I have focused my research on the thermal safety and thermal management of retired ternary lithium-ion batteries, especially those used in electric vehicle battery packs. As electric vehicles become increasingly popular, a large number of lithium-ion batteries retire from electric vehicle battery packs each year. Although these batteries no longer meet the requirements of high-power electric vehicle applications, they still possess considerable residual capacity and can be reused in lower-rate scenarios. Among retired batteries, ternary lithium batteries with high energy density are attractive but also face higher thermal runaway risks because of their internal material structures. In my work, I systematically studied retired ternary lithium batteries with different nickel contents, including NCM523, NCM622, and NCM811. I first evaluated their reusability and inconsistency, then investigated their thermal-electrochemical behaviors and thermal runaway mechanisms under adiabatic and abusive conditions, and finally developed a silicon nitride-based composite phase change material for thermal management of an electric vehicle battery pack module.
The motivation for my research is clear. Resource shortages and environmental pollution are two major global challenges. Electric vehicles and hybrid electric vehicles have received widespread attention as solutions to reduce carbon dioxide emissions. The electric vehicle battery pack is the core energy storage component, and its performance directly determines driving range, power, and safety. Lithium-ion batteries are widely used in electric vehicle battery packs because of their high specific energy, low self-discharge, long cycle life, and no memory effect. However, as the specific energy of batteries increases, the heat generated by internal irreversible chemical reactions also increases, which raises the probability of thermal runaway. Thermal runaway is one of the most serious safety hazards in electric vehicle battery packs. Therefore, understanding the thermal runaway mechanism and developing effective thermal management strategies are essential for the safe application of retired ternary lithium batteries.
I selected three types of 18650 ternary lithium-ion cells with different nickel contents. The technical specifications are summarized in Table 1. These cells were retired from electric vehicle battery packs and provided a realistic basis for my study.
| Parameter | NCM523 | NCM622 | NCM811 |
|---|---|---|---|
| Nominal voltage (V) | 3.6 | 3.6 | 3.6 |
| Nominal capacity (Ah) | 2.6 | 2.75 | 3.05 |
| Charge cut-off voltage (V) | 4.2 | 4.2 | 4.2 |
| Discharge cut-off voltage (V) | 2.75 | 2.5 | 2.5 |
| Maximum charge rate (C) | 1 | 1 | 1 |
| Maximum discharge rate (C) | 3 | 3 | 2 |
| Mass energy density (Wh/kg) | 180 | 206 | 234 |
| Internal resistance (mΩ) | ≤40 | ≤35 | ≤35 |
To evaluate the reusability of retired cells from electric vehicle battery packs, I performed a comprehensive screening process. I first inspected the appearance of 150 cells. Ten cells showed cracks, corrosion, or electrolyte leakage, accounting for 6.7%. Seven cells had extremely low or zero voltage, accounting for 4.7%. Three cells had low voltage or high internal resistance, accounting for 2.0%. The remaining 130 cells were qualified, accounting for 86.7%. The screening results are shown in Table 2.
| Screening stage | NCM523 | NCM622 | NCM811 | Total | Percentage |
|---|---|---|---|---|---|
| Appearance failure | 2 | 3 | 5 | 10 | 6.7% |
| Voltage failure | 2 | 2 | 3 | 7 | 4.7% |
| Internal resistance failure | 1 | 1 | 1 | 3 | 2.0% |
| Qualified cells | 45 | 44 | 41 | 130 | 86.7% |
I then measured the capacity distribution of the qualified cells. The average discharge capacities were 2414.89 mAh, 2711.16 mAh, and 2790.15 mAh for NCM523, NCM622, and NCM811, respectively. The capacity of NCM811 was 15.5% higher than that of NCM523 and 2.9% higher than that of NCM622. The capacity ranges were 227.17 mAh, 122.51 mAh, and 89.89 mAh, corresponding to 9.4%, 4.4%, and 3.2% of the average capacity. These results indicate good capacity consistency after screening, which is beneficial for regrouping retired cells into a new electric vehicle battery pack for cascade utilization.
I also tested the internal resistance of the screened cells. The average internal resistances were 22.62 mΩ, 22.45 mΩ, and 22.33 mΩ for NCM523, NCM622, and NCM811, respectively. Compared with fresh cells, the internal resistance increased because of aging and damage to internal materials. The differences among individual cells reflect the amplification of initial microscopic differences after repeated charge-discharge cycles.
Capacity retention and recovery are important indicators for the reusability of retired batteries. I stored the cells at room temperature for 30 days and then measured their capacity. The capacity retention rate and recovery rate were calculated using the following equations:
$$ \text{Capacity retention rate} = \frac{C_1}{C_0} \times 100\% $$
$$ \text{Capacity recovery rate} = \frac{C_2}{C_0} \times 100\% $$
where \(C_0\) is the initial capacity, \(C_1\) is the capacity after storage, and \(C_2\) is the capacity after the third discharge. The results showed that 90.8% of the retired cells had a capacity retention rate above 95%, and 79.0% had a recovery rate above 90.0%. This confirms that the retired cells still have good charge retention capability and can be reused in less demanding applications.
Inconsistency among cells is a critical issue for electric vehicle battery packs. I analyzed the causes of inconsistency, including manufacturing differences, storage environment, usage environment, and aging. These factors lead to inconsistent capacity, internal resistance, temperature, and voltage curves. I summarized the causes in Table 3.
| Cause | Description |
|---|---|
| Manufacturing differences | Minor differences in materials and manufacturing tolerances cause variations in performance. |
| Storage environment | Temperature, humidity, and storage time affect battery degradation differently. |
| Usage environment | Different operating conditions and loads lead to different aging degrees. |
| Aging | Repeated cycling causes material degradation and performance divergence. |
I randomly selected five cells from each group and performed a 1.0C discharge temperature rise test. The maximum surface temperatures were 43.1 °C, 45.9 °C, and 46.5 °C for NCM523, NCM622, and NCM811, respectively. The temperature consistency after screening was good, and NCM811 showed the highest temperature, which is consistent with its higher nickel content and higher energy density.
I also measured the voltage curves of nine cells, three from each group. The voltage curves were consistent after screening. During charge and discharge, NCM811 reached the same voltage plateau in the shortest time, and it also completed discharge faster. The time to reach the cut-off voltage decreased as nickel content increased. This behavior indicates that higher nickel content leads to higher discharge capacity but also faster voltage drop, which can affect the overall performance of an electric vehicle battery pack.
After evaluating the reusability and inconsistency, I moved to the thermal-electrochemical characteristics and thermal runaway behavior of the three types of cells. I used an accelerating rate calorimeter (ARC) to study the thermal behavior under adiabatic conditions. I tested the cells at 25 °C and 55 °C, with charge at 1.0C and discharge at 0.5C, 1.0C, and 2.0C. The results are summarized in Table 4.
| Condition | Parameter | NCM523 | NCM622 | NCM811 |
|---|---|---|---|---|
| 25 °C, 2.0C discharge | Maximum temperature (°C) | 56.7 | 68.8 | 70.9 |
| 25 °C, 2.0C discharge | Temperature rise (°C) | 31.7 | 43.8 | 45.9 |
| 25 °C, 2.0C discharge | Temperature rise rate (°C/min) | 1.026 | 1.436 | 1.5559 |
| 55 °C, 2.0C discharge | Maximum temperature (°C) | 78.6 | 91.5 | 96.6 |
| 55 °C, 2.0C discharge | Temperature rise (°C) | 23.6 | 36.5 | 41.6 |
| 55 °C, 2.0C discharge | Temperature rise rate (°C/min) | 0.761 | 1.217 | 1.4597 |
At low discharge rates, the maximum temperature and temperature rise rate were similar for all three cells because the heat generation rate was lower than the heat dissipation rate. However, at high discharge rates, NCM811 exhibited the highest maximum temperature and the fastest temperature rise. At 2.0C and 25 °C, the maximum temperature and temperature rise rate of NCM811 were 70.9 °C and 1.5559 °C/min, which were 25% and 51.6% higher than those of NCM523 and NCM622, respectively. At 55 °C, the maximum temperature of NCM811 was 96.6 °C, and the temperature rise rate was 1.4597 °C/min, which was 1.92 times that of NCM523 and 1.2 times that of NCM622. These results indicate that high-nickel batteries generate more heat, especially under high-temperature and high-rate conditions, increasing the thermal risk in an electric vehicle battery pack.
I then studied the thermal runaway behavior triggered by thermal abuse. The key parameters are the self-heating onset temperature \(T_1\), the thermal runaway trigger temperature \(T_2\), and the maximum temperature \(T_3\). The results are shown in Table 5.
| Parameter | NCM523 | NCM622 | NCM811 |
|---|---|---|---|
| \(T_1\) (°C) | 81.3 | 80.8 | 78.4 |
| \(T_2\) (°C) | 184.3 | 171.7 | 139.9 |
| \(T_3\) (°C) | 574.94 | 670.63 | 676.44 |
| Maximum temperature rise rate (°C/min) | 4264.71 | 9069.1 | 9420.33 |
NCM811 had the lowest \(T_1\) and \(T_2\), but the highest \(T_3\) and the fastest temperature rise rate. This means that high-nickel batteries are more prone to thermal runaway at lower temperatures and release more heat once thermal runaway occurs. The thermal runaway process includes several stages. First, the solid electrolyte interphase (SEI) decomposes and generates heat. Then, the separator shrinks, causing micro-internal short circuits and a sudden voltage drop. The safety valve ruptures, releasing high-temperature gas and reducing the temperature temporarily. After that, the active materials react with the electrolyte and oxygen, causing a rapid temperature increase. The maximum temperature rise rate can exceed 1 °C/s, indicating severe thermal runaway. I calculated the temperature rise rate using:
$$ \text{Temperature rise rate} = \frac{\Delta T}{\Delta t} $$
where \(\Delta T\) is the temperature change and \(\Delta t\) is the time interval. The high temperature rise rate of NCM811 confirms its poor thermal stability.
After thermal runaway, I disassembled the cells in an argon-filled glove box and examined the internal structure. The external appearance showed that NCM811 suffered the most severe damage. The top cap was deformed, and the internal active materials leaked more seriously. The negative electrode current collector of NCM811 was the most damaged, with the coiled structure almost destroyed. I used scanning electron microscopy (SEM) to observe the negative electrode active materials before and after thermal runaway. The results are summarized in Table 6.
| Sample | Before thermal runaway | After thermal runaway |
|---|---|---|
| NCM523 | Active materials are tightly arranged and evenly distributed. | Active materials slightly peel off, conductive network is slightly disordered. |
| NCM622 | Active materials are compact and uniform. | Peeling is more obvious, structure becomes disordered. |
| NCM811 | Active materials are dense and ordered. | Severe peeling, large gaps, and highly disordered structure. |
The SEM images showed that after thermal runaway, the active material particles become rough and loose, with increased gaps. NCM811 exhibited the most severe agglomeration and structural damage. This irreversible damage to the electrode materials is the fundamental reason for the decline in electrochemical performance and the increase in heat generation. The destruction of the electrode material morphology and crystal structure hinders normal electrochemical reactions and reduces the energy storage capacity of the electric vehicle battery pack.
To address the heat dissipation problem in electric vehicle battery packs, I developed a high-insulation and high-thermal-conductivity composite phase change material (CPCM). I used paraffin (PA) as the phase change matrix, expanded graphite (EG) as the thermal conductive skeleton, epoxy resin (ER) as the anti-leakage additive, and silicon nitride (Si\(_3\)N\(_4\)) as the insulating filler. The preparation process involved melt blending and curing. I prepared six CPCM samples with different Si\(_3\)N\(_4\) contents, as shown in Table 7.
| Sample | PA (wt%) | EG (wt%) | ER (wt%) | Si\(_3\)N\(_4\) (wt%) |
|---|---|---|---|---|
| CPCM-1 | 167 | 3 | 30 | 0 |
| CPCM-2 | 163 | 3 | 30 | 4 |
| CPCM-3 | 159 | 3 | 30 | 8 |
| CPCM-4 | 155 | 3 | 30 | 12 |
| CPCM-5 | 151 | 3 | 30 | 16 |
| CPCM-6 | 147 | 3 | 30 | 20 |
I characterized the thermal stability of the CPCM using thermogravimetric analysis (TGA). The TGA curves showed that pure paraffin began to decompose at around 200 °C. The CPCM samples exhibited two degradation stages, and the mass loss decreased as Si\(_3\)N\(_4\) content increased. At 800 °C, the final mass loss rates were 91%, 88%, 86%, 84%, and 80% for CPCM-2 to CPCM-6. The addition of Si\(_3\)N\(_4\) improved the thermal stability of the composite.
I used differential scanning calorimetry (DSC) to measure the phase change temperature and latent heat. The results are shown in Table 8. The latent heat of the CPCM decreased as the paraffin content decreased. CPCM-5, with 8 wt% Si\(_3\)N\(_4\), had a latent heat of 147.14 J/g, which is suitable for battery thermal management because the phase change temperature is close to the optimal operating temperature range of lithium-ion batteries.
| Sample | Si\(_3\)N\(_4\) (wt%) | Peak temperature (°C) | Latent heat (J/g) |
|---|---|---|---|
| Pure PA | 0 | 48.8 | 225.8 |
| CPCM-1 | 0 | 47.8 | 174.06 |
| CPCM-2 | 2 | 48.06 | 163.19 |
| CPCM-3 | 4 | 47.46 | 159.37 |
| CPCM-4 | 6 | 47.18 | 154.64 |
| CPCM-5 | 8 | 47.3 | 147.14 |
| CPCM-6 | 10 | 47.96 | 144.58 |
I also tested the high-temperature cycling leakage performance. The leakage rate was calculated using:
$$ K_{\text{leakage}} = \frac{M_0 – M_1}{M_0} \times 100\% $$
where \(M_0\) is the initial mass and \(M_1\) is the mass after each measurement. After 12 hours at 70 °C, the total leakage rate of all CPCM samples was less than 0.2%. The epoxy resin and expanded graphite effectively prevented paraffin leakage. The addition of Si\(_3\)N\(_4\) further promoted the shape stability of the composite. This is crucial for practical applications in an electric vehicle battery pack because excessive leakage can aggravate safety issues.
I measured the thermal conductivity and volume resistivity of the CPCM. The thermal conductivity was measured using a thermal conductivity analyzer, and the volume resistivity was calculated using:
$$ \rho_V = R_X \times \frac{S}{t} $$
where \(\rho_V\) is the volume resistivity, \(R_X\) is the measured insulation resistance, \(S\) is the effective area of the electrode (21.24 cm\(^2\)), and \(t\) is the thickness of the sample. The results are shown in Table 9.
| Sample | Si\(_3\)N\(_4\) (wt%) | Thermal conductivity (W·m\(^{-1}\)·K\(^{-1}\)) | Volume resistivity (×10\(^{11}\) Ω·cm) |
|---|---|---|---|
| CPCM-1 | 0 | 1.11 | 3.37 |
| CPCM-2 | 2 | 1.65 | 4.95 |
| CPCM-3 | 4 | 1.98 | 5.95 |
| CPCM-4 | 6 | 2.21 | 6.95 |
| CPCM-5 | 8 | 2.39 | 8.93 |
| CPCM-6 | 10 | 2.53 | 6.44 |
The thermal conductivity increased with Si\(_3\)N\(_4\) content. The volume resistivity first increased and then decreased. CPCM-5 with 8 wt% Si\(_3\)N\(_4\) achieved the highest volume resistivity of \(8.93 \times 10^{11}\) Ω·cm, which meets the insulation requirement (\(>1 \times 10^9\) Ω·cm). The thermal conductivity of CPCM-5 was 2.39 W·m\(^{-1}\)·K\(^{-1}\). When Si\(_3\)N\(_4\) content exceeded 8 wt%, the agglomeration of Si\(_3\)N\(_4\) particles and interface polarization reduced the electrical insulation performance. The XRD results confirmed that the addition of Si\(_3\)N\(_4\) did not change the crystal structure of paraffin or expanded graphite, indicating physical mixing rather than chemical reaction.
I also evaluated the mechanical properties of the CPCM. The tensile strength and flexural strength were measured using a universal testing machine. The results are shown in Table 10.
| Sample | Tensile strength (MPa) | Flexural strength (MPa) |
|---|---|---|
| CPCM-1 | 1.42 | 2.15 |
| CPCM-2 | 1.58 | 2.54 |
| CPCM-3 | 1.76 | 2.98 |
| CPCM-4 | 1.89 | 3.32 |
| CPCM-5 | 2.04 | 3.87 |
| CPCM-6 | 1.92 | 3.55 |
CPCM-5 exhibited the highest tensile strength and flexural strength, which were 43.6% and 79.9% higher than those of CPCM-1, respectively. When Si\(_3\)N\(_4\) content exceeded 8 wt%, the mechanical strength decreased due to agglomeration. The SEM images showed that CPCM-5 had a dense and continuous microstructure, while CPCM-6 had agglomerated Si\(_3\)N\(_4\) particles. The element mapping of CPCM-5 confirmed the uniform distribution of C, N, and Si. The dense microstructure improves heat transfer and mechanical strength.
Finally, I applied the CPCM to the thermal management of a retired NCM811 electric vehicle battery pack module. I selected CPCM-1 and CPCM-5 for comparison with natural air convection cooling. The battery module consisted of six 18650 NCM811 cells in a 6P1S configuration. I tested discharge rates of 0.5C, 1.0C, and 2.0C at 25 °C. The temperature results are shown in Table 11.
| Cooling method | 0.5C peak temperature (°C) | 1.0C peak temperature (°C) | 2.0C peak temperature (°C) | 0.5C max temperature difference (°C) | 1.0C max temperature difference (°C) | 2.0C max temperature difference (°C) |
|---|---|---|---|---|---|---|
| Natural air convection | 36.1 | 48.9 | 67.7 | 1.2 | 2.1 | 3.6 |
| CPCM-1 | 34.2 | 43.5 | 56.7 | 0.4 | 0.8 | 2.4 |
| CPCM-5 | 33.9 | 41.6 | 48.2 | 0.15 | 0.23 | 0.37 |
The CPCM-5 cooling system showed the lowest peak temperature and the smallest temperature difference. At 2.0C discharge, the peak temperature was 48.2 °C, which is 28.8% lower than that of natural air convection. The maximum temperature difference was only 1.3 °C, indicating excellent temperature uniformity. In comparison, the CPCM-1 module had a peak temperature of 56.7 °C and a maximum temperature difference of 2.4 °C, while the natural air convection module had a peak temperature of 69.5 °C and a maximum temperature difference of 3.6 °C. The temperature rise curves of the CPCM-5 module were smoother than those of the air-cooled module. The temperature difference curves were more fitted, meaning that the temperature distribution was more uniform.
I further conducted uninterrupted charge-discharge cycling tests. The battery module was charged at 1.0C and discharged at 2.0C for ten cycles. The CPCM-5 cooling system maintained the battery temperature within the optimal range even at high discharge rates. The peak temperature remained stable at around 48.8 °C, with a maximum temperature difference between adjacent cycles of only 0.7 °C. These results demonstrate that the CPCM-5 cooling system has excellent temperature control and temperature uniformity capabilities, which are beneficial for extending the life of an electric vehicle battery pack and improving its safety.
The heat transfer process in the CPCM can be described by the energy equation:
$$ \rho C_p \frac{\partial T}{\partial t} = \nabla \cdot (k \nabla T) + Q $$
where \(\rho\) is the density, \(C_p\) is the specific heat capacity, \(T\) is the temperature, \(t\) is time, \(k\) is thermal conductivity, and \(Q\) is the heat source term. The effective thermal conductivity of the composite can be estimated by the Maxwell model:
$$ k_{\text{eff}} = k_m \frac{2k_m + k_p + 2\phi_p (k_p – k_m)}{2k_m + k_p – \phi_p (k_p – k_m)} $$
where \(k_m\) is the thermal conductivity of the matrix, \(k_p\) is the thermal conductivity of the filler, and \(\phi_p\) is the volume fraction of the filler. The improvement in thermal conductivity is attributed to the formation of a continuous conductive network by EG and Si\(_3\)N\(_4\). The insulation performance is maintained by the high intrinsic resistivity of Si\(_3\)N\(_4\) and the blocking effect of the network.
In summary, I have systematically studied retired ternary lithium batteries with different nickel contents. I found that the screening process is essential for reusing retired cells from electric vehicle battery packs. NCM811 has the highest capacity but also the poorest thermal stability and the most severe thermal runaway. The microstructure damage after thermal runaway is the fundamental cause of performance degradation. To improve the safety of an electric vehicle battery pack, I developed a high-insulation and high-thermal-conductivity CPCM based on paraffin, expanded graphite, silicon nitride, and epoxy resin. CPCM-5 with 8 wt% Si\(_3\)N\(_4\) exhibits optimal comprehensive performance, including a thermal conductivity of 2.39 W·m\(^{-1}\)·K\(^{-1}\), a volume resistivity of \(8.93 \times 10^{11}\) Ω·cm, a latent heat of 147.14 J/g, and excellent mechanical properties. When applied to a retired NCM811 electric vehicle battery pack module, CPCM-5 effectively reduces the peak temperature and improves temperature uniformity under different discharge rates and cycling conditions. The results provide valuable guidance for the thermal safety design and thermal management of retired ternary lithium batteries in electric vehicle battery packs.
For future work, I suggest studying more types of batteries and more abuse conditions, such as mechanical and electrical abuse, to broaden the applicability of the findings. In addition, the flammability of paraffin should be addressed. The addition of flame retardants may reduce the latent heat, so developing efficient flame-retardant phase change materials is a promising direction. The thermal management system can also be integrated with liquid cooling or heat pipes to further improve the cooling performance of an electric vehicle battery pack. Overall, my research contributes to the safe and sustainable utilization of retired ternary lithium batteries and provides a foundation for advanced thermal management technologies in electric vehicle battery packs.