Driven by global decarbonization goals, the application of li ion batteries has expanded dramatically into critical sectors such as energy storage power stations, grid support, and electric mobility. Their high energy density, long cycle life, and decreasing cost make them the cornerstone of the modern clean energy transition. However, the persistent risk of thermal runaway—an uncontrolled, self-accelerating exothermic reaction within the cell—poses a severe threat to equipment integrity and personal safety. Catastrophic failures, often involving fire, explosion, and the release of toxic gases, underscore the critical need to deepen our understanding of this failure mode. Therefore, systematic research into the safety characteristics, particularly the thermodynamics and triggering mechanisms of thermal runaway, holds immense theoretical and practical value for designing safer batteries and reliable management systems. In our study, we focus on a widely used commercial cell type to elucidate the fundamental electrochemical evolution during aging and the consequent impact on thermal instability under various energetic states.
This work employs a commercial 18650 cylindrical li ion battery with a LiNi0.6Co0.2Mn0.2O2 (NCM622) cathode and a graphite anode as the primary research object. We subjected these li ion battery units to controlled charge-discharge cycling to simulate different states of health (SOH). A combination of electrochemical characterization techniques, including direct current (DC) internal resistance measurement via the Hybrid Pulse Power Characterization (HPPC) method, Electrochemical Impedance Spectroscopy (EIS), and capacity fade analysis, was used to quantify degradation. The core safety evaluation was conducted using Accelerating Rate Calorimetry (ARC), which provides an adiabatic environment essential for studying the self-sustaining nature of thermal runaway in a li ion battery. By testing cells at different states of charge (SOC: 0%, 50%, 100%), we aimed to map the intrinsic thermal hazard against their stored electrochemical energy.
The degradation of the li ion battery through cycling reveals a clear trajectory of performance decay. The discharge capacity, a direct measure of a li ion battery’s energy storage capability, fades with cycle number. Our data shows this fade is initially more rapid before asymptotically approaching a more stable but gradually declining rate. This external manifestation of aging is intrinsically linked to internal electrochemical changes. The DC internal resistance (RDC), a crucial parameter influencing heat generation during operation and thermal runaway propagation, was observed to increase with cycling. The fresh li ion battery exhibited the lowest RDC, while the aged cells showed higher values, indicating a reduction in overall efficiency and stability.
| Cycle Count | Remaining Capacity (%) | DC Internal Resistance @ 50% SOC (mΩ) | Ohmic Resistance, Rs (mΩ) | SEI Film Resistance, RSEI (mΩ) | Charge Transfer Resistance, Rct (mΩ) |
|---|---|---|---|---|---|
| 0 (Fresh) | 100 | 32.30 | 17.3 | 1.1 | 7.7 |
| 100 | ~95 | 32.74 | 23.6 | 2.7 | 47.2 |
| 200 | ~90 | 33.82 | 26.7 | 2.9 | 93.8 |
| 300 | ~85 | 34.16 | 28.3 | 4.6 | 125.0 |
To deconvolute the contributions to the increasing impedance of the aging li ion battery, we performed EIS analysis. The Nyquist plots were fitted to a common equivalent circuit model for a li ion battery: Rs(RSEICPESEI)(RctCPEdl). The fitted parameters, summarized in Table 1, reveal a systematic increase in all resistive components. The ohmic resistance (Rs), related to electrolyte ionic conductivity and bulk electrode electronic conductivity, increases moderately. The resistance associated with the Solid-Electrolyte Interphase (SEI) layer (RSEI) grows, indicating continual SEI reformation and thickening, which consumes active lithium. The most dramatic change is observed in the charge transfer resistance (Rct), which increases by over an order of magnitude. This surge in Rct points to significant degradation in the electrode/electrolyte interface kinetics, likely due to surface film buildup, particle cracking, or loss of electrical contact within the electrodes. The overall impedance (Z) of a li ion battery can be conceptually represented as the sum of these frequency-dependent components:
$$ Z(\omega) = R_s + \frac{R_{SEI}}{1 + (j\omega R_{SEI} C_{SEI})^{\alpha_{SEI}}} + \frac{R_{ct}}{1 + (j\omega R_{ct} C_{dl})^{\alpha_{dl}}} + Z_W $$
where $C$ and $\alpha$ are capacitance and dispersion factors for constant phase elements (CPE), and $Z_W$ is the Warburg diffusion impedance. The growth in $R_{ct}$ dominantly increases the overall polarization ($\eta$) during operation, described by the Butler-Volmer equation, reducing efficiency and increasing joule heating:
$$ \eta = \frac{RT}{\alpha nF} \ln\left(\frac{j}{j_0}\right) \approx R_{ct} \cdot j $$
where $j$ is current density and $j_0$ is exchange current density. This increased polarization and internal resistance fundamentally lower the thermal stability threshold of the aged li ion battery, making it more susceptible to triggering under abuse conditions.
The propensity for thermal runaway in a li ion battery is profoundly influenced by its State of Charge (SOC), which dictates the amount of stored chemical energy and the reactivity of electrode materials. Our ARC tests on fresh cells at 0%, 50%, and 100% SOC quantified this relationship. Key thermal milestones were identified: the onset temperature of self-heating (Tonset), the temperature at venting/jet (Tvent), the temperature marking the beginning of uncontrollable thermal runaway (TTR, defined by a heating rate ≥ 2 °C/min), and the maximum surface temperature (Tmax). The results, detailed in Table 2 and depicted in the temperature-time profiles, present a stark correlation.
| Parameter | 0% SOC | 50% SOC | 100% SOC |
|---|---|---|---|
| Open Circuit Voltage (V) | 2.96 | 3.76 | 4.18 |
| Onset Temperature, Tonset (°C) | 115 | 95 | 85 |
| Venting/Jet Temperature, Tvent (°C) | 128.5 | 135 | 110.2 |
| Thermal Runaway Start, TTR (°C) | 183.4 | 152.9 | 128.9 |
| Maximum Temperature, Tmax (°C) | 290.9 | 288.3 | 186.5 |
| Time from Tonset to Tvent (min) | 649 | 1073 | 817 |
| Time from Tvent to TTR (min) | 433 | 101 | 82 |
| Post-Test Residual Mass (g) | 41.1 | 14.1 | 11.1 |
The data clearly shows that as the SOC of the li ion battery increases, all characteristic temperatures decrease. A fully charged li ion battery (100% SOC) begins self-heating at a significantly lower temperature (85°C) compared to a discharged one (115°C). This is primarily because at high SOC, the graphite anode is highly lithiated (LixC6, x ≈ 1), making it thermodynamically more reactive. The metastable SEI layer is also less stable against decomposition at elevated temperatures when the anode is lithiated. The onset of exothermic reactions, such as SEI decomposition, can be described by an Arrhenius-type relationship where the reaction rate constant $k$ for a given side reaction is faster when the reactant (lithiated graphite) concentration is high:
$$ k = A \exp\left(-\frac{E_a}{RT}\right) $$
The lower Tonset at high SOC implies a lower effective activation energy ($E_a$) barrier for the initial exothermic reactions in that state. Once triggered, the thermal runaway proceeds more violently and rapidly in the high-SOC li ion battery. The time from the onset of self-heating to the onset of full thermal runaway (TTR) is shortest for the 100% SOC cell. The voltage profile during the ARC test provides further insight. In all cases, a sudden voltage drop to near 0V coincides with Tvent, indicating massive internal shorting due to separator meltdown (typically polyethylene, melting point ~135°C). The violent jet observed, particularly at 50% and 100% SOC, results from rapid gas generation from electrolyte decomposition and cathode reactions, building immense internal pressure. Interestingly, the 50% SOC li ion battery exhibited the most violent mechanical rupture, suggesting a possible optimal pressure-building scenario from combined gas generation and liquid electrolyte content before full consumption. The total energy released during the thermal runaway event of a li ion battery is a function of its SOC and the extent of chemical reactions. The maximum temperature (Tmax) reached, while lower for the 100% SOC cell in our test (potentially due to rapid mass ejection limiting subsequent combustion), does not negate the higher total heat release that likely occurred more intensely in a shorter time. The significantly lower residual mass for the 50% and 100% SOC cells confirms more complete consumption of internal materials.
The sequence of exothermic reactions leading to thermal runaway in a commercial NCM/graphite li ion battery can be modeled as a series of successive, accelerating steps. The overall heat generation rate ($\dot{Q}_{gen}$) is the sum of contributions from individual reactions (SEI decomposition, anode-electrolyte reaction, cathode decomposition, electrolyte decomposition, etc.):
$$ \dot{Q}_{gen} = \sum_i \Delta H_i \cdot A_i \exp\left(-\frac{E_{a,i}}{RT}\right) [C_i]^{n_i} $$
where $\Delta H_i$, $A_i$, $E_{a,i}$, $[C_i]$, and $n_i$ are the enthalpy, pre-exponential factor, activation energy, concentration, and reaction order for reaction $i$. At high SOC, $[C_i]$ for reactions involving lithiated graphite or delithiated cathode is high, leading to a larger $\dot{Q}_{gen}$ at any given temperature once the reactions are activated. Under adiabatic ARC conditions, this generated heat cannot dissipate, leading to a temperature rise described by:
$$ m C_p \frac{dT}{dt} = \dot{Q}_{gen} $$
where $m$ is mass and $C_p$ is heat capacity. The positive feedback loop—heat generation increases temperature, which exponentially increases reaction rates, generating more heat—defines the thermal runaway of the li ion battery. Our results quantitatively show how the initial conditions (SOC) set the parameters in these equations, determining the trajectory and severity of the failure.
Based on our findings regarding the performance decay of the li ion battery and its thermal hazard profile, several practical recommendations can be made to enhance safety. Firstly, for transportation and storage, maintaining li ion battery packs at a low state of charge (ideally below 30% SOC) is a critical safety measure that significantly raises the thermal abuse threshold. Secondly, in operational use, strategies that minimize stress factors contributing to impedance growth should be employed. This includes avoiding consistent operation at extreme states of charge (very high or very low), minimizing exposure to high temperatures, and using charging protocols that prevent excessive currents which accelerate degradation. Thirdly, end-of-life management is crucial. A li ion battery with significantly degraded capacity (e.g., below 80% of original capacity) and elevated internal resistance presents a higher inherent risk. Implementing state-of-health (SOH) monitoring and establishing clear retirement criteria are essential. For repurposing in secondary life (e.g., stationary storage), a thorough safety re-assessment, including impedance checks and possibly updated thermal management, is mandatory for aged li ion battery modules.
In conclusion, our systematic investigation into the thermal runaway performance of a commercial NCM li ion battery has yielded fundamental insights linking electrochemical aging to thermal instability. We have quantitatively demonstrated that the degradation of a li ion battery through cycling is externally manifested as capacity fade but is internally driven by a significant increase in electrochemical impedance, particularly the charge transfer resistance. This increase elevates operational polarization and reduces thermal stability. Furthermore, we have unequivocally established that the thermal hazard of a li ion battery escalates with increasing state of charge. Higher SOC lowers the onset temperature for exothermic reactions and accelerates the progression to violent thermal runaway. These findings underscore the importance of sophisticated Battery Management Systems (BMS) that not only monitor voltage and temperature but also track state-of-health parameters like impedance for early failure预警. Future work should focus on integrating such real-time diagnostics with predictive models for thermal runaway, ultimately enabling fail-safe designs and management protocols for the next generation of safer li ion battery systems across all applications.