This comprehensive experimental investigation delves into the thermal runaway (TR) characteristics of commercial 21700-format lithium-ion batteries (Li-ion batteries) across a full spectrum of states of charge (SOC). Understanding the intrinsic hazard profile of a single Li-ion battery cell is a fundamental prerequisite for designing safer battery packs and modules for electric vehicles and energy storage systems. While previous research has established a general correlation between SOC and thermal stability, a detailed quantitative analysis of hazard parameters—such as trigger times, mass loss, released energy, and derived explosive equivalents—for this prevalent high-energy-density cell format is essential. This study systematically quantifies these critical safety metrics, providing valuable data for thermal hazard assessment, fire safety protocols, and the development of advanced battery management and safety systems.
1. Introduction and Background
The widespread adoption of lithium-ion batteries as the premier energy storage solution for portable electronics and electric mobility is primarily driven by their high energy density, long cycle life, and declining cost. However, the significant chemical energy stored within a Li-ion battery can, under abusive conditions such as overheating, overcharging, or internal short circuits, be released in an uncontrolled exothermic cascade known as thermal runaway. This process involves the breakdown of metastable components, leading to rapid self-heating, gas generation, venting, and often fire or explosion. The severity of a thermal runaway event is not constant; it is profoundly influenced by the battery’s state of charge, which determines the amount of active lithium and the electrochemical potential of the electrodes.
The core components of a typical Li-ion battery, such as the one studied here with a Li(Ni0.8Co0.1Mn0.1)O2 (NCM811) cathode and graphite anode, undergo sequential decomposition. The exothermic reactions begin with the solid electrolyte interphase (SEI) layer decomposition on the anode, followed by reactions between the anode and electrolyte, cathode material decomposition, and electrolyte decomposition. The heat released from these reactions, if not sufficiently dissipated, accelerates further decomposition, creating a positive feedback loop. The SOC dictates the quantity of lithium intercalated in the graphite anode; a higher SOC means more lithiated graphite, which is more reactive with the electrolyte, and a higher oxidation state of the cathode materials, which are thermally less stable. Consequently, a fully charged Li-ion battery represents a significantly greater thermal hazard than a partially discharged one. Previous studies on other form factors (e.g., 18650 cells) have confirmed this trend, but detailed data on larger, higher-capacity 21700 cells, which are increasingly common in modern applications, is crucial for accurate risk modeling.
2. Materials and Experimental Methodology
2.1 Cell Specification and Pre-Testing
The test subject was a commercially available cylindrical 21700 Li-ion battery with a nominal capacity of 5.0 Ah. The key specifications of this Li-ion battery are summarized in Table 1.
| Parameter | Value |
|---|---|
| Dimensions (Diameter × Height) | 21.0 mm × 70.0 mm |
| Mass | Approx. 69.0 g |
| Nominal Capacity | 5.0 Ah |
| Nominal Voltage | 3.65 V |
| Charge / Discharge Cut-off Voltage | 4.2 V / 2.5 V |
| Cathode Material | Li(Ni0.8Co0.1Mn0.1)O2 (NCM811) |
| Anode Material | Graphite |
Prior to thermal abuse testing, the maximum available capacity (Cmax) of each cell was determined at 25°C using a standard characterization procedure on a battery cycler. This step ensured an accurate baseline for setting the desired SOC. The cell was first discharged to 2.5 V, then charged under constant-current constant-voltage (CC-CV) conditions to 4.2 V (with a C/50 cutoff current), followed by a constant-current discharge to 2.5 V. The discharge capacity from this cycle was recorded as Cmax. Cells were then charged to their target SOCs (0%, 25%, 50%, 75%, and 100%) based on this measured capacity. The actual capacities for the test cells are listed in Table 2.
| Target SOC (%) | Rated Capacity (Ah) | Max. Available Capacity, Cmax (Ah) | Actual Charge Content, Cfact (Ah) |
|---|---|---|---|
| 100 | 5.0 | 4.98 | 4.98 |
| 75 | 5.0 | 4.95 | 3.71 |
| 50 | 5.0 | 4.96 | 2.48 |
| 25 | 5.0 | 4.94 | 1.24 |
| 0 | 5.0 | 4.98 | 0 |
2.2 Thermal Runaway Test Setup and Procedure
Thermal runaway was induced using the external heating method, a common technique for evaluating thermal stability. Each test cell was placed horizontally on a ceramic fiber board inside a sealed stainless-steel chamber designed to contain explosions. A nickel-chromium heating wire was tightly wound around the mid-section of the cell. The heating power was supplied by a DC power source set to a constant 50 W output.
Cell surface temperature was monitored at two points (top and bottom) using K-type sheath thermocouples fixed with stainless-steel clamps. Cell voltage was measured directly via the terminals. All temperature and voltage data were recorded at a 1 Hz sampling rate by a high-speed data acquisition unit. A high-definition camera recorded the event through a viewport.
The experimental procedure was consistent for all SOC levels: initiate heating, continue until the cell undergoes complete thermal runaway (characterized by violent venting, flame ejection, and a sudden voltage drop to zero), and terminate heating shortly after. Post-test, the chamber was allowed to cool, and the cell remnants were collected for mass loss measurement.
3. Results and Discussion
3.1 Thermal Runaway Triggering and Combustion Behavior
The thermal response of the Li-ion battery to external heating varied dramatically with SOC. The time from the start of heating to the onset of thermal runaway (tTR) decreased exponentially with increasing SOC, as detailed in Table 3. The 100% SOC cell failed in just 603 seconds, whereas the 25% SOC cell took 1,472 seconds, a 59.1% increase in time to failure. The 0% SOC cell did not undergo thermal runaway under this heating protocol; it only vented gas mildly through its safety valve.
| SOC (%) | TR Trigger Time, tTR (s) | Ignition & Flame | Flame Duration (s) | Qualitative Violence |
|---|---|---|---|---|
| 100 | 603 | Yes | 40 | Most violent explosion, loudest sound, brightest flame. |
| 75 | 836 | Yes | 31 | Violent explosion and sustained flame. |
| 50 | 1,169 | Yes | 23 | Significant explosion and flame. |
| 25 | 1,472 | Yes | 11 | Moderate explosion, shorter flame. |
| 0 | N/A (No TR) | No | 0 | Minor venting only, cell structure intact. |
The combustion behavior was directly linked to SOC. Higher SOC cells produced more intense fireballs immediately after case rupture, with longer flame durations. The 100% SOC cell’s flame persisted for 40 seconds, compared to only 11 seconds for the 25% SOC cell. This indicates a larger volume of combustible gases (e.g., from electrolyte and electrode decomposition) and more reactive ejected materials in high-SOC Li-ion batteries.
Mass loss, measured as the percentage difference in mass before and after the test, serves as a direct indicator of material ejection and consumption during thermal runaway. As shown in Table 4, the mass loss correlates strongly with SOC. The 100% SOC cell lost over 85% of its original mass, with the jellyroll completely ejected and combusted. In contrast, the 0% SOC cell lost less than 7%, primarily from vented gases, with its structure remaining largely coherent.
| SOC (%) | Average Mass Loss, Δmloss (%) |
|---|---|
| 100 | 85.92 |
| 75 | 79.66 |
| 50 | 65.88 |
| 25 | 51.33 |
| 0 | 6.89 |
3.2 Temperature and Voltage Evolution
The surface temperature profiles revealed distinct stages of the thermal runaway process, with characteristic temperatures shifting with SOC. A representative profile for a 100% SOC cell is shown in Figure 1, with stages defined as follows:
Stage I (Heating): Linear temperature rise due to external heating.
Stage II (Self-heating onset): Temperature T1 marks the beginning of exothermic side reactions (e.g., SEI decomposition). The heating rate increases.
Stage III (Venting): Internal pressure builds from gas generation, leading to safety vent opening at temperature Tv, often causing a slight temperature dip.
Stage IV (Thermal Runaway): At temperature T2, separator collapse leads to large-scale internal short circuit, causing a temperature spike to the peak temperature Tpeak.
Stage V (Cooling): After energy is expended, the cell cools down.
The key temperatures T1, Tv, and T2 all decreased with increasing SOC, indicating reduced thermal stability. For instance, T2 dropped from ~224°C at 25% SOC to ~173°C at 100% SOC. Conversely, the peak temperature Tpeak increased with SOC, reaching over 500°C for the fully charged Li-ion battery, as summarized in Table 5.
| SOC (%) | Self-heat onset T1 (°C) | Vent Open Tv (°C) | TR Trigger T2 (°C) | Peak Temp. Tpeak (°C) |
|---|---|---|---|---|
| 100 | 81.5 | 101.3 | 172.5 | 500.1 |
| 75 | 100.4 | 111.2 | 190.9 | 489.6 |
| 50 | 111.6 | 123.5 | 204.9 | 415.9 |
| 25 | 128.3 | 141.1 | 224.4 | 342.8 |
| 0 | 139.7 | 153.4 | N/A | 203.8* |
*For 0% SOC, this is the maximum temperature reached during heating, not a TR peak.
The voltage response was synchronous with the temperature stages. After a gradual decline during initial heating and self-heating, the voltage underwent an abrupt collapse to zero at a specific temperature Td. This event signifies the onset of a severe internal short circuit and is a clear marker for the beginning of the violent thermal runaway stage. Td was also found to be SOC-dependent, occurring at lower temperatures for higher SOC cells, consistent with the T2 trend.
3.3 Quantitative Hazard Analysis: Energy Release and TNT Equivalent
To quantify the thermal hazard, the total energy released during the main thermal runaway event (Stage IV) was estimated. Assuming the cell mass and specific heat capacity remain approximately constant during the rapid temperature rise, the released energy ΔH can be approximated by:
$$ \Delta H = m \cdot c_p \cdot (T_{peak} – T_2) $$
where \( m \) is the initial cell mass (kg), \( c_p \) is the average specific heat capacity (J kg-1 K-1), \( T_{peak} \) is the peak temperature (K), and \( T_2 \) is the thermal runaway trigger temperature (K). Using \( c_p = 913.62 \, \text{J kg}^{-1} \text{K}^{-1} \) for this type of 21700 Li-ion battery, the energy release was calculated.
To contextualize this energy in terms of explosive potential, the TNT equivalent mass \( W_{TNT} \) was computed:
$$ W_{TNT} = \frac{\eta \cdot \Delta H}{H_{TNT}} $$
where \( \eta \) is an explosion efficiency factor (assumed to be 1 for a conservative estimate) and \( H_{TNT} \) is the heat of combustion of TNT, taken as 4,500 J/g.
Finally, a simplified model for the blast effect radius \( R \) was applied, often used in hazard distance estimation:
$$ R = H_R \cdot \left( \frac{W_{TNT}}{1000} \right)^{1/3} $$
where \( H_R \) is a scaled distance factor, taken as 13.6 m/kg1/3 based on empirical correlations for pressure damage.
The results of these calculations, presented in Table 6, demonstrate a stark increase in hazard with SOC. The energy released by a 100% SOC Li-ion battery is nearly three times that of a 25% SOC cell. In explosive terms, a single failing high-SOC 21700 cell can release energy equivalent to over 4.5 grams of TNT, with a potential blast effect radius approaching 2 meters. This underscores the critical importance of SOC management for safety, especially during storage, transportation, and recycling of Li-ion batteries.
| SOC (%) | Released Energy, ΔH (kJ) | TNT Equivalent, WTNT (g) | Estimated Blast Effect Radius, R (m) |
|---|---|---|---|
| 100 | 20.65 | 4.58 | 1.85 |
| 75 | 18.83 | 4.18 | 1.79 |
| 50 | 13.31 | 2.96 | 1.58 |
| 25 | 7.46 | 1.66 | 1.27 |
4. Conclusions
This experimental study provides a detailed quantitative analysis of the thermal runaway behavior of a commercial 21700 NCM811/graphite Li-ion battery across its full SOC range. The key conclusions are as follows:
1. The state of charge is the dominant factor governing the severity and hazard of thermal runaway in this Li-ion battery. Higher SOC leads to significantly reduced thermal stability, evidenced by lower characteristic onset temperatures (T1, Tv, T2) and drastically shorter times to failure under thermal abuse.
2. The violence of the event, measured by combustion intensity, flame duration, and post-event mass loss, scales strongly with SOC. A fully charged cell can lose over 85% of its mass in a violent ejection and combustion event, whereas a fully discharged cell may only vent gases without catastrophic failure under the same heating condition.
3. A quantitative hazard assessment reveals that the thermal energy released during the runaway phase increases non-linearly with SOC. The potential explosive yield, expressed as TNT equivalent, can exceed 4.5 grams for a single 100% SOC 21700 Li-ion battery, corresponding to a blast effect radius of approximately 1.85 meters under the applied model.
These findings have direct implications for the safe handling and design of systems using high-energy-density Li-ion batteries. They reinforce the critical need for battery management systems to accurately monitor and control SOC, not only for performance but for safety. Furthermore, safety protocols for storage, transport, and end-of-life processing must prioritize the SOC of Li-ion batteries, with clear guidelines to maintain cells at the lowest practical SOC to mitigate inherent thermal hazards. The data presented here can serve as valuable input for modeling thermal runaway propagation in battery packs and for developing more robust safety standards.