[Technology] Joule: Challenging Traditional Slow Manufacturing to Improve Battery Cycle Life Through Rapid Manufacturing

[Research Background]

Formation is a critical step in the manufacturing of lithium-ion batteries. During the formation process, the electrolyte is reduced at the anode, resulting in the formation of a Solid Electrolyte Interphase (SEI) layer. To ensure the stability of the SEI layer, traditional formation processes are typically conducted at low current densities, which are time-consuming and expensive. Optimizing the formation process is essential for balancing battery performance and manufacturing efficiency. However, the optimization of the formation process faces numerous challenges due to the complex relationship between formation parameters and battery aging mechanisms, as well as the high dimensionality of the experimental parameter space and the long feedback cycles. 

While traditional views suggest that slower formation conditions help prevent lithium plating, some studies have shown that lithium plating does not necessarily have a negative impact on cycling performance or thermal safety. Therefore, rapid formation may offer advantages. Data-driven approaches, integrating interpretable machine learning and differential voltage analysis, can effectively navigate the complex parameter space to optimize formation conditions and gain a deeper understanding of the coupling between formation and battery aging mechanisms. Analyzing different datasets not only aids in optimizing formation schemes but also enhances the understanding of battery performance and lifespan.

The electrolyte is composed of EC, EMC, and DMC in a volume ratio of 1:1:1, with 1 M LiPF6 and 2 wt% of VC added.

[Content Summary]

This paper develops a data-driven workflow for experimental design and generates mechanistic insights to guide future optimizations. A comprehensive exploration was conducted within a six-dimensional parameter space, resulting in a dataset of 186 single-crystal Li[Ni0.5Mn0.3Co0.2]O2 (SC-NMC 532) / artificial graphite (AG) soft-pack batteries formed under 62 different conditions. The effects of formation conditions on cycle life ranged from 400 to 1,300 cycles. 

Interpretable machine learning techniques, specifically SHAP analysis, were applied to establish relationships between formation conditions, electrode-specific utilization, cycle life, and resistance growth. The research not only confirms the role of the formation process in regulating SEI characteristics but also reveals how adjustments to the electrode voltage profile can influence battery cycle life. Enhanced performance at higher formation currents is attributed to the shift in electrode-specific utilization range and the optimization of SEI characteristics at elevated formation temperatures. The two mechanisms uncovered in this study can be utilized to optimize formation protocols and design the optimal operating ranges for batteries.

【Results and discussion】

                                       Figure 1. Experimental Design Grouping

To gain a deeper understanding of the voltage-dependent formation of the SEI layer, a two-step charging formation template was designed. This template systematically adjusted formation temperature (T), two-step formation charging currents (CC1 and CC2), transition voltage (Vtransition), the number of verification cycles after the first charge (nverification), and the open-circuit voltage rest time (tOCV). Previous literature has often mentioned the need for multiple formation cycles; thus, nverification was introduced to determine the number of cycles required for complete formation of the battery (Figure 1A).

A Latin hypercube sampling method was employed to generate 62 different formation conditions, ensuring wide coverage of the parameter space while minimizing correlations between parameters (Figure 1B). At a specific formation temperature, the battery was initially charged at a rate of C/5 to 1.5 V and maintained at this voltage for 24 hours to prevent corrosion of the Cu current collector and enhance the wettability of the electrolyte. The battery was then subjected to two-step charging, followed by a constant voltage (CV) hold at the charge peak (4.4 V) for 1 hour, which is crucial for forming a uniform SEI. Afterward, the battery discharged at a rate of C/5 down to 3 V. 

Following the initial formation cycle, the battery underwent an additional 5 verification cycles between 3 V and 4.4 V, omitting the CV hold and rest periods. After formation, the battery was held at 3 V for 1 hour, then degassed and resealed in an Ar glovebox. The battery was subsequently aged in a temperature-controlled room at 30 °C, which included charging at 1C to 4.4 V, followed by CV holding at C/20 and then discharging at a rate of 3C/4 to 3 V. To further investigate the battery’s low-rate capacity and internal resistance, diagnostic tests such as Reference Performance Testing (RPT) and Hybrid Pulse Power Characterization (HPPC) were conducted. The C/20 discharge data was also utilized for differential voltage analysis to estimate electrode-specific capacity and the state of charge (SOC) of the battery.

Figure 2. Discharge Capacity and Resistance Degradation of the Entire Dataset

The initial examination focused on the capacity and resistance degradation of the batteries after formation. Figures 2A and 2C illustrate the discharge capacity degradation trajectories of all batteries under low rate (C/20 RPT cycles) and high rate (3C/4 aging cycles) conditions, indicating that formation parameters significantly impact aging performance. The variation range of initial C/20 discharge capacity (0.24-0.27 Ah) is broader than that of 3C/4 discharge capacity (0.24-0.25 Ah), which is primarily driven by the formation conditions. Cycle life is defined as the point at which the discharge capacity during the 3C/4 aging cycle declines to 80% of its initial value. Figure 2D showcases the repeatability of each formation scheme. The baseline formation scheme at 40 °C with C/20 charge and discharge shows a performance improvement of approximately 110% in total energy compared to the baseline scheme. 

For batteries with shorter lifetimes, an initial low resistance is observed, but their resistance sharply increases after approximately 490 cycles, resulting in a rapid capacity decline (Figure 2B). This phenomenon suggests that early resistance measurements may not be reliable predictors of ultimate cycle life.

To further understand the impact of formation parameters on degradation, interpretable machine learning techniques (SHAP analysis) combined with a random forest model were employed to quantify the effects of formation cycle parameters on battery performance. Figure 3A depicts the relationship between cycle life and formation time, indicating that formation time can be reduced without compromising battery performance. The key influencing factors identified were formation current and temperature. Figure 3B provides a further visualization of the importance of these parameters on performance metrics. High formation temperatures (55 °C) and high rates (with CC1 and CC2 close to 1C) are defined as “fast formation” schemes. Although traditional views suggest that fast formation may induce lithium plating, potentially shortening battery life, the results of this study indicate that high formation current and temperature can extend cycle life through different mechanisms.

Figure 4. By analyzing the formation capacity and voltage difference of batteries formed below 55°C, the relationship between specific electrode utilization and cycle life was determined. Fast-formed batteries experienced significant lithium inventory loss during formation. High charge currents (CC1 and CC2) generally resulted in higher charge capacities (Figure 4A), with some fast-formed batteries showing over 20% increased charge capacity. When only one charge current was high, the capacity increase was less pronounced than when both currents were high (Figure 4B). Additionally, the increased initial charge capacity of fast-formed batteries was accompanied by a decrease of up to 7% in C/5 formation discharge capacity, indicating significant lithium inventory loss. There was an inverse relationship between formation charge capacity (Qch) and post-formation lithium inventory. The larger charge capacity of fast-formed batteries during the 1-hour CV hold might stem from lithium plating reactions. This suggests that greater lithium inventory loss caused by formation side reactions is associated with thicker SEI layers, lithium plating, increased resistance, and reduced discharge capacity, which negatively impact capacity and energy flux during cycling. Nevertheless, fast-formed batteries did not experience significant resistance increase, and their cycle life was on average approximately 50% longer than that of batteries in the baseline scheme. Low post-formation lithium inventory led to changes in specific electrode utilization. By using differential voltage analysis on the post-formation full-cell RPT C/20 discharge voltage curves, the state of charge (SOC) of the positive electrode (PE) at the bottom of discharge for fast-formed batteries was estimated. Figure 4C shows that the utilization range of the electrodes is determined by their SOC at the top of charge (4.4 V) and the bottom of discharge (3 V). Due to lithium inventory and cutoff voltage losses, only part of the theoretical voltage range of the electrodes is usable, with the gray dashed lines indicating the unused electrode voltage ranges beyond the full-cell voltage range. Figure 4D plots the relationship between post-formation PE SOC (3 V) and cycle life, showing that fast-formed batteries (blue) had a PE SOC at the bottom of discharge up to 8% lower than non-fast-formed batteries (gray). This indicates a negative correlation between PE SOC and cycle life. As further verification, full cells were disassembled at different voltages, and coin half-cells were fabricated with lithium metal electrodes to compare the individual electrode voltages of fast and slow-formed batteries. The cathodes of fast-formed full cells in the discharged state showed relatively higher voltage readings, confirming lower lithium inventory. These findings further confirm that fast formation reduces PE utilization in deep lithiation regions.

Figure 5. Causes and impacts of specific electrode utilization shifts.

During battery formation, efficiency differences between electrodes arise from distinct mechanisms. In the positive electrode (PE), lithium cannot fully re-embed during discharge, primarily due to slow reaction kinetics at high lithiation states; in the negative electrode (NE), lithium cannot be fully extracted during charging due to side reactions (such as SEI formation or lithium plating). Even without side reactions in the NE, incomplete lithium re-embedding occurs, leading to a decrease in full-cell discharge capacity. In the first cycle, the Coulombic efficiency (CE) of the NMC 532 PE (90.5%) is slightly lower than that of the AG NE (92.5%), indicating that the reversibility of the embedding reaction in NMC is lower than that in graphite. To effectively utilize Li in the PE and prevent over-lithiation of the graphite NE, the PE and NE voltage curves should ideally align when the battery is fully discharged after formation, as shown in Figure 5A. High charging currents induce more irreversible side reactions, limiting the full-cell lithium inventory and thus affecting discharge capacity (Figure 5B). Although slow and fast-formed batteries exhibit similar 3C/4 discharge capacities after formation, there are significant differences in specific electrode utilization ranges. Figures 5A and 5B show that batteries with more side reactions have lower lithiation degrees in the PE at the bottom of discharge and in the NE at the top of charge, indicating that fast-formed batteries improve the utilization ranges of both PE and NE. Changes in specific electrode utilization have two main impacts on battery performance: (1) For the positive electrode (PE), this change avoids the kinetic limitation region at high lithiation states. Since layered oxide-based PE dominates the resistance of the full cell at high lithium stoichiometry, utilization shifts can improve its kinetic performance; (2) Although the voltage change of the negative electrode (NE) at the top of full-cell charging is small, the lower lithium ion stoichiometry reached by the NE reduces the possibility of future lithium plating and continuous lithium inventory loss. Despite the reduced lithium inventory in fast-formed batteries, the difference in initial cycle aging discharge capacity between fast and slow-formed batteries at 3C/4 rate is less than 1%, but the total energy throughput is overall improved. Slow-formed batteries exhibit higher discharge capacity at low rates (C/20) because their positive electrodes can enter the kinetic limitation region at high lithiation states (Figure 5C). However, at high rates (3C/4), both types of formed batteries face high overpotentials, making the kinetic limitation region unreachable, thus minimizing discharge capacity differences. A narrower cycling voltage window may make the benefits of fast-formed batteries in cycle life less obvious but will maintain higher rate performance. Unlike adjusting battery design parameters, the significant changes in specific electrode utilization of fast-formed batteries are mainly achieved through more side reactions. Under high-temperature effects, the first charge capacity of fast-formed batteries decreases with temperature, while that of slow-formed batteries increases with temperature, which is related to lithium plating. Visual inspection of the electrodes shows that fast-formed batteries experience substantial lithium plating. Although lithium inventory loss gradually occurs during aging, accompanied by resistance growth and other degradations, changes in specific electrode utilization result in less resistance growth in subsequent cycles (Figure 5D).

Figure 6. Plots of PE SOC shift, resistance, and cycle life as functions of formation temperature.

Additionally, due to improvements in the composition and structure of the SEI layer, high-temperature formation can enhance battery cycling stability (Figures 6A and 6C). Unlike fast formation, high-temperature formation improves cycle life by enhancing SEI stability and overall battery performance rather than through adjustments in electrode utilization (Figure 6B). These findings indicate that high temperature and fast formation cannot be combined to achieve optimal performance, as high temperatures inhibit lithium plating—the very mechanism through which fast formation induces changes in electrode utilization.

【Conclusions 】

This study developed a data-driven workflow for designing, generating, and analyzing a dataset of 186 SC-NMC532/AG batteries evaluated under different formation conditions but identical cycle aging conditions. We systematically explored a broad parameter space to investigate the impact of these parameters on battery performance. It was found that high polarization charge currents and temperature are key factors affecting battery cycle life. Although high charge currents induce significant side reactions, they improve cycle life by up to 70% through altering the specific electrode utilization range. This shift avoids kinetic limitation regions on the positive electrode (PE) and reduces the likelihood of lithium plating on the negative electrode (NE), thereby enhancing cycle life.

——End——

Canrd Brief Introduce：Canrd use high battery R&D technology(core members are from CATL) and strong Chinese supply chain to help many foreign companies with fast R&D. 

We provide lab materials, electrodes, custom dry cells, material evaluation, perfomance and test, coin/pouch/cylindrical cell equipment line, and other R&D services.

Email:Eden@canrd.com

Phone/Wechat/WhatsApp/Skype:+86 13713092381

Website : www.canrud.com