Comparing Different Thermal Runaway Triggers for Two Automotive Lithium-Ion Battery Cell Types

Inside the reactor, the temperature behavior of the cell is recorded with type-K thermocouples on both sides of the cell surface (type ♯1 and ♯2), the cell tabs (type ♯1) and close to the burst plate of the cell (type ♯2). Additionally, the temperature inside the TR reactor (type ♯1 and ♯2) is measured. Since the pouch cell (type ♯1) design has no defined burst plate like the prismatic metal can cell, three thermocouples were positioned on the welded pouch foil on the long side of the cell. The thermocouple positions for both cell types are shown in Fig. 5. Different thermocouple patterns were used because of the different geometries of both cell types. Safety relevant parameters are:

• ${\bar{T}}_{cell}^{v1}$ (°C) the average measured temperature of all thermocouples on the cell surface when the first venting starts
• ${T}_{cell}^{crit}$ (°C) the temperature of the one cell-surface thermocouple, which is the first to exceed the temperature rate of 10 °C min⁻¹ (detailed description in Ref. 20)
• ${\bar{T}}_{cell}^{v2}$ (°C) the average measured temperature of all thermocouples on the cell surface when the second venting starts
• ${T}_{cell}^{{\rm{\max }}}$ (°C) the maximum recorded temperature of one of the thermocouples on the cell surface
• ${T}_{vent}^{{\rm{\max }}}$ (°C) the maximum recorded temperature of one of the thermocouples at the venting positions
• ${\bar{T}}_{reactor}$ (°C) the average gas temperature inside the reactor.

With GEMS 3300B06B0A05E000 pressure sensors inside the reactor, the pressure increase due to gas generation is recorded during the whole TR experiment. With the ideal gas law, the amount of produced, not condensed vent gas can be calculated. The same calculation is used as presented in Ref. 20 Safety relevant parameters are:

• n_v (mol or liter) the total amount of released gas (in STP: 298.15 K, 100 kPa)
• n_v1 (mol) the amount of gas produced starting at ${T}_{cell}^{v1}$ and ending at the ${T}_{cell}^{v2}$
• n_v2 (mol) the gas produced after ${T}_{cell}^{v2}$ and during the TR
• ̇_ch (mol s⁻¹ or l s⁻¹) the characteristic venting rate based on the minimal duration Δt_50% (s) when 50% of the venting gas n_ch50% (mol) is produced.

The vent gas composition is measured with Fourier Transform Infrared (FTIR) spectrometer and gas chromatograph (GC) in parallel after the TR. The pipes from the reactor towards the gas analysis are closed during the TR experiment and opened after the TR happened. They are heated to 130 °C. Safety relevant parameters are the quantified gas concentrations of gases like: H₂, CO, CO₂, CH₄, C₂H₆, C₂H₄, C₂H₂, DEC, DMC, EC, EMC, H₂O, C₆H₁₄, HF, C₄H₁₀, C₃H₈ and O₂. Since N₂ is not produced by battery failures, N₂ is used as inert gas.

The reactor gas consists of the inert gas N2 and the vent gas, which is added by the failing cell. Since the produced vent gas does not contain N₂, the amount of N₂ in the reactor gas can be subtracted to calculate the concentration of each component of the vent gas only. The concentration of any gas component (c_v/%) in the vent gas is calculated with the measured concentration of this gas component in the reactor gas (c_m) and the measured N₂ concentration (c_N2) in the reactor gas:

$$\begin{eqnarray}&&{{\rm{c}}}_{{\rm{v}}}=\left(\left({c}_{m}\,* \,100\right)/\left(100-{{\rm{C}}}_{{\rm{N}}2}\right)\right)\end{eqnarray} \tag{ 1 }$$

A Bruker MATRIX-MG01 FTIR is used with 0.5 cm⁻¹ wavenumber resolution and N₂(l) cooled MCT detector. The measurement chamber itself is heated to 190 °C and the interior space of the spectrometer is purged with N₂(g) for at least 2 h. For the background measurement 100 scans are averaged. A number of 40 scans are used for each data point.

For gas analysis with GC, the 3000 Micro GC (G2802A) is used with three columns and TCD detectors. The three-channel system includes Molsieve (10 m × 320 μm × 12 μm), Plot U (8 m × 320 μm × 30 μm) and OV1 (8 m × 150 μm × 2,0 μm). The injector temperature and the sample inlet temperature are set to 100 °C for all three channels. The column temperature of the Molsieve channel is 80 °C (at 30 psi) and 60 °C for the Plot U and OV1 channel (40 psi each). Injection time for Molsieve and Plot U is 15 ms and 10 ms for the OV1 channel.

The duration of the venting during TR is defined as the time between the start of the TR (at the second venting when the pressure increase inside the reactor exceeds 200 mbar s⁻¹) until the maximum pressure in the reactor is reached.

After finishing the vent gas composition measurement, the experiment after-treatment is started. This experiment after-treatment consists of the following steps: the reactor is heated up to 200 °C, evacuated to about 1 kPa absolute pressure and flushed with N₂ several times before the reactor is opened again. Because each TR trigger experiment is repeated, for the safety relevant parameters, such as the maximum reached cell surface temperature, the average value of the repeated experiments is presented with a standard deviation.

(Figures 5a and 5d) the temperature of the cell surface increased constantly due to the homogenous heating from both sides of the sample holder. In Table II the thermal parameters are compared for both cell types and all three TR trigger. The first venting ${\bar{T}}_{cell}^{v1}$ of the cell type ♯1 happened at a lower average cell surface temperature (difference of 18 °C) than of cell type ♯2. The second venting ${\bar{T}}_{cell}^{v2}$ for type ♯1 is (204 ± 1) °C, which is higher than for cell type ♯2 at (190 ± 2) °C. The maximum recorded temperature ${T}_{cell}^{{\rm{\max }}}$ is similar for both cell designs. ${T}_{vent}^{{\rm{\max }}}$ for cell type ♯1 reached 584 °C, which is a lower temperature than measured at the cell surface. This is due to the undefined pouch foil opening. In comparison, ${T}_{vent}^{{\rm{\max }}}$ for cell type ♯2 with the defined cell opening reached more than 1200 °C.

Table II.  Thermal parameters of two automotive cell types in overtemperature, overcharge and nail-penetration abuse tests in comparison.

	Overtemperature		Overcharge		Nail-penetration
	cell type ♯1 pouch	cell type ♯2 hard case	cell type ♯1 pouch	cell type ♯2 hard case	cell type ♯1 pouch	cell type ♯2 metal can
${\bar{T}}_{cell}^{v1}$ (°C)	121 ± 1	138 ± 1	56 ± 1	66 ± 9	—	—
${T}_{cell}^{crit}$ (°C)	206 ± 1	192 ± 1	—	—	—	—
${\bar{T}}_{cell}^{v2}$ (°C)	204 ± 1	190 ± 2	82 ± 17	96 ± 4	—	—
${T}_{cell}^{{\rm{\max }}}$ (°C)	819 ± 5	782 ± 50	906 ± 146	579 ± 139	783 ± 1	743 ± 33
${T}_{vent}^{{\rm{\max }}}$ (°C)	584 ± 1	1169 ± 39	1044 ± 271	1021	482 ± 127	850 ± 185

In Fig. 5b for the pouch cell and (e) for the hard case cell, a first venting happened for cell type ♯1 at (56 ± 1) °C and for cell type ♯2 at (66 ± 9) °C. The first venting is detected as a small increase of the temperature signal and cross checked with video recordings inside the test chamber. ${\bar{T}}_{cell}^{v1}$ in overcharge experiments is not as meaningful as in overtemperature experiments, but also shows that cell type ♯2 can withstand higher internal pressure than the soft pouch foil, although the aluminum hard case is a better heat conductor than the pouch foil. The second venting in the overcharge trigger is measured at (82 ± 17) °C for cell type ♯1 and at (96 ± 4) °C for cell type ♯2. In comparison to overtemperature trigger, in overcharge the temperature of second venting decreases due to the destabilization of the cathode and the Li plating at the anode side. The maximum recorded temperature ${T}_{cell}^{{\rm{\max }}}$ is for type ♯1 (906 ± 146) °C and for type ♯2 (579 ± 139) °C. In the case of cell type ♯2 the jelly roll was found outside of the aluminum housing after the experiment. Therefore, the thermocouples on the aluminum hard case measured a lower maximum cell-casing surface temperature because the reacting active material was not inside the case anymore. ${T}_{vent}^{{\rm{\max }}}$ for cell type ♯1 reached in one experiment also temperatures above 1000 °C. ${T}_{vent}^{{\rm{\max }}}$ for cell type ♯2 reached 1021 °C in one experiment and in the second experiment all thermocouples at the vent position were destroyed during the TR and the vent gas temperature could not be measured.

Figure 5c for the pouch cell and (f) for the hard case cell the first venting happened at the nail-penetration itself and both cell designs heated up exothermally immediately after the nail-penetration of the cells. The maximum recorded temperature ${T}_{cell}^{{\rm{\max }}}$ is similar for both cell designs. For cell type ♯1 ${T}_{vent}^{{\rm{\max }}}$ is lower than the maximum recorded cell surface temperature. ${T}_{vent}^{{\rm{\max }}}$ for cell type ♯2 reached again temperatures above 1000 °C.

Independent of the trigger and the cell design, in each experiment the maximum measured cell-case temperature was above 718 °C, which is higher than the melting temperature of aluminum. The highest ${T}_{cell}^{{\rm{\max }}}$ $=\,\left(906\,\pm \,146\right)^\circ {\rm{C}}$ was measured for cell type ♯1 in the overcharge triggered experiment (see Table II and Fig. 7a).

As reported by Pfrang et al.¹⁸ we observed that metal can cells can withstand a higher internal pressure until they open compared to pouch cells. This can be observed in overtemperature and overcharge experiments. We also measured the maximum expansion of the cell during the experiment against the spring force of the sample holder with a micrometer screw. Then we calculated the force with which the cell presses against the upper pressure plate. This force divided by the area of the cell gives the pressure. This pressure was evaluated for the pouch cells and the metal can cells. For the pouch cell we observed values between 223 kPa–411 kPa and for the metal can cell 386 kPa–855 kPa before the first venting. With video recordings we can prove that the maximum expansion of the cell is reached before the first venting. These observed pressure values also prove, that the pouch cells open at a lower internal pressure than the metal can cells. The temperature at the first venting of the pouch cell is consistent with the temperature of 120 °C reported by Ren et al. on overheated 24 Ah pouch cells in Ref. 35. When the burst plate of the hard case cells opens, at the first venting gas is ejected. At the second venting and TR at both cell types, gas and particles with high temperatures are ejected.³ The second venting ${\bar{T}}_{cell}^{v2}$ and the critical temperature ${T}_{cell}^{crit}$ for cell type ♯1 was observed in overtemperature trigger at a higher temperature than for cell type ♯2. This indicates that cell type ♯1 can withstand overtemperatures longer than type ♯2. Reason for the higher ${T}_{cell}^{crit}$ and ${\bar{T}}_{cell}^{v2}$ for cell type ♯1 might be the different delithiation percent of the cathode and consequently the cathode stability, the stacking of the active layers or the earlier opening of the pouch foil. The percent of delithiation of the cathode and lithiation of the anode at 100% SOC are not known and are not analysed in this study. What is known is, that both cell types have NMC cathode and graphite anode and are series products from two different cell manufacturer. The two cells types have different electrolyte composition, different packaging material, the active material is stacked inside the pouch foil and rolled inside the hard case.

${T}_{vent}^{{\rm{\max }}}$ for cell type ♯2, the hard case cell, reached in all three trigger types higher maximum values than the pouch cell design type ♯1. This can be explained due to the defined venting of the hard case cell. The vent gas temperature can be measured more easily for hard case cells with a defined burst plate, than for pouch cells, because the pouch foil can open at all pouch welded sides simultaneously. Nevertheless, the nail-penetration experiment showed that the vent gas of the pouch cell (type ♯1) can also reach temperatures above 1000 °C. It needs to be mentioned that the maximum measured vent gas temperature might also be affected by the ejection of hot particles.

The maximum recorded temperature ${T}_{cell}^{{\rm{\max }}}$ does not vary significantly for both cell types. But comparing the average ${T}_{cell}^{{\rm{\max }}}$ for all three triggers in Figs. 7a and 7e the cell type ♯1 reached higher values than cell type ♯2, especially in the overcharge experiment. Huang et al. also observed in overcharge experiments that the pouch cell reaches higher maximum temperature on the cell surface than the prismatic cell.¹⁹ Additionally, the slightly higher ${T}_{cell}^{{\rm{\max }}}$ for cell type ♯1 might also indicate the influence of the higher energy density of cell type ♯1 compared to type ♯2. The influence of the energy density on the ${T}_{cell}^{{\rm{\max }}}$ is published in Ref. 36.

Although several thermocouples on the cell surface are used in the experiment, the thermocouple positions in our test setup are limited in amount and in position. At some experiments the thermocouple is close to the hottest position of the cell, and sometimes not. The hottest position on the cell surface is not known before the experiment. Therefore, sometimes a higher deviation of the recorded ${T}_{cell}^{{\rm{\max }}}$ and also ${T}_{vent}^{{\rm{\max }}}$ from one experiment to the other experiment was observed. Other temperature values such as ${\bar{T}}_{cell}^{v1},$ ${\bar{T}}_{cell}^{v2}$ and ${T}_{cell}^{crit},$ have a good reproducibility (see Table II).

The pressure inside the reactor increased slowly after the first venting and opening of the cells ending up in n_v1 = 0.15 mol of gas before the TR. The continuous pressure increase is due to the ongoing evaporation of electrolyte out of the opened cell. During the main TR reaction, both cell types released significant additional amount of gas and the pressure inside the sealed reactor increased within seconds up to a maximum value shown in Fig. 6a for cell type ♯1 and (d) for cell type ♯2. The duration of the venting during TR of cell type ♯1 was (3.5 ± 0.1) s. In this time the main amount of gas was produced with a characteristic venting rate of ̇_ch = (34 ± 2) l s⁻¹. The cell type ♯2 released gas in less time (2.0 ± 0.1) s than type ♯1. The characteristic venting rate of cell type ♯2 was therefore higher ̇_ch = (67 ± 4) l s⁻¹. Cell type ♯1 produced in total (1.56 ± 0.04) l Ah⁻¹ in the overtemperature experiment. For cell type ♯2 also (1.56 ± 0.05) l Ah⁻¹ were measured. The produced gas amount fits to reported values in literature between 1.2–2 l Ah^−117,22,23 for overtemperature triggered NMC cells.

Zoom In Zoom Out Reset image size

The first venting ended up in n_v1 = 0.13 mol of gas for both cell types. During the main TR reaction, both cell types released significant amounts of gas and the pressure inside the sealed reactor increased up to a maximum value shown in Fig. 6 for (b) cell type ♯1 and (e) cell type ♯2. The duration of the venting during TR for cell type ♯1 was (5.2 ± 0.8) s with a characteristic venting rate of ̇_ch = (47 ± 4) l s⁻¹. The cell type ♯2 released during the TR in (1.1 ± 0.1) s gas with consequently much higher characteristic venting rate of ̇_ch = (250 ± 56) l s⁻¹. Cell type ♯1 and ♯2 produced in total significantly more gas during overcharge than during overtemperature. Cell type ♯1 released (2.79 ± 0.02) l Ah⁻¹ and cell type ♯2 (2.65 ± 0.06) l Ah⁻¹.

Both cell types released significant amounts of gas immediately after the nail penetrated the cell. The pressure inside the sealed reactor increased up to a maximum value shown in Fig. 6 for (c) cell type ♯1 and (f) cell type ♯2. The duration of the venting during TR of the stacked cell type ♯1 in nail-penetration was (3.1 ± 0.9) s, which was faster than in overtemperature or overcharge. The characteristic venting rate of cell type ♯1 in nail-penetration is therefore ̇_ch = (182 ± 65) l s⁻¹. The cell type ♯2 behaves differently than the pouch cell, because the gas release took (11.5 ± 2.5) s during the TR. In Fig. 6f a stepwise increase of the pressure inside the reactor can be seen. The characteristic venting rate of cell type ♯2 in nail-penetration is ̇_ch = (140 ± 76) l s⁻¹. Cell type ♯1 produced in total (1.71 ± 0.07) l Ah⁻¹ in nail-penetration experiments and cell type ♯2 also (1.77 ± 0.03) l Ah⁻¹. In nail-penetrated TR, less gas is produced than in the overcharge triggered TR, but more gas than in the overtemperature triggered TR (see Figs. 7b and 7f). Therefore, we do not confirm the statement of Diaz et al., that for nail-penetrated cells lower vent gas amounts are produced than in thermal abused cells. The divergence of the observations might be because Diaz et al. did not observe a TR.

Zoom In Zoom Out Reset image size

Note, that since the characteristic venting rate is defined as the minimum duration when 50% of the gas is produced during the TR, the numeric value of the characteristic venting rate can exceed the numeric value of the maximum amount of released gas. The minimum duration to produce 50% of the vent gas during the TR is plotted in Fig. 7c for cell type ♯1 for all three TR trigger in comparison and in Fig. 7g for cell type ♯2.

The TR duration (pressure increase during the TR) is in overtemperature and overcharged experiments for cell type ♯1 longer than for cell type ♯2. In Fig. 6e the pressure increases inside the reactor during overcharge of cell type ♯2 (hard case). For the hard case cell, the pressure increase is not as smoothly as it is for the pouch cell, and two peaks are visible before reaching the maximum pressure. This observation of two peaks can be seen even more clearly for the nail-penetrated hard case cell in Fig. 6f. The nail-penetrated 8 mm into each cell. The cell thickness of the cell type ♯2 is 28 mm; more than twice the nail-penetrated length. At the disassembling of the test setup after the TR, for cell type ♯2 two jelly rolls were visible inside the hard case. Therefore, the stepwise increasing pressure for the nail-penetrated hard case cell indicates that the penetrated jelly roll was triggered into the TR first, and the TR propagated to the second jelly roll after some seconds. A closer look to the thermocouples positioned on the bottom side of the cell confirms that the temperature at the bottom side (opposite to the nail-penetration position) increased some seconds after the temperature on the top of the cell, where the nail penetrated the cell.

The measured vent gas amounts per Ah of cell type ♯1 and cell type ♯2 are very similar for both cell types, if the TR trigger is the same! In this case, both cell types have the same capacity (Ah) and also similar cathode and anode chemistry, but different electrolyte composition (see Table I). Therefore, we assume that the exact electrolyte composition does not influence the produced vent gas amount. The minimum duration to produce 50% of the vent gas during the TR is different for both cell types (Figs. 7c and 7g). For cell type ♯1 the venting time for 50% of the vent gas is higher than for cell type ♯2, especially in the overtemperature and the overcharge trigger. Also, the duration of the venting during TR for cell type ♯1 in overtemperature and in overcharge is longer than for the hard case cell. This also results for cell type ♯1 in lower characteristic venting rates and less intense pressure peaks during the TR. In the nail-penetration experiment the duration of the venting during TR for cells with more than one jelly roll is longer, but a high characteristic venting rate is still possible.

Therefore, the cell and venting design as well as the packaging of active material layers seems to influence the reaction time of the TR, and consequently the safety relevant parameters such as the characteristic venting rate and maximum pressures reached inside a sealed volume.

The main gas components are in descending order for cell type ♯1: 30% CO₂, 26% CO, 15% H₂, 10% C₂H₄, 5% CH₄, 4% H₂O, 3% EMC and below 1% C₄H₁₀, C₂H₆, C₂H₂. Cell type ♯2 produced in descending order: 30% CO, 21% CO₂, 15% H₂, 5% C₂H₄, 5% CH₄, 4% H₂O, 2% DMC, 2% C₄H₁₀, and lower 1% EMC, C₂H₆, C₂H₂, C₆H₁₄ (see Fig. 8 red bars). Differences in gas concentrations from both cell types are observed in C₂H₄, CO, CO₂ and DMC. These differences will be discussed in the comparison of the results of both cell types. The average substance concentration values published by Koch et al. in Ref. 17 over 51 overtemperature triggered NMC LIBs fit to the presented results in this paper: 28% CO, 37% CO₂, 22%H₂, 6% C₂H₄ and 5% CH₄.¹⁷ Our results are within the error bars of each gas component presented in Ref. 17, Fig. 4. The gases produced during the TR can be explained by SEI decomposition,^18,38 electrolyte decomposition,^39,40 NMC (cathode) degradation and reaction of the solvent with the cathode.^27,32

Beside higher amounts of gas than in overtemperature trigger both cell types produced significant higher amounts of CO and H₂, and a little higher amount of CO₂ and CH₄. To emphasize the higher amounts of CO, H₂ in overcharge experiments compared to overtemperature, Fig. 9 presents the main gas composition in mol for all three TR trigger. Cell type ♯1 produced in overcharge two times as much CO ((2.27 ± 0.05) mol) and three times as much H₂ ((1.77 ± 0.03) mol) than in overtemperature and nail trigger. Cell type ♯2 also produced 1.6 times higher amounts of CO ((1.93 ± 0.06) mol) and 2.8 times higher H₂ ((1.71 ± 0.04) mol) in overcharge experiments compared to overtemperature. In Fig. 8 yellow bars the relative vent gas composition is presented. Additional to the above-mentioned reactions producing gases, the lithium metal deposit on the anode surface reacts with the electrolyte under generation of heat and gas.³³ According to Ohsaki et al. CO and CO₂ were mainly produced at the cathode side during overcharge and on the anode side the main component is H₂ and small amounts of CH₄, C₂H₄, C₂H₆, CO and CO₂ were measured⁴¹.

Zoom In Zoom Out Reset image size

Both cell types produced the lowest CO volume percent values and the highest C₂H₄ value of all three triggers. The H₂ values are lower than in the overcharge experiment, but higher than in overtemperature (see grey bars in Fig. 8 and Fig. 9). CO₂ values are increased for the cell type ♯2 compared to the overtemperature trigger. Diaz et al. observed for nail-penetrated cells without TR less toxic gas than in thermal abuse using cells with the same state-of-charge (SOC) (100%).¹² Although Diaz et al. refers to gas results without TR, we also observed lower CO values in the nail-penetration experiments than in overtemperature experiments.

The gas compositions measured in the repeated experiments show high agreement and for most gas components deviations between the two measured values were below 3%. Exceptions are in some experiments the values of H₂O, higher hydrocarbons with low concentrations such as C₂H₂, C₂H₄ and C₂H₆ and in nail-penetration tests the CO concentration.

We assume that the trigger and consequently the reaction mechanism and the reached temperatures influences the decomposition reactions and the resulting vent gas composition. The influence of temperature inside the cell during the reactions on the gas composition was also stated by Fernandez et al. in Ref. 13.

For both cell types significant amounts of CO were produced, especially in overtemperature and overcharge experiments. In overcharge experiments the CO amount was higher than the CO₂ amount. The observation of CO₂:CO ratio is less than one can be explained by the high gravimetric energy density of both cell types. As observed by Koch et al. the CO₂:CO ratio decreases with increasing gravimetric energy density of the cells.¹⁷ This can also be observed in the overheated LCO/NMC cell by Golubkov et al., which produced significant higher CO values than the NMC cells with lower energy density.³⁷

The exact gas composition varies with different electrolyte composition. In this case, the active material of both cell types is very similar, but different electrolyte compositions are used. Internal investigations of car manufacturers reveal that in cell type ♯1 the main electrolyte components are EC:EMC (1:1) and in cell type ♯2 EC:DMC:EMC (2:3:3). In Figs. 8a and 8b different electrolyte vapor is identified for both cell types. For cell type ♯1 EMC was identified, but no EC could be identified. EC is very unlike to be measured with the presented gas analysis setup. We assume that parts of EC decompose and parts condensate inside the TR reactor or before the gas analysis section, because the boiling point of EC is higher than the gas temperature inside the reactor. For cell type ♯2 DMC and EMC could be measured. Typical decomposition reactions of electrolyte components such as EC are CO₂, CO, C₂H₄, EMC, DEC,^29,39 for EMC: CO₂, CO, DMC, DEC, for DEC CO₂, C₂H₆,³⁹ and for DMC: CO, CO₂, CH₄, C₂H₆, H₂O.^29,31,38 Major varying gas concentrations between both cell types are observed in C₂H₄, CO, CO₂ and DMC values. Onuki et al. observed that C₂H₄ was only formed from EC.³⁹ Cell type ♯1 produced significant higher C₂H₄ concentrations at all three triggers (twice as much than cell type ♯2) and also CO₂ values were higher than for cell type ♯2. This observation correlates with the significant higher amount of EC in cell type ♯1 (twice as much) compared to cell type ♯2. Therefore, we assume that the higher C₂H₄ amount in the vent gas composition of cell type ♯1 is because of the higher amount of decomposed EC. Only minor varying gas concentrations were measured in H2 and CH₄ values.

In the presented results no hydrogen fluoride (HF) was identified with the FTIR spectrometer inside the FTIR gas measurement chamber. Although it is assumed that small amounts of HF were released by the cells according to,^42,43 but the highly reactive HF reacted with materials inside the test reactor, the analysis pipes and the ejected particles. Additionally, we assume that modern mass produced EV cells produce less HF than older cells. We measured inside our test setup for an older aged 18 Ah cell with NMC/LTO chemistry 66 ppm (0.396 mmol) HF after the TR²¹ and for another modern automotive pouch cell 0 ppm HF.³

At each tested TR trigger cell type ♯1 lost higher amounts of mass than cell type ♯2. This observation is not consistent with the identified higher mass loss of hard case cells compared to pouch cells of Koch et al. in Ref. 17. Koch et al. explains the observation of higher mass loss at hard case cells with the higher stream velocities through the burst plate.¹⁷ We assume, that the different casing and the different gravimetric energy density influences the mass loss. For cell type ♯2, the metal casing has a higher mass than the pouch foil and this metal casing contributes to the measurement of the weight. Beside the casing, the pouch cell type ♯1 might lost additionally higher amounts of mass than the hard case cell type ♯2, because cell type ♯1 has a higher gravimetric energy density.

In each experiment independent of the trigger and the cell design the maximum measured temperature was above 718 °C. Under consideration of the standard deviation of measured temperature values and different cell design all three TR trigger resulted for both cell geometries in similar ${T}_{cell}^{{\rm{\max }}}$ and ${T}_{vent}^{{\rm{\max }}}.$ In overtemperature and in nail-penetration the SOC is 100% and therefore, we assume that the delithiation of the cathode and the lithiation of the anode are the same. Here the ${T}_{cell}^{{\rm{\max }}}$ values are comparable. Cell type ♯1 reached in nail-penetration the lowest ${T}_{cell}^{{\rm{\max }}}$ values and in overcharge the highest (see Figs. 7a and 7e). Cell type ♯2 reached the lowest ${T}_{cell}^{{\rm{\max }}}$ values in overcharge experiments due to the ejection of the jelly roll during the overcharge and the highest value in overheated TR.

Cell type ♯1 and ♯2 produced in total significantly more gas in overcharge than in overtemperature and nail-penetration experiments. Overtemperature triggered cells produced the lowest gas amounts of the three tested triggers. For the amount of produced gas for overtemperature triggered NMC cells in literature values between 1.2–2 l Ah⁻¹ are reported.^17,22,23 The measured values of both cell types in this publication fit into the reported gas amount per liter for heat trigger. No literature was found presenting the gas amount per Ah at NMC cells triggered by overcharge or nail-penetration. But the presented experiments show that overcharge trigger forces the batteries to produce significantly more gas per Ah than using other trigger methods. The venting rates ̇_ch are for overtemperature TR lower than for overcharge or nail-penetration trigger. Concerning the venting rate ̇_ch nail-penetration trigger as well as overcharge for hard case cells might challenge closed battery packs most.

The trigger influences the vent gas composition, especially the CO, CO₂, H₂, C₂H₄, CH₄, and electrolyte concentrations. Beside higher amounts of gas are produced in overcharge tests, for both cell types the highest H₂ values are measured in the overcharge failure case and the least H₂ values in the overtemperature trigger. Nail-penetration trigger forced both cell types to produce higher of C₂H₄ and CO₂ vol% values and lower CH₄ vol% values that in the other two triggers.

The highest mass loss was observed for overcharged cell and the lowest mass loss for nail-penetrated cells.