt should be slower thereby increasing cycling life and delaying cell failure. Thus, one of the benefits of lower temperature cycling (reduced rate of parasitic reactions) can now be realized when cycling is initialized with hot formation.

Zoom In Zoom Out Reset image size

In order to better understand why hot formation is improving cell performance, it is important to consider the changes occurring during hot formation. Hot formation should allow for improved initial lithium morphology due to the slower charge rate and the higher temperature at which the lithium can be more easily constrained. Additionally, reactions occurring within the cell will be accelerated during the high temperature formation compared to lower temperature cycling. For instance, the reactions generating gas within a cell are often significantly accelerated at higher temperature. Figure S2 shows the gas generation after 50 cycles at different temperatures for anode-free cells with 0.6M LiDFOB 0.6MLiBF₄ FEC:DEC 1:2 electrolyte cycled with a C/5 charge and C/2 discharge between 3.6 and 4.5 V. The gas volume was measured via the Archimedes' method described previously.²⁴ Predictably, the gas generation increases rapidly with temperature. The LiDFOB salt is known to oxidize at high voltage with CO₂ being one of the byproducts.²⁵ GC-MS analysis of the gas produced during cycling of LiDFOB/LiBF₄ anode free cells revealed CO₂ as the only quantifiable gas detected to a significant degree (Figure S3). CO₂ has been has been demonstrated as an effective additive for Si anodes and has been used as a SEI film forming additive in lithium-ion and lithium metal cells.^26,27 Thus, some of the benefit of hot formation may be from the additional CO₂ generated at the cathode and the subsequent beneficial cross-talk reactions. To explore this idea, further experiments were conducted to better quantify the impact of gas generated during formation with the resulting cycling data shown in Figure 2b.

The pink curves in Figure 2b represents results for control cells cycled (C/5 charge C/2 discharge) at 20°C. The light blue curves correspond to cells which were charged (C/5) once at 20°C, moved to a 40°C chamber and held for 8 hours to generate approximately 0.2 mL of gas, and then cycled (C/5 charge C/2 discharge) at 20°C. The production of gas after the first charge results in a noticeable but very minor improvement in capacity retention, which means gas generation alone without hot formation has minimal impact. The dark blue curves show results for cells that underwent a single C/5 hot formation cycle at 40°C and were subsequently degassed before being cycled (C/5 charge C/2 discharge) at 20°C. This single hot formation cycle, even after degassing, still has a large effect on capacity retention, more than double the number of cycles until 50% retention. This means that when isolated, hot formation has a more significant impact than gas generation. Finally, the green curves in Figure 2b corresponds to the combination of a C/5 hot formation cycle followed by an 8 hour hold to generate 0.2 mL of gas. When hot formation and gas generation are combined the best capacity retention is achieved. This is reasonable as when cycling lithium metal the SEI is constantly being partially broken and reformed during so having an additional reservoir of a helpful additive should be beneficial.

The effect of the length of the formation protocol was also investigated. The capacity retention curves for the 20°C control cells and single cycle hot formation with degas cells are reproduced in Figure 2c along with those for a two cycle (C/10 charge C/2 discharge) hot formation protocol in both the degassed and non-degassed condition. The longer two cycle hot formation protocol shows a definite benefit over the single cycle formation when comparing cell performance in the degassed condition. The slower charge may contribute to superior initial lithium morphology and the longer overall formation time could allow for more complete SEI formation. The performance improvement of the degassed two-cycle hot formation cells over the 20°C control is quite remarkable. In this condition all of the gas generated during formation is removed so it is really just the changes that occurred during those first two cycles which are having a profound impact on the subsequent cycle life, an impact that lasts beyond cycle 100. When the two cycle hot formation cells are not degassed there is an additional benefit to the cycling performance from the reservoir of CO₂. It's interesting to note that the benefit that comes from not degassing the cell seems to be less substantial with the longer formation protocol compared to the single cycle formation, again suggesting that the longer process offers a more complete formation. A two cycle formation at even higher temperature (55°C) was also attempted, but this showed no benefit over the 40°C formation (Figure S4).

As mentioned, the failure of these anode free cells cycling with LiDFOB/LiBF₄ dual salt electrolyte can be attributed to salt consumption.²¹ Therefore in order to achieve the best possible performance under current conditions we evaluated a high concentration dual salt electrolyte (1.8M LiDFOB 0.4M LiBF₄ FEC:DEC 1:2) cycling in anode free cells at 20°C following a two cycle hot formation (Figure 2d). At low pressure (75 kPa) the high concentration cells maintain achieve 95–100 cycles before falling below 80% retention. With high concentration and high pressure after the two cycle hot formation protocol the anode-free cell reaches 195 cycles to 80% retention. The absolute capacity retention and difference between average charge and discharge voltages (called ΔV here) of these cells are shown in Figure S5. With a 3.6 V discharge cutoff at 20°C these cells leave approximately 30% excess lithium on the anode after the first cycle. Taking this into consideration, close to 80% retention at 200 cycles with only 30% excess lithium is among the best performance reported for lithium metal cells with limited excess. This corresponds to an average lithium metal cycling efficiency of 99.67% over 200 cycles.

Figure 3 shows SEM images of the lithium anode at the top of charge (4.5 V) were for cells with and without hot formation. Figures 3a and 3b show the lithium anode after two cycles (C/5 Charge, C/2 Discharge) at 20°C with low (75 kPa) applied pressure. In the higher magnification image (Figure 3b) there are some smooth, nodular lithium grains 2–4 μm in diameter, but these are dispersed amongst what appears to be a more irregular shaped and porous lithium morphology. The smooth lithium grains are regions of higher contrast, while the more porous lithium appears as lighter regions. Thus, one can clearly see in the lower magnification image (Figure 3a) that these smooth high contrast regions are the minority of the sample. Conversely, the morphology of the lithium anode after two cycles of hot formation (C/10 charge; C/2 discharge at 40°C) at low (75 kPa) pressure looks much better (Figures 3c and 3d). After two cycles at the higher temperature the lithium has a smooth and flat columnar morphology with lithium grains 5–10 μm in diameter. Not only are the lithium grains larger at higher temperature, but they are also packed much more tightly leading to much less porosity. The lower magnification image (Figure 3c) shows that this columnar morphology is uniform over a larger area. It certainly appears that two cycles at higher temperature leads to superior lithium morphology and so the lithium quality after hot formation is far superior to that demonstrated after two conventional 20°C cycles.

Zoom In Zoom Out Reset image size

While the hot formation provides a better starting point, it's important to determine if this superior morphology can be retained after subsequent lower temperature cycling. Figures 3e and 3f show the lithium anode after 20 cycles (C/5 charge C/2 discharge) at 20°C in a cell that did not have hot formation. The higher magnification image (Figure 3f) displays many irregular geometries with no smooth flat grains leading to an overall porous morphology. The lower magnification image (Figure 3e) shows that the low contrast regions of poor morphology represent the majority of the sample. This poor quality high surface area lithium explains the fast capacity fade shown by the cells cycled at 20°C without formation.

Figures 3g and 3h show the lithium anode after 20 cycles (C/5 charge C/2 discharge) at 20°C for a cell that had hot formation. The difference between the morphology with and without hot formation is pretty dramatic. The low magnification image (Figure 3g) shows that the darker high contrast regions of flat lithium columns now represent the majority of the sample. At higher magnification (Figure 3h) these regions appear as large flat lithium grains >20 μm in diameter, quite different from the irregular geometries and porous structure shown after 20 cycles without formation (Figures 3e and 3f). Even with hot formation there seems to be a porous lithium morphology growing between the large grains, however at this point it represents a small portion of the sample. The SEM analysis confirms what the cycling data suggested; the benefits established during the two hot formation cycles are carried forward and have a substantial impact the quality of lithium plated on subsequent cycles and in turn on the cycling stability of the anode free cells.