3.2 Mechanical integrity and melting point characteristics of separators
(1) DSC melting point test DSC results show that the PP|PE|PP separator exhibits two endothermic peaks, corresponding to the melting point of the PE layer (≈152℃) and the melting point of the PP layer (≈167℃); the melting point of the PE separator is ≈128.9℃; the melting point of the PP ceramic separator is ≈131.5℃ – although the ceramic coating does not increase the melting point, it can reduce shrinkage.
(2) Jelly Roll Mechanical Integrity Testing: The jelly roll test results showed significant differences from DSC: The PE separator began to melt at 156°C, and UR1=0.5 V (complete collapse) was reached at 195°C; PP|PE|PP began to melt at 190°C, and UR2=0.5 V was reached at 195°C; PP ceramic began to melt at 190°C, but UR3 did not reach 0.5 V by the end of the test (incomplete collapse). This difference arises from the compact structure of the jelly roll and the heat diffusion effect – the separator is compressed by the electrodes inside the battery, and heat transfer takes time, hence the complete collapse temperature is much higher than the DSC melting point.
More crucially, the complete collapse temperatures of the three separators are all concentrated around 195℃, which is close to the ISC trigger temperatures (215℃, 224℃) in non-CID batteries. This confirms that the complete collapse of the separator is a prerequisite for ISC, and the slight differences in collapse temperatures among different separators provide an explanation for the insignificant differences in subsequent TR.
3.3 Five-stage division of thermal runaway
Based on the ARC test results, this study divides the TR process into five stages:
(1) Stage I: Safe Zone (Test Starting Temperature – Tonset) In this stage, ARC heats at a rate of 5℃ per step, with a temperature rate < 0.02℃/min (ARC sensitivity threshold). The battery does not self-heat, and only experiences passive temperature rise.
(2) Stage II: Anode reaction zone (Tonset-Tventing). At Tonset (the first exothermic temperature), the temperature rate is ≥0.02℃/min, and the battery enters the self-heating stage. After the decomposition of the SEI film, the intercalated lithium comes into contact with the electrolyte and regenerates the SEI film (exothermic reaction). The generation of gas and evaporation of the electrolyte lead to an increase in internal pressure, activating the CID (standard battery), and the voltage is pulled to zero by the parallel resistor.
(3) Stage III: When the internal pressure in the venting area (Tventing-Tonset TR) reaches the rupture threshold, the safety valve opens (Tventing), and high-temperature gas and electrolyte vapor are discharged, causing a temporary decrease in temperature; however, the decomposition of the negative electrode continues, and the temperature gradually rises again.
(4) Stage IV: When the temperature rate exceeds 1°C/s during the acceleration of the TR region (Tonset TR-Tmax), the separator completely collapses, triggering a large-scale ISC. This releases the remaining electrochemical energy, initiating violent decomposition of the positive electrode and electrolyte, causing the temperature to rapidly rise to Tmax (maximum temperature). Some batteries may even combust as a result.
(5) Stage V: Cooling Zone (Tmax – End of Test) When the ARC detects Tmax, maximum temperature rate, or a sudden temperature drop, the cooling system is activated, and the battery temperature gradually decreases.
3.4 Impact of separator on thermal runaway
(1) ARC test results for three types of separator batteries show that the initial exothermic temperature is close (PE 42℃, PP|PE|PP 47℃, PP ceramic 47℃), with the difference stemming from the 5℃ heating step size and sensitivity of the ARC; the trend in CID activation temperature and exhaust temperature is “PE < PP|PE|PP < PP ceramic” (PE: CID ≈ 95℃, exhaust 147℃; PP|PE|PP: CID ≈ 131℃, exhaust 166℃; PP ceramic: CID ≈ 125℃, exhaust 165℃), speculated to be related to the separator thickness (PE 16μm, the latter two 20μm) – a thicker separator may delay gas diffusion and postpone the rate of internal pressure increase; the average temperature at which TR starts to accelerate is 226℃, close to the complete collapse temperature of the separator (195℃), and there is no significant difference in Tmax among the three types of separators (PE ≈ 480℃, PP|PE|PP ≈ 479℃, PP ceramic ≈ 283℃), but the PP ceramic battery did not burn.
(2) In the thermal abuse test comparison (1.5°C/min temperature rise), the temperature curves of the three separator batteries were consistent: due to the heating rate being higher than the self-heating rate of SEI decomposition, the first exothermic temperature was not detected; the CID activation temperature (PE 98°C, PP|PE|PP 131°C, PP ceramic 125°C) and exhaust temperature (PE 141°C, PP|PE|PP 166°C, PP ceramic 165°C) trends were consistent with ARC; the accelerated TR initiation temperature was close (PE 193°C, PP|PE|PP 198°C, PP ceramic 193°C), but the TR of PP|PE|PP and PP ceramic was delayed by 15.2 minutes and 10.7 minutes, respectively, compared to PE; Tmax showed a trend of “PE (438°C) > PP|PE|PP (367°C) > PP ceramic (293°C)”, and only the PP ceramic battery did not burn (supplementary videos V1a-c).
In summary, the type of separator has no significant impact on the TR core stages (acceleration TR, Tmax), but only affects CID activation, exhaust temperature, and combustion behavior through its thickness and coating characteristics. Further research is needed to explore the correlation mechanism between the separator and internal pressure.
4 Conclusion
This study achieved “single variable control” by preparing 18650 laboratory batteries, and systematically revealed the correlation between ISC and TR under thermal abuse conditions for the first time. The main conclusions are as follows:
The essence of voltage drop: The early voltage drop in standard 18650 batteries due to thermal abuse originates from CID activation (disconnection of the positive tab from the top cover caused by increased internal pressure), rather than ISC; in batteries without CID, ISC triggers a voltage drop (≈215℃), and it occurs just a few seconds before the accelerated TR interval, confirming that large-scale ISC is the direct cause of accelerated TR.
Key characteristics of the separators: The complete collapse temperature of the three types of separators (PE, PP|PE|PP, PP ceramic) is around 195℃ (as determined by the jelly roll test), which is significantly higher than the DSC melting point. TR testing indicates that the separator has no significant effect on the accelerated TR initiation temperature or Tmax, and only affects CID activation and exhaust temperature through its thickness.
Innovation in testing method: The developed three-electrode structure (LFP reference electrode) can accurately distinguish between positive and negative electrode potential changes, eliminating CID interference. The mechanical integrity test of the separator, combined with jelly roll compactness and thermal diffusion, is more closely aligned with the actual working environment of batteries.
This study provides a clear direction for the safety design of lithium-ion batteries: there is no need to overly rely on high-priced separators, and the focus should be on optimizing the internal pressure threshold and response speed of CID; meanwhile, the BMS should combine dual signals of temperature (>200℃) and voltage drop to determine ISC, enhancing the accuracy of early warning. Subsequent research will focus on “fixed-point triggering of ISC”, simulating energy release and TR pathways in real-life scenarios.