In the rapidly evolving landscape of energy storage, the demand for high-performance and consistent lithium-ion batteries has never been greater. As a researcher deeply involved in battery technology, I have observed that the application of lithium-ion batteries in electric vehicles, commercial transport, and large-scale energy storage systems places immense emphasis on cell-to-cell uniformity. The performance of battery packs, often comprising hundreds or thousands of individual cells, is critically dependent on this consistency. A primary factor undermining this uniformity is the inadvertent introduction of impurities during raw material processing or cell manufacturing. Metallic contaminants such as iron, stainless steel, copper, and aluminum can be introduced from equipment surfaces or during electrode slitting processes. These impurities can catalyze deleterious side reactions, leading to accelerated self-discharge, capacity fade, and in severe cases, internal short circuits that compromise safety. Therefore, in this comprehensive study, I aimed to systematically investigate the effects of specific metallic impurities—copper and stainless steel—on the self-discharge characteristics of nickel-cobalt-manganese (NCM) ternary cathode lithium-ion batteries. My goal was not only to quantify the impact but also to develop a predictive model for impurity-induced failure, thereby informing better manufacturing controls and aging protocols for lithium-ion battery production.
The core of my experimental work involved the fabrication of pouch-type lithium-ion batteries with a nominal capacity of 0.24 Ah. The positive electrode was based on LiNi0.6Co0.2Mn0.2O2 (NCM622) active material. The electrode slurry was formulated with a mass ratio of 96.5% NCM622, 1.5% conductive carbon black (Super P), and 2.0% polyvinylidene fluoride (PVDF) binder. This was coated onto a 12 μm aluminum foil current collector using a multi-zone drying process. The resulting cathode film had a loading corresponding to a density of 2.45 mg/cm³ after calendering. The negative electrode consisted of artificial graphite (95.6%), Super P (1.0%), sodium carboxymethyl cellulose (CMC, 1.4%), and styrene-butadiene rubber (SBR, 2.0%) coated onto an 8 μm copper foil, calendered to 1.62 mg/cm³. A ceramic-coated polypropylene separator (14 μm PP + 2 μm ceramic) was used. The electrolyte was a 1.2 mol/L LiPF6 solution in a blend of ethylene carbonate (EC), diethyl carbonate (DEC), ethyl methyl carbonate (EMC), dimethyl carbonate (DMC), and propylene carbonate (PC) (25:20:45:5:5 by mass), with additives including vinylene carbonate (VC), propylene sulfite (PS), lithium difluorophosphate (LiPO2F2), and ethylene sulfate (DTD).
To simulate contamination, I introduced metallic impurities between the cathode and the separator. Two types of impurities were selected: pure copper particles and 316 stainless steel particles. The copper particles had a nominal diameter of (120 ± 20) μm. The stainless steel particles were separated into two size groups: “large” with (100 ± 20) μm and “small” with (80 ± 20) μm. For each test cell, exactly 20 particles were meticulously placed on a single cathode sheet with a spacing greater than 5 mm between them. A glass plate was used to apply uniform pressure (approximately 800 N for 5 cycles) to embed the particles into the electrode surface, ensuring they would not dislodge during subsequent assembly. Cells without any added impurities served as the baseline or control group. The experimental matrix was designed to evaluate the interaction between impurity type and formation voltage, a critical step in lithium-ion battery activation. The detailed test matrix is summarized in Table 1.
| Group | Formation Cut-off Voltage (V) | Impurity Type | Number of Cells |
|---|---|---|---|
| 1 | 3.80 | None (Blank) | 5 |
| Copper (120±20 μm) | 5 | ||
| Stainless Steel – Large (100±20 μm) | 5 | ||
| Stainless Steel – Small (80±20 μm) | 5 | ||
| 2 | 3.90 | None (Blank) | 5 |
| Copper (120±20 μm) | 5 | ||
| Stainless Steel – Large (100±20 μm) | 5 | ||
| Stainless Steel – Small (80±20 μm) | 5 | ||
| 3 | 4.05 | None (Blank) | 5 |
| Copper (120±20 μm) | 5 | ||
| Stainless Steel – Large (100±20 μm) | 5 | ||
| Stainless Steel – Small (80±20 μm) | 5 | ||
| 4 | 4.20 | None (Blank) | 5 |
| Copper (120±20 μm) | 5 | ||
| Stainless Steel – Large (100±20 μm) | 5 | ||
| Stainless Steel – Small (80±20 μm) | 5 |
All assembled lithium-ion batteries underwent a standardized testing protocol. First, a pre-formation step was conducted: after a 5-minute rest, a constant current (CC) charge at 0.10C rate to 3.50 V, followed by another CC charge at 0.20C to 4.20 V, with a 120-minute time limit. Voltage-time profiles were recorded to analyze initial electrochemical behavior. Subsequently, the main formation step was performed: a 5-minute rest, then a CC charge at 0.20C to the specified cut-off voltage (3.80 V, 3.90 V, 4.05 V, or 4.20 V), followed by a constant voltage (CV) hold until the current decayed to 0.02C or a 400-minute limit was reached. Finally, cells were subjected to a high-temperature aging at 45°C for 12 days. The open-circuit voltage (OCV) of each lithium-ion battery was measured daily to track self-discharge, characterized by the voltage drop rate or K-value (mV/day).
The analysis of the pre-formation voltage-time curves revealed distinct behaviors. For lithium-ion batteries containing stainless steel impurities, the curves were nearly indistinguishable from those of the blank cells. The total pre-formation time showed no statistically significant increase. In contrast, lithium-ion batteries contaminated with copper particles exhibited a 9.1% longer total charging time during pre-formation compared to the blank group. This initial finding suggests that the electrochemical oxidation of copper begins during this early stage, consuming additional charge. The more active components of stainless steel (iron, chromium, nickel) likely dissolve and react rapidly once their oxidation potentials are reached, completing the dissolution-migration-deposition cycle quickly without extending the charging time noticeably.
The formation step, targeting different upper cut-off voltages, provided deeper insights. The formation curves for cells with stainless steel impurities again overlapped closely with the blank cells across all voltages. However, for copper-contaminated lithium-ion batteries, the total formation time was substantially longer. Specifically, compared to the blank group, the increase in total formation time was 63% at 3.80 V, 66% at 3.90 V, 36% at 4.05 V, and 31% at 4.20 V. This voltage-dependent trend indicates that copper dissolution and subsequent reactions are ongoing processes that consume significant capacity. The relatively smaller percentage increase at higher voltages (4.05 V and 4.20 V) might be attributed to a higher degree of initial copper dissolution during pre-formation or the onset of internal micro-shorts that bypass some charge. The stark difference between copper and stainless steel underscores the critical role of the metal’s electrochemical potential within the operating window of a lithium-ion battery.
The most critical data emerged from the high-temperature aging study, which directly probed self-discharge. All lithium-ion batteries containing copper impurities, regardless of the formation voltage, developed a full internal short circuit within the first 24 hours of aging. Their voltage dropped rapidly to near zero, indicating a severe failure mode. For lithium-ion batteries with stainless steel impurities, the behavior was more nuanced and voltage-dependent. At the lowest formation voltage of 3.80 V, none of the cells short-circuited during the 12-day aging period. However, at 3.90 V, 4.05 V, and 4.20 V, cells began to short circuit at varying rates. The higher the formation voltage, the earlier the onset of voltage drop and the faster the self-discharge rate. The daily voltage drop (K-value) was calculated to quantify self-discharge speed. The K-value curves confirmed that for stainless steel, the self-discharge rate increased with formation voltage, while the difference between the two particle size ranges (80-120 μm) was minimal. This suggests that for the size scales studied, the particle size of stainless steel impurity is less influential than the electrochemical driving force (voltage) on the subsequent self-discharge behavior in these lithium-ion batteries.
Post-mortem analysis of disassembled cells provided visual evidence of the failure mechanism. In copper-contaminated cells, dark metallic deposits were observed on both sides of the separator, but the deposit on the anode-facing side was consistently larger in area. For stainless steel-containing cells, the number of dark spots increased with formation voltage, and smaller particles tended to produce more spots, though the count was always below 20 (the number initially added), likely due to non-uniform pressure during cell stacking. The morphology of the deposits was telling: copper deposits tended to form more cylindrical, through-thickness structures, while stainless steel deposits formed truncated cone-like shapes within the separator pores, not always fully penetrating.
Based on these observations, I developed a model to describe the metal impurity dissolution-deposition process leading to a short circuit in a lithium-ion battery. When a metallic particle is embedded in the cathode, during charging, the cathode potential rises. Once it exceeds the oxidation potential of the metal (M), the metal dissolves into ions (Mn+):
$$ M \rightarrow M^{n+} + n e^- $$
These ions migrate through the electrolyte-filled pores of the separator towards the anode. As the anode potential drops during lithiation, the metal ions are reduced and deposited back as metal within the separator pores or on the anode surface:
$$ M^{n+} + n e^- \rightarrow M $$
This deposition can grow progressively, eventually bridging the separator and creating an electronic conduction path—an internal short. The growth geometry depends on kinetics. Copper, with a higher oxidation potential, dissolves steadily, and its ions deposit in a manner that favors linear growth through the separator. Stainless steel components dissolve more readily but may deposit in a more dispersed, lateral manner initially.
To quantify this, I modeled the metal deposit within a separator pore as a truncated cone (frustum). The volume of deposited metal \( V_d \) is approximated by the volume of this frustum adjusted by the separator porosity \( \phi \). The volume of a frustum with height equal to the separator thickness \( h \), and radii \( a \) and \( b \) for the top (cathode-side) and bottom (anode-side) circles is:
$$ V_f = \frac{\pi h}{3} (a^2 + ab + b^2) $$
Thus, the metal deposit volume is:
$$ V_d = V_f \cdot \phi $$
Assuming the deposited metal originates from a spherical impurity particle of diameter \( D \), the total volume of the particle is \( V_p = \frac{\pi D^3}{6} \). The short circuit occurs when a critical volume \( V_{crit} \) (which is \( V_d \) from a fully developed bridge) has been deposited. The dissolution rate \( R \) (volume per unit time per unit surface area) can be estimated from the experimental short-circuit time \( t_{sc} \) and the initial particle surface area \( A_p = \pi D^2 \):
$$ R = \frac{V_{crit}}{A_p \cdot t_{sc}} $$
Using measurements from the largest observed deposits to represent worst-case scenarios, I calculated the apparent dissolution rates for copper and stainless steel at different formation voltages. The results are summarized in Table 2.
| Formation Voltage (V) | Impurity Type | Estimated Deposit Volume \(V_d\) (10-9 cm³) | Assumed Particle Diameter \(D\) (μm) | Short-Circuit Time \(t_{sc}\) (hours) | Dissolution Rate \(R\) (10-9 cm³ μm⁻² h⁻¹) |
|---|---|---|---|---|---|
| 3.80 | Copper | 8.5 | 120 | 24 | 1.56 |
| 3.90 | Copper | 8.7 | 120 | 24 | 1.60 |
| 4.05 | Copper | 8.6 | 120 | 24 | 1.58 |
| 4.20 | Copper | 8.4 | 120 | 24 | 1.55 |
| 3.90 | Stainless Steel | 2.1 | 100 | 168* | 0.025 |
| 4.05 | Stainless Steel | 3.5 | 100 | 120 | 0.058 |
| 4.20 | Stainless Steel | 4.8 | 100 | 96 | 0.099 |
*Estimated onset time for detectable self-discharge at 3.90V, as full short did not occur within test period.
The data shows that the dissolution rate \( R \) for copper is roughly an order of magnitude higher than for stainless steel and is relatively independent of formation voltage above 3.80 V. For stainless steel, \( R \) increases significantly with formation voltage, explaining the accelerated self-discharge at higher states of charge in these lithium-ion batteries.
This model can be inverted to estimate the minimum high-temperature aging time required to screen for impurity-induced self-discharge in lithium-ion batteries. By setting \( V_{crit} \) as the minimum deposit volume needed for a short (derived from the smallest observed bridging deposit) and using the calculated dissolution rates, one can solve for the time \( t_{min} \) required for a spherical impurity of a given minimum diameter \( D_{min} \) to cause a detectable short. The formula is:
$$ t_{min} = \frac{V_{crit}}{R \cdot \pi D_{min}^2} $$
Assuming a reasonable minimum detectable impurity size (e.g., \( D_{min} \) = 50 μm based on manufacturing control limits), and using the conservative dissolution rates from the experiments, I calculated the necessary aging durations for effective screening. The results provide practical guidance for lithium-ion battery production:
- For copper impurities: If formation is conducted at 3.80 V or above, a high-temperature aging period of approximately 4 days is necessary to reliably identify cells with critical copper contamination that will lead to short circuits.
- For stainless steel impurities: Screening at formation voltages below 3.90 V is ineffective, as the self-discharge rate is too slow to manifest within typical aging periods. For formation at 4.05 V and 4.20 V, aging periods of 7 to 8 days are required to ensure cells with stainless steel impurities above the critical size are identified through their voltage drop.
These findings have profound implications for the quality control processes in lithium-ion battery manufacturing. The study clearly demonstrates that even small metallic impurities, on the order of tens of microns, can have catastrophic effects on cell stability, particularly for copper. The difference in behavior between copper and stainless steel highlights the importance of impurity speciation, not just size detection. General magnetic separation methods might catch ferromagnetic stainless steel but would miss non-magnetic copper, underscoring the need for comprehensive material handling and cleanliness protocols. Furthermore, the formation voltage is a key process lever. While higher formation voltages might be desirable for achieving specific performance characteristics in a lithium-ion battery, they also accelerate the failure mechanism induced by certain impurities. Therefore, a balanced approach considering both performance and quality screening is essential.
In conclusion, through a detailed first-person experimental investigation and modeling effort, I have elucidated the significant impact of copper and stainless steel metallic impurities on the self-discharge and internal short-circuit behavior of NCM622 ternary cathode lithium-ion batteries. Copper impurities act rapidly, causing severe shorts soon after formation, necessitating a relatively short but critical aging screen. Stainless steel impurities induce a more voltage-dependent, gradual self-discharge, requiring longer aging times for detection, especially at higher formation cut-off voltages. The developed model, based on a truncated cone deposition geometry, provides a quantitative framework to estimate dissolution rates and minimum screening times. This work contributes valuable insights for designing more robust manufacturing and quality assurance processes for lithium-ion batteries, ultimately aiming to enhance the safety and longevity of energy storage systems that rely on these essential power sources. Future work should expand to other cathode chemistries, smaller impurity sizes, and the effects of multiple impurity particles to further refine the model and its predictive capability for real-world lithium-ion battery production challenges.