Study on thermal safety and thermal management of retired ternary lithium traction battery pack

The rapid development of electric vehicles has brought about a dramatic increase in the number of retired lithium-ion batteries from traction battery packs. These retired cells, especially those with high nickel content and high energy density, still possess considerable residual value. However, their intrinsic material instability and the thermal hazards induced by internal electrolyte decomposition, solid electrolyte interface (SEI) breakdown, and cathode phase transition pose elevated risks of thermal runaway (TR). This work systematically investigates the thermal-electrochemical characteristics of retired ternary lithium-ion cells with three different nickel contents, namely Li(Ni₀.₅Co₀.₂Mn₀.₃)O₂/C (NCM523), Li(Ni₀.₆Co₀.₂Mn₀.₂)O₂/C (NCM622), and Li(Ni₀.₈Co₀.₁Mn₀.₁)O₂/C (NCM811), from both macroscopic and microscopic perspectives. After a rigorous screening procedure, 130 out of 150 cells were classified as qualified, corresponding to an acceptance ratio of 86.7%. The capacity consistency of the screened cells was satisfactory, with NCM811 showing 15.5% and 2.9% higher discharge capacity than NCM523 and NCM622, respectively. Under adiabatic heating tests, NCM811 exhibited the highest peak temperature and temperature rise rate under both room-temperature and high-temperature environments. Moreover, in thermal runaway experiments triggered by oven heating, NCM811 showed the lowest triggering temperature (T₂ = 139.9 °C) but the highest peak temperature (T₃ = 676.44 °C) and the fastest temperature rise rate (9420.33 °C·min⁻¹). Microstructural characterizations of the anodes and active materials after TR revealed that the damage severity is positively correlated with the nickel content, with NCM811 suffering the most severe structural degradation, particle cracking, and active material detachment from the current collector. To address the heat dissipation bottleneck in a retired traction battery pack, a novel high-insulation, high-thermal-conductivity composite phase change material (CPCM) was developed. The CPCM consists of paraffin (PA), expanded graphite (EG), silicon nitride (Si₃N₄), and epoxy resin (ER), where a binary Si₃N₄/EG thermally conductive skeleton was designed to improve both the thermal conductivity and electrical insulation. Six CPCM formulations were synthesized by varying the Si₃N₄ content from 0 to 10 wt%. Experimental results demonstrate that the thermal conductivity rises monotonically from 1.11 to 2.53 W·m⁻¹·K⁻¹ as the Si₃N₄ loading increases. In contrast, the volume resistivity increases first and then decreases, reaching its peak value of 8.93 × 10¹¹ Ω·cm at 8 wt% Si₃N₄ loading, which is far above the insulation threshold of 1 × 10⁹ Ω·cm. The optimized CPCM-5 also retains a high latent heat of 147.17 J·g⁻¹ and exhibits a leakage ratio below 0.2% during the high-temperature cycling test. Its tensile and flexural strengths simultaneously reach the maximum values at the same Si₃N₄ loading. Finally, the CPCM-5 was applied to a 6-series-1-parallel module assembled from retired NCM811 18650 cells. Thermal management experiments at discharge rates of 0.5C, 1.0C and 2.0C show that CPCM cooling can significantly reduce the peak surface temperature compared with natural air convection. At a 2.0C discharge rate, the peak temperature of the CPCM-5 cooled module is 48.2 °C, while that of the air-cooled module reaches 67.7 °C. The maximum temperature difference inside the module decreases from 3.6 °C (air) to 1.3 °C (CPCM-5). In continuous charge-discharge cycles under high-rate conditions, the CPCM-5 system is able to keep the battery temperature within a safe range, demonstrating outstanding temperature uniformity and thermal regulation capability.

1. Introduction

The depletion of fossil fuels and the deterioration of the global climate have accelerated the transition toward clean energy. Electric vehicles (EVs) and hybrid electric vehicles (HEVs) are regarded as effective solutions to reduce CO₂ emissions and air pollutants. As the core energy storage component of EVs, the traction battery pack determines the driving range, dynamic performance, and overall safety. Among various battery chemistries, lithium-ion batteries (LIBs) are the most widely adopted because of their high specific energy, long cycle life, low self-discharge rate, and negligible memory effect. However, the increasing demand for longer cruising range has pushed the energy density of cathodes toward higher nickel contents. Ternary cathodes such as NCM523, NCM622, and NCM811 have become mainstream in modern EV traction battery packs.

The retirement wave of EV traction battery packs is approaching. Typically, a battery is retired when its capacity decreases to 70–80% of its initial value. Nevertheless, retired batteries can still be used in less demanding applications, such as stationary energy storage, low-speed vehicles, and backup power. This cascade utilization can maximize the economic and environmental benefits. However, safety concerns for retired traction battery packs are more significant than those for fresh cells, particularly for high-nickel chemistries. The internal degradation of electrodes, loss of electrolyte, and growth of lithium dendrites increase the risk of internal short circuits. Moreover, the thermal stability of nickel-rich cathode materials is intrinsically poorer than that of low-nickel or lithium iron phosphate cathodes. Therefore, understanding the thermal-electrochemical behaviors and thermal runaway mechanisms of retired ternary cells is essential for safe cascade use.

Thermal runaway (TR) is the most catastrophic failure mode of LIBs. It can be triggered by various abuse conditions, including mechanical abuse, electrical abuse, and thermal abuse. The fundamental cause of TR is the imbalance between heat generation and heat dissipation. Once the internal temperature reaches a critical point, exothermic side reactions, such as SEI decomposition, anode–electrolyte reaction, cathode decomposition, and electrolyte oxidation, become self-sustaining and lead to a rapid temperature rise. For a traction battery pack, TR in one cell may propagate to adjacent cells, causing fire or explosion. Thus, a comprehensive understanding of the TR characteristics of individual cells is a prerequisite for the design of safe battery modules and packs.

Battery thermal management systems (BTMSs) play a vital role in maintaining the temperature of a traction battery pack in an optimal range, typically 20–50 °C. The existing cooling methods include air cooling, liquid cooling, heat pipe cooling, and phase change material (PCM) cooling. Air cooling is simple but inefficient for high discharge rates. Liquid cooling is effective but suffers from leakage risk and added weight. Heat pipe cooling has high thermal conductivity but may lead to contact resistance issues. PCM cooling, especially with composite phase change materials (CPCMs), offers passive temperature regulation, high latent heat, and excellent temperature uniformity. Nonetheless, pure paraffin has low thermal conductivity (~0.2–0.3 W·m⁻¹·K⁻¹) and suffers from leakage during solid-liquid phase transition. To overcome these defects, researchers often blend paraffin with expanded graphite, carbon nanotubes, metal foams, or ceramic fillers. Among ceramic fillers, silicon nitride (Si₃N₄) has attracted considerable interest due to its comparatively high thermal conductivity (140–180 W·m⁻¹·K⁻¹), excellent electrical insulation, and low dielectric loss.

The aim of this work is to provide a systematic approach, from cell characterization to thermal management material development and module-level application, for retired ternary traction battery packs. The novelty lies in three aspects: (1) a comparative study of macro-scale TR characteristics and micro-scale electrode structural degradation among NCM523, NCM622, and NCM811 after TR; (2) the rational design of a Si₃N₄/EG hybrid thermally conductive and electrical insulating framework inside paraffin/epoxy CPCM; and (3) the experimental validation of the optimized CPCM in a retired NCM811-based traction battery pack module.

2. Screening and reusability of retired ternary cells

2.1 Experimental samples

The retired 18650-type ternary cells with three different cathode stoichiometries were selected. Their nominal specifications are listed in Table 1. Cells were obtained from a retired EV traction battery pack after service.

Table 1. Specification of the tested cells with different nickel contents

2.2 Screening procedure

The screening procedure is presented schematically below:

1. Visual inspection: collecting cells with no cracks, leakage, or severe corrosion.
2. Open-circuit voltage and internal resistance check: cells with voltage < 2 V or resistance > 30 mΩ were discarded.
3. Capacity measurement: three charge–discharge cycles were applied; the last discharge capacity was used as reference.
4. Self-discharge / charge retention test: after storage for 30 days, the capacity retention and recovery rates were calculated.

The capacity test program is given in Table 2 for NCM811.

Table 2. Capacity detection program for a retired cell (NCM811 example)

The capacity retention rate and capacity recovery rate were evaluated as follows:

$$
\text{Capacity retention (\%)} = \frac{C_1}{C_0} \times 100\% \tag{1}
$$

$$
\text{Capacity recovery (\%)} = \frac{C_2}{C_0} \times 100\% \tag{2}
$$

where C₀ is the initial discharge capacity before storage, C₁ is the discharge capacity after 30 days of storage, and C₂ is the discharge capacity after a subsequent full charge–discharge cycle.

2.3 Screening results and reuse potential

The quantities of defective cells from the visual inspection, voltage inspection, and resistance inspection are summarized in Table 3.

Table 3. Summary of screening results

According to Figure 2-10 of the original study, the capacity distribution of the selected cells is shown in Table 4.

Table 4. Capacity distribution of screened cells

It is evident that NCM811 delivers higher residual capacity than NCM523 and NCM622. The internal resistance distribution is given in Table 5.

Table 5. Internal resistance distribution

As seen in 2.4.4, after 30 days of storage, 90.8% of the retired cells had capacity retention above 95%, and 79.0% showed capacity recovery above 90%. These results demonstrate that the retired ternary traction battery cells possess good reusability for secondary applications, provided that a careful selection is carried out.

3. Thermal runaway behavior and microstructure evolution of cells with different nickel contents

3.1 Adiabatic heat generation characteristics

In order to evaluate the heat generation under actual charge/discharge conditions, adiabatic tests were performed in an EV-Accelerating Rate Calorimeter (EV-ARC). Cells at 100% SOC were discharged at 0.5C, 1.0C, and 2.0C in both room temperature (25 °C) and high temperature (55 °C) adiabatic environments. The measured peak temperatures and temperature rise rates are summarized in Tables 6 and 7.

Table 6. Adiabatic heat generation of cells under 25 °C environment

Table 7. Adiabatic heat generation of cells under 55 °C environment

At a lower C-rate (0.5C and 1.0C), the heat generation rate is lower than the heat dissipation rate; hence, the cell temperature rises slowly and the peak temperature differences among different nickel contents are not obvious. In contrast, at a high discharge rate (2.0C), heat accumulation dominates, and the peak temperature rises monotonically with nickel content. For example, at 25 °C and 2.0C discharge, NCM811 presents a peak temperature of 70.9 °C, which is 25.0% and 3.1% higher than NCM523 and NCM622, respectively. The temperature rise rate of NCM811 is 51.6% higher than that of NCM523.

3.2 Thermal runaway characteristics under heat abuse

The thermal runaway experiments were carried out by heating the cells inside an EV-ARC under adiabatic conditions. The criterion for detecting self-heating was 0.03 °C·min⁻¹, while the TR triggering was considered when the temperature rise rate exceeded 1 °C·s⁻¹. During TR, the cell voltage, surface temperature, and pressure were recorded. Three characteristic temperatures are defined:

– T₁: the self-heating onset temperature (when dT/dt ≥ 0.03 °C·min⁻¹);
– T₂: the thermal runaway triggering temperature (when dT/dt ≥ 1 °C·s⁻¹);
– T₃: the maximum temperature reached during TR.

Table 8. Key thermal runaway parameters for three cells

The higher nickel content in the cathode leads to a lower T₂, which means that less thermal energy is needed to trigger the runaway of NCM811. Simultaneously, the peak temperature T₃ and the temperature rise rate increase significantly with nickel content. This indicates that high-nickel cathodes have less thermal stability and release more energy during the decomposition reaction. These results highlight the need for stronger thermal management systems and safer separator/electrolyte designs for high-energy-density traction battery packs.

3.3 Post-TR morphological changes

After the TR tests, the cells were disassembled in an argon-filled glovebox. Macroscopic photographs revealed that the NCM811 cell suffered the most severe deformation and rupture of the can; the cap of the cell was completely damaged. The internal winding structure of NCM523 remained relatively intact, while NCM622 and NCM811 showed severe damages. Furthermore, scanning electron microscopy (SEM) observations of anode materials before and after TR are summarized as follows:

– Before TR: The graphite particles are tightly packed and uniformly distributed in all three cell types.
– After TR: The active material layer peels off from the copper foil, and cracks appear in the graphite particles. The extent of detachment increases with nickel content. SEM images show severe agglomeration and large voids on the NCM811 anode after TR.

These microstructural degradations disrupt the electron-conducting network and Li⁺ diffusion channels, leading to reduced capacity and increased heat generation in subsequent cycles. Therefore, the irreversible destruction of the electrode microstructure is the root cause for the degradation of electrochemical performance and the escalation of heat generation after a TR event.

4. Development and characterization of high-insulation and high-thermal-conductivity composite phase change materials

4.1 Design and fabrication

To improve the thermal conductivity of paraffin/expanded graphite (PA/EG) CPCMs while maintaining excellent electrical insulation, a Si₃N₄/EG binary skeleton was designed to replace the conventional EG-only or metallic skeleton. The CPCMs were synthesized by a melt-mixing and gelation method. The components include PA (melting point = 48.8 °C, latent heat = 225.8 J·g⁻¹), EG, epoxy resin ER (prepolymer and curing agent), and Si₃N₄ particles (average diameter 2 μm, theoretical thermal conductivity ≈ 177 W·m⁻¹·K⁻¹). The formulations are listed in Table 9.

Table 9. Formulation and designations of CPCM samples

During fabrication, the PA is first melted in a water bath at 70 °C. EG is added under mechanical stirring for 60 min, followed by the successive addition of Si₃N₄ powder and ER. The mixture is then cast into acrylic molds and cured at 90 °C for 10 h.

4.2 Thermophysical properties

The phase change behavior was analyzed using differential scanning calorimetry (DSC), and the latent heats are given in Table 10.

Table 10. DSC results of CPCM samples

The latent heat values are proportional to the PA content. CPCM-5 retains a latent heat of 147.14 J·g⁻¹, which is sufficiently high for battery thermal management.

Thermogravimetric analyses (TGA) showed that all CPCMs have two major decomposition steps: the first from 200 to 300 °C (PA evaporation) and the second from 300 to 400 °C (ER decomposition). The total mass loss at 800 °C decreased with Si₃N₄ content, indicating that Si₃N₄ contributes to better thermal stability.

The leakage behavior was evaluated in a high-temperature cycling test at 70 °C for 12 h. The leakage ratio is defined as:

$$
K_{leakage} = \frac{M_0 – M_t}{M_0} \times 100\% \tag{3}
$$

All CPCMs showed leakage below 0.2%, as shown in Table 11.

Table 11. Leakage ratio after 12-h high-temperature test

Since the leakage ratio is extremely low, the CPCM has adequate shape stability for the traction battery pack application.

4.3 Thermal conductivity and electrical insulation

The thermal conductivity \(\lambda\) and volume resistivity \(\rho_v\) were measured at 50 °C. The volume resistivity was calculated by:

$$
\rho_v = R_x \cdot \frac{S}{t} \tag{4}
$$

where Rₓ is the measured insulation resistance, S is the effective electrode area (21.24 cm²), and t is the sample thickness. The results are listed in Table 12.

Table 12. Thermal conductivity and volume resistivity of CPCMs

The twofold improvement in thermal conductivity from 1.11 to 2.39 W·m⁻¹·K⁻¹ for CPCM-5 is attributed to the formation of a continuous Si₃N₄/EG thermally conductive network. The volume resistivity of CPCM-5 exceeds \(8.9 \times 10^{11}\ \Omega\cdot cm\), which is two orders of magnitude greater than the acceptable threshold for insulating materials (\(10^9\ \Omega\cdot cm\)). The decrease in resistivity at 10 wt% Si₃N₄ can be explained by particle agglomeration and interface polarization effects, as observed in SEM images. At 8 wt% Si₃N₄, a uniform dispersion of Si₃N₄ particles occurs, which can intercept the electron transport paths in the EG network while providing a more efficient phonon transport pathway.

4.4 XRD and morphology

X-ray diffraction (XRD) patterns confirmed that the characteristic peaks of PA (21.52° and 24.51°), EG (26.77°), and Si₃N₄ (36.14°, 40.7°, 47.76°) are all present without shifts, which suggests that the manufacturing process only involves physical mixing, maintaining the chemical stability of the components.

SEM characterization showed that CPCM-1 possesses a porous and loose morphology with many micro-voids due to volume shrinkage of the ER and PA. For CPCM-5, these micro-voids are effectively filled by Si₃N₄ particles, leading to a denser and more continuous structure. In CPCM-6, agglomeration of Si₃N₄ particles becomes visible, which reduces the effective interface area and deteriorates the mechanical and insulating properties.

4.5 Mechanical properties

The tensile and flexural strengths of the CPCMs are shown in Table 13.

Table 13. Mechanical strength of CPCMs

At 8 wt% Si₃N₄, the tensile and flexural strengths reach their maximum values, improving by about 43.6% and 79.9% compared with CPCM-1. This enhancement originates from the reinforcing effect of well-dispersed Si₃N₄ particles and the denser microstructure. When excessive Si₃N₄ is used, agglomeration weakens the internal framework, thus reducing the mechanical properties.

5. Application of the optimal CPCM in thermal management of a retired NCM811 traction battery module

5.1 Module design and experimental setup

To evaluate the practical cooling performance of the CPCM, the module-level experiments were conducted using retired NCM811 18650 cells (nominal capacity 3.05 Ah, nominal voltage 3.6 V). The module is composed of six cells in parallel (6P1S) with a nominal module capacity of 18.3 Ah. Three cooling configurations were compared:

– Module A: natural air convection cooling (no PCM);
– Module B: CPCM-1 (without Si₃N₄) cooling;
– Module C: CPCM-5 (with 8 wt% Si₃N₄) cooling.

The CPCMs were machined into cylinders with an outer diameter of 38.5 mm, an inner hole of 18.5 mm, and a height of 60 mm, into which the cells were inserted. The measurements were carried out in a thermostat chamber at 25 °C. The charge/discharge tests were performed using a battery test system, and the surface temperatures were monitored by T-type thermocouples and recorded by a data acquisition device.

The thermal management performances at different discharge rates are presented in Table 14.

Table 14. Peak temperature of modules under different discharge rates

At a high discharge rate of 2.0C, CPCM-5 suppresses the peak temperature to 48.2 °C, which is 19.5 °C lower than that of the air-cooled module. Moreover, the CPCM-5 module presents a smoother temperature rise curve, indicating that the melting of paraffin efficiently absorbs the generated heat and delays the temperature increase.

5.2 Temperature uniformity

The maximum temperature difference within each module is a key indicator for uniformity. Values for the 2.0C discharge are given in Table 15.

Table 15. Maximum temperature difference at 2.0C discharge

The temperature difference of 1.3 °C in CPCM-5 module is well within the desired uniform temperature standard of less than 2–3 °C for a traction battery pack. This improved uniformity results from the combination of the high thermal conductivity of the Si₃N₄/EG skeleton and the high latent heat of paraffin, which quickly spreads the local heat and dampens thermal gradients.

5.3 Cyclic charge–discharge behavior

To further verify the long-term cooling capability of the CPCM-5, a continuous charge-discharge cycling experiment was performed. The module was charged at 1.0C to 4.2 V in CC-CV mode, then discharged at 2.0C to 2.5 V. Up to 10 cycles were conducted without any rest intervals. The measured peak temperatures for the last cycle are listed in Table 16.

Table 16. Peak temperature during the 10th continuous cycle (after 2.0C discharge)

The maximum cycle-to-cycle temperature variation for the CPCM-5 module was only 0.7 °C, proving that the composite phase change material can repeatedly melt and solidify without significant performance degradation. Since the latent heat of CPCM-5 is sufficiently high (147.14 J·g⁻¹), it can absorb the repeated generated heat and release it during the rest or charging period without completely losing its thermal storage capability.

6. Conclusion

Retired ternary lithium-ion cells with different nickel contents were first screened and analyzed. Higher nickel content leads to higher residual capacity but also causes more severe thermal safety concerns. The key results are summarized below:

1. From 150 retired cells, 86.7% were qualified for reuse. NCM811 had the highest residual capacity, but also the widest internal resistance distribution. Charge retention tests showed that 90.8% of the cells retained more than 95% of their capacity after 30-day storage, indicating promising reuse potential for low-rate applications.

2. In adiabatic discharge tests at high rates, NCM811 exhibited a maximum temperature of 70.9 °C and a temperature rise rate of 1.556 °C/min at 25 °C and 2.0C discharge. Under 55 °C and 2.0C, the peak temperature of NCM811 reached 96.6 °C with a temperature rise rate of 1.460 °C/min. The temperature rise is more severe at a higher nickel content, particularly under high rates.

3. Thermal runaway triggered by heat abuse revealed that higher nickel content lowers the self-heating onset temperature T₁ (78.4 °C for NCM811) and the TR triggering temperature T₂ (139.9 °C), simultaneously increasing the peak temperature T₃ (676.44 °C) and the maximum temperature rise rate (9420.33 °C·min⁻¹). Therefore, high-nickel cells are more susceptible to TR and should be protected with robust thermal management strategies.

4. Post-TR microstructure characterizations demonstrated that the anode active material becomes loosely packed, cracked, and severely detached from the copper current collector. NCM811 underwent the most severe irreversible structural damage, which exacerbates the thermal-electrochemical aging and reduces the energy storage capability of the retired cells.

5. A novel Si₃N₄/EG-based CPCM was fabricated. At the optimal Si₃N₄ loading of 8 wt%, the CPCM-5 achieved a thermal conductivity of 2.39 W·m⁻¹·K⁻¹, a volume resistivity of 8.93 × 10¹¹ Ω·cm, a latent heat of 147.14 J·g⁻¹, and excellent shape stability with a leakage ratio below 0.2%. Its tensile and flexural strengths were also the highest among all formulations.

6. Module-level experiments on a 6P1S retired NCM811 traction battery module demonstrated that CPCM-5 cooling provides excellent temperature suppression and uniformity. At a high 2.0C discharge rate, the peak temperature was lowered from 67.7 °C (air cooling) to 48.2 °C, and the maximum temperature difference was reduced from 3.6 °C to 1.3 °C. In continuous cycles, CPCM-5 maintained the module temperature within a safe range.

These findings provide new insights into the thermal hazard analysis of retired ternary lithium-ion traction battery packs and contribute to the development of efficient, safe, and reliable thermal management systems for their cascade utilization. Future work should extend the experimental matrix to include all-electric abuse conditions, larger format cells, and possibly flame-retardant CPCM systems to further enhance the safety of high-energy-density traction battery packs.