A Case Study on Gas Venting Events in NCM523 Batteries During Thermal Runaway Under Different Pressures in a Sealed Chamber ^†

Abstract

The venting process is one of the most important events during the thermal runaway (TR) of lithium-ion batteries (LIBs) in determining fire accidents, while different ambient pressures will exert an influence on the venting events as well as the TR. Ternary nickel–cobalt–manganese (NCM) batteries with a 75% state of charge (SOC) were employed to conduct TR tests under different ambient pressures in a sealed chamber with dilute oxygen. It was found that elevated ambient pressure results in milder ejections in terms of jet temperature and mass loss. Gas venting characteristics were also obtained. Additionally, the amount of carbon dioxide (CO₂), hydrogen (H₂), methane (CH₄), and ethylene (C₂H₄) released increase with ambient pressure, while carbon monoxide (CO) varies inversely with ambient pressure. The higher the ambient pressure is, the greater the flammability risk is. The molar amount of C, H, O, and total gases released shows a positive correlation with the maximum battery temperature and ambient pressure. This study will support the design of safety valves and help reveal the effects of venting events on the evolution of TR.

1. Introduction

LIBs have been widely employed in the field of new energy vehicles and energy storage power stations operating at various altitudes because of their huge advantages, such as their long life cycles, high specific energy density, and low self-discharge rate [1,2,3]. However, under abuse conditions, including electrical abuse, thermal abuse, and mechanical abuse, LIBs may suffer from TR, one of their most important safety issues [4,5,6]. The TR of LIBs has attracted public attention, and establishing how to mitigate or prevent thermal hazards is becoming more and more crucial. During the process of TR, a venting event is another way that matter and energy can be exchanged between the interior of the battery and the external environment, resulting in the release of a large number of combustible gases [5,6,7,8]. In addition, vented particles at elevated temperatures are also potential sources of arc failure in the battery system, as they could ignite combustibles and possibly lead to fire accidents or explosions [9,10]. Local atmospheric pressures can vary with different altitudes; therefore, venting events should be different during the process of TR.

The venting process is a necessary pressure relief design in order to avoid the collapse of the battery shell. When a battery is involved in the evolution of TR, several exothermic reactions may occur successively among its cathode materials, anode material, electrolyte, and other components [4,6]. Gases released from these reactions will accumulate gradually inside the sealed battery until the battery valve opens, resulting in a venting event, when the internal pressure of the battery reaches a certain value [11]. Some solid materials and unreacted liquid electrolyte are eventually ejected out of the battery along with these gases during the process of venting. These ejections possess a large amount of heat and flammability and are considered a source of extremely hazardous combustions or explosions [12,13]. Therefore, research on the venting behavior of batteries is of great importance.

There has been abundant research on the characteristics of venting behaviors, such as the composition of ejections and the total amount of gas vented. Vented ejections were described as appearing gaseous, liquid, and solid based on the sampling temperature and detection temperature [11,14,15,16]. The TR gas analysis of a 50 Ah prismatic NCM622 battery with a 100% SOC revealed at least 31 chemical compositions in the vented ejections, which were discovered in a sealed chamber initially filled with nitrogen, while 17 compounds were found to have a boiling temperature below 25 °C, accounting for approximately 90% of the detected gases [15]. Most of the vented gases were H₂, CO, CO₂, and short-chain hydrocarbons such as CH₄, ethane (C₂H₆), and C₂H₄. TR tests of 51 ternary nickel–cobalt–manganese (NCM) batteries in a gas-tight autoclave filled with air were conducted and the seven most common gas species—H₂, CO, CO₂, CH₄, C₂H₆, C₂H₄, and propylene (C₃H₆)—were summarized, with these making up over 99% of the total gases released [17]. Similar conclusions were also reached about the gas compositions of nickel–cobalt–aluminum (NCA), lithium cobalt oxide (LCO), and lithium iron phosphate (LFP) batteries at different SOCs in a sealed vessel that was filled with nitrogen before the TR tests [16,18]. Besides the gaseous and liquid ejections, metallic and nonmetallic elements were found in the vented particles, most of which were Ni, Co, Mn, Li, Al, Cu, C, H, P, and O [9,19,20,21,22]. In terms of the total amount of gas released, roughly 0.03 mol/Ah, 0.08 mol/Ah [16,18], and 0.09 mol/Ah [17] were obtained from LFP, NCA, and NCM batteries with a 100% SOC at normal temperature and pressure, respectively. By taking most of the species of vented gases into account and adopting Le Chatelier’s mixing rule, upper and lower flammability limits were calculated to evaluate the flammability characteristics of LIBs with different cathodes [23,24,25].

A venting event is mainly shaped by pressure differences between the inside and outside of the battery. Therefore, different ambient pressures should result in different gas venting behaviors during TR. TR tests of 2.6 Ah LCO batteries with 0, 50%, and 100% SOC under 64.3 kPa and 100.8 kPa were carried out in an open environment and their jet behavior and combustion hazards were characterized [26]. The results showed that a lower pressure could lead to a faster ignition and a smaller mass loss and heat release rate during TR. Nevertheless, venting behaviors in the open environment may depend on the content and concentration distribution of oxygen to some extent, because oxygen can participate in several reactions during the venting event.

Although sufficient efforts have been made towards studying venting behaviors with regard to battery type, SOC, capacity, etc., limited studies have investigated venting behaviors under different pressures. Furthermore, a comprehensive understanding of the whole venting process without the influence of oxygen is yet to be achieved.

In this study, TR tests of 50 Ah NCM523 batteries with a 75% SOC were carried out in a sealed chamber filled with nitrogen. The tested batteries were prompted to enter TR by lateral heating and the ambient pressure inside the chamber was set to 20 kPa, 100 kPa, or 260 kPa of absolute pressure. Thermocouples (TCs) and pressure sensors were adopted to monitor thermodynamic parameters. After the TR tests terminated, the released gases were collected and analyzed by a gas chromatography analyzer (GC). Thermal parameters, venting characteristic parameters and gas compositions were mainly adopted to describe gas venting behaviors. This research will benefit pressure relief design of the safety valves and provide reference to battery safety evaluation for application scenarios at different altitudes.

2. Experiment

Compared to pouch cells, the safety valve on prismatic batteries enables a relatively vertical jet flow during gas discharge. Moreover, the burst pressure of the safety valves in the same batch of batteries is consistently uniform, ensuring reliable performance under critical conditions [11,27,28]. In this study, prismatic batteries were selected to investigate venting behaviors. As shown in

Table 1. Information of the tested battery.

Parameters	Value
Nominal capacity (Ah)	50
State of charge (%)	75
Nominal weight (g)	865
Size (mm)	148 × 27 × 102
Type	Prismatic
Cathode	Li(Ni0.5Co0.2Mn0.3)O2
Anode	Graphite
Specific energy density (Wh/kg)	212
Ambient pressure (kPa)	20, 100, 260

, the tested battery with a Li(Ni_0.5Co_0.2Mn_0.3)O₂ cathode has a nominal capacity of 50 Ah, a nominal weight of 865 g, and a specific energy density of 212 Wh/kg. Before each test, the examined battery was discharged to 0% SOC and then charged to 75% SOC using a constant current mode. If the SOC of the examined battery is significantly below the required threshold (e.g., 0% SOC), violent TR reactions will not occur even if a venting event takes place. By contrast, if the SOC is extremely high (e.g., 100% SOC), TR will occur once the safety valve opens, which complicates the analysis of the relationship between the venting event and the TR.

To comprehensively understand the characteristics of the entire venting process in LIBs, TR tests were conducted on 50 Ah NCM523 batteries with a 75% SOC within a sealed chamber filled with nitrogen. It facilitates the precise adjustment of ambient pressure and, to a significant extent, prevents subsequent reactions with oxygen, thereby enabling the observation of the intrinsic venting event. During the TR process, combustible emissions released from LIBs pose a high risk of fire accidents or explosions. Given that the oxygen concentration and its distribution can significantly influence the combustion process, different operation conditions are expected to lead to varying thermal hazard behaviors. Therefore, an inert and sealed piece of equipment was constructed to avoid the effects of oxygen and enhance comparability of test results. The tested batteries were triggered to undergo TR by lateral heating with a heating power of 400 W. Prior to each test, the chamber was evacuated to a pressure of 20 kPa to control the residual oxygen content. Subsequently, the internal pressure was adjusted to the desired value. To simulate diverse pressure conditions, the ambient pressure inside the chamber was set to 20 kPa, 100 kPa, and 260 kPa in terms of absolute pressure, respectively. TCs and pressure sensors were adopted to monitor the thermodynamic parameters inside the chamber, as well as the jet temperature. Both pressure and temperature were detected throughout the entire process in order to calculate the total gas amount. In addition to analyzing the variations in battery temperature and jet temperature, gas compositions as well as mass loss ratios were investigated at different pressures. By introducing a real-time gas amount calibration method, the venting characteristics were systematically evaluated.

As shown in Figure 1, a sealed and gas-tight chamber with a volume of 230 L was adopted to carry out TR tests and collect the vented gases. On the outside surface of the chamber, at least 3 K-type TCs were arranged to measure the battery temperature, the jet temperature (approximately 30 mm above the safety valve) and the ambient temperature, while pressure sensors with a full-scale error of 0.5% were installed to monitor the internal pressure of the sealed chamber. The tested battery was placed in the chamber by a fixture initially. A total torque of 4 N·m was applied uniformly at the four corners to create a consistent gap structure between the tested battery and the fixture for each test, in order to ensure that the heat conduction and heat dissipation conditions remained nearly unchanged. After the chamber was locked, a vacuum pump and a pure nitrogen cylinder were utilized to exchange the atmosphere inside the chamber. Firstly, the air inside the chamber was slowly removed with the vacuum pump until the vacuum degree reached 0.8. A few minutes were required to reach an equilibrium state, and the chamber was filled with fresh nitrogen with the pure nitrogen cylinder to the given pressure, typically atmospheric pressure. The atmosphere exchange operation was repeated to control the oxygen concentration below 1%. Then, the heating plate was turned on to initiate the thermal runaway process in the examined battery. The signals of the temperature and pressure were recorded with a data acquisition instrument for the whole process with a frequency of 1 Hz. When the venting event eventually terminated, gases inside the chamber were collected with gas collection bags for subsequent gas analysis. The gas chromatography analyzer, Trace 1300 (Thermo Fisher, Singapore), was primarily calibrated for H₂, CO, CO₂, CH₄, C₂H₆, C₂H₄, and C₃H₆, which were the most common gas species of the vented gases. Two thermal conductivity detectors (TCDs) were utilized to detect common permanent gases, while two flame ionization detectors (FIDs) were employed to analyze short-chain hydrocarbons. Argon (Ar) and helium (He) served as carrier gases. By calculating the total gas amount, the influence of ambient pressures on elemental contents of C, H and O in the vented gases were also explored. Additionally, flammability characteristics were analyzed based on gas analysis.

3. Results

3.1. Thermal Behavior

It has been observed that the beginning of a venting event during the process of TR is primarily governed by the pressure difference between the inside and outside of the battery, because the burst pressure of the safety valve is approximately consistent. Consequently, the venting event will appear earlier at a lower ambient pressure. However, owing to manufacturing variations in the burst pressure of the safety valve and potential discrepancies in the reaction process under varying pressure, the evolution of internal pressure inside the battery may exhibit slight differences. Therefore, it is crucial to recognize that the start time of venting events is not the only judgement for determining venting behaviors.

As the tested battery was laterally heated to the TR, it is evident that the battery temperature on the heating side is more sensitive than that on the opposite side. Figure 2 presents the temperature curves on the heating side of the tested batteries under different pressures. For comparison, the moment when the heating plate was activated was set to t = 0 s. Approximately 20 min elapsed before the venting event occurred. The tested battery under 260 kPa reached the maximum temperature (827 °C) of the TR at t = 1381 s, while the maximum temperature of the tested battery under 100 kPa achieved the lowest value (749.9 °C) at t = 1447 s. The maximum temperature shows a positive correlation with ambient pressure, suggesting that a higher pressure results in a more violent TR, consistent with the previous results [18]. Compared with the case at 20 kPa, although the tested battery under 100 kPa achieved the maximum temperature at the latest, the maximum temperature was elevated. Meanwhile, the onset of the venting event under 20 kPa was significantly earlier than the other two cases. Hence, it may imply that the lower burst pressure of the safety valve can mitigate the hazards of the TR and provide earlier warning signals, such as characteristic gases. However, the internal pressure of the examined battery will increase to a certain degree during normal cycles. As a consequence, an optimal value of the burst pressure should exist in terms of early warning and hazard prevention for the TR.

Figure 3 illustrates the variations of jet temperature approximately 30 mm above the safety valve under different pressures. It is worth noting that it is difficult to avoid measurement errors as a result of the location of the measurement point and the heat capacity of the TCs. For the case at 20 kPa, jet temperature was first detected at t = 1105 s and reached the first peak (113.5 °C) at t = 1194 s and the second peak (309.2 °C) at t = 1368 s. For the case at 100 kPa, the jet flow appeared at t = 1209 s and its temperature reached the first peak (107.7 °C) at t = 1237 s and the second peak (239.4 °C) at t = 1440 s; for the case at 260 kPa, the significant rise in jet temperature appeared at t = 1201 s and reached the first peak (135.8 °C) at t = 1209 s and the second peak (200.2 °C) at t = 1305 s. It seems that the maximum jet temperature decreases with the elevated ambient pressures. The reason why all cases exhibited two peaks in the jet temperature curves was that two ejections occurred during the venting process. During the first ejection, the saturated vapor pressure of electrolyte contributed the most, but TR was not triggered. With the elevated battery temperature, the main triggering reactions commenced and released a great number of gases; thus, the second ejection took place. The time interval between the second peak and the onset of the jet was employed to describe the response time, which varied conversely with the ambient pressure, namely 263 s under 20 kPa, 231 s under 100 kPa and 104 s under 260 kPa. The variations of jet temperature implied an earlier warning and a longer response time under a lower pressure. Characteristic temperatures of the tested battery and the jet during the TR are listed in

Table 2. Characteristic temperature of the battery and jet field during TR.

Parameters	P0 = 20 kPa	P0 = 100 kPa	P0 = 260 kPa
Maximum battery temperature (°C)	749.9	778.4	827
Maximum jet temperature of 1st peak (°C)	113.5	107.7	135.8
Maximum jet temperature of 2nd peak (°C)	309.2	239.4	200.2

.

3.2. Gas Venting Behavior

3.2.1. Evolution of Pressure in the Chamber

For the lithium-ion battery with an SOC near 50%, TR may not occur when the safety valve opens, because the battery temperature is still below the critical point to trigger TR reactions. Therefore, the burst of the safety valve and the TR event lead to two distinct venting processes. The first venting event is primarily driven by the increasing internal pressure of the battery as a consequence of continuous electrolyte evaporation, while the second venting event is probably induced by violent reactions of the TR. The battery samples adopted in this study conform to this scenario.

The pressure variations in the sealed chamber represent the intensity of the venting process to a degree. Figure 4 illustrates the diagrammatic sketch of a typical pressure curve of the examined batteries and defines several parameters to describe the whole venting process from the aspect of pressure variations inside the chamber. Prior to the opening of the safety valve, the pressure within the chamber is approximately equal to the initial pressure, $P_0$. Similar to Figure 3, there are two pressure peaks, namely $P_{\mathrm{P}1}$ and $P_{\mathrm{P}2}$, which may result from the dynamic pressure and the rapid cooling. After each pressure peak, the pressure inside the chamber gradually approaches a constant value, $P_1$ and $P_2$. The pressure variation between the peak pressure and the onset pressure is defined as the pressure impact to represent the venting intensity, such as ${∆P}_{\mathrm{P}1}$ and ${∆P}_{\mathrm{P}2}$. Two parameters calculated at the stable stage, ${∆P}_1$ and ${∆P}_2$, are positively correlated with the amount of gas released and represent the pressure effects of each venting process.

Figure 5 presents the four parameters, ${∆P}_1$, ${∆P}_2$, ${∆P}_{\mathrm{P}1}$ and ${∆P}_{\mathrm{P}2}$, against ambient pressure inside the chamber. During the first venting process, exothermic or gas-producing reactions were weak, and as a result, only a few gases were released from the battery. Hence, the pressure effects of the first venting event under various ambient pressures were significantly smaller than those of the second venting event. It should be noted that the interval between the two venting events was much shorter for the case of 260 kPa. This could explain why ${∆P}_1\mathrm{a}\mathrm{t}{P}_0=260\mathrm{k}\mathrm{P}\mathrm{a}$ was lower. In terms of pressure effects, ${∆P}_1$ was only 9.3%, 9.5%, and 6.8% of ${∆P}_1$ for 20 kPa, 100 kPa, and 260 kPa, respectively, implying that most of the vented gases were produced by TR reactions and the pressure rise of the first venting event could hardly serve as a signal for the early warning of TR. However, organic electrolyte vapor and a small number of vented gases may serve as better indicators for predicting the occurrence of TR. In contrast, there is a great difference in the pressure impact between the two venting events. For the first venting peak, the pressure impact ${∆P}_{\mathrm{P}1}$ was approximately 2~3 times the pressure effect ${∆P}_1$, while the value was approximately 1.5~3 for the second venting peak. The venting intensity of the second venting event was much more violent than that of the first venting event. Dynamic pressure shocks and ejections with the elevated temperature significantly contribute to the second pressure peak. In addition to the pressure rise due to the accumulated gases during the venting process, it is necessary to consider the margin of safety design to improve structure design and pressure relief design of the battery pack in consideration of pressure impacts. It is evident from Figure 5 that the pressure effect is considered as a weakly positive correlation with ambient pressure, while the pressure impact is strongly correlated with ambient pressure. The influence of ${∆P}_{\mathrm{P}2}$ is greater than that of ${∆P}_{\mathrm{P}1}$. At low ambient pressures, both the pressure effect and pressure impact are at a lower level. Protection design against pressure rise becomes more complicated at a higher ambient pressure.

3.2.2. Evaluation of Gas Amount

Thermodynamic parameters of the chamber interior, pressure and temperature, are potentially influenced by test conditions and heat dissipation conditions. The variations of pressure and temperature could be disparate without given situations. Instead, the gas amount versus time, calculated by the ideal gas law based on the pressure and temperature in the chamber as expressed in Equation (1), represents the venting behaviors.

$$n={\displaystyle \frac{P_{\mathrm{a}}{\mathrm{V}}_0}{\mathrm{R}{T}_{\mathrm{a}}}}$$

(1)

n is the total gas amount inside the chamber, in mol; P_a denotes the average of ambient pressure, in kPa; V₀ is the effective volume of the chamber, 230 L; R is the universal gas constant, 8.314 J/mol/K; T_a denotes the average of ambient temperature, in K.

The venting characteristic parameters are derived from the gas amount curve in Equation (1). Nevertheless, according to the pressure curve inside the chamber, the pressure peak means a non-equilibrium state. Near this point, the ideal gas law may not be applicable; hence, an evaluation method was proposed to calibrate the real-time gas amount, as shown in Figure 6. Before the venting event, the total gas amount was approximately a constant due to the gas-tight chamber, and the ideal gas law was satisfied. For each venting event, it could be further divided into two stages, namely the non-equilibrium stage and the quasi-equilibrium stage. Take the second venting process induced by the TR, for instance, in the non-equilibrium stage; the uneven temperature field and dynamic pressure led to an inaccuracy of the gas amount. At the point (t₂, n₂), the first venting process adequately weakened, and the gas amount by the ideal gas law was reliable. The situation was also applicable after t = t₃. Based on the gas amount calculated directly by Equation (1), valuable datasets were available, namely (t₂, n₂), (t₃, n₃) and (t₃, n₃^′), where n₃^′ is the slope at t = t₃. A unique formula of quadratic polynomial was determined as Equations (2)–(4). In Figure 6, the black solid line and blue dotted line describe the whole real-time gas amount together, and corresponding venting characteristic parameters are acquired, such as the gas amount of each venting, the maximum gas release rate, the total gas amount, and the venting duration. Compared to the linear fitting method, the quadratic fitting method can satisfy the decaying tendency of each venting event in practice.

$$n(t)=a_i{(t-t_{2i-2})}^2+b_i(t-t_{2i-2})+n_{2i-2},t_{2i-2}<t<t_{2i-1},i=1,2$$

(2)

$$a_i={\displaystyle \frac{n_{2i-2}-n_{2i-1}+n_{2i-1}^{’}(t_{2i-1}-t_{2i-2})}{{(t_{2i-1}-t_{2i-2})}^2}},i=1,2$$

(3)

$$b_i={\displaystyle \frac{2(n_{2i-1}-n_{2i-2})}{(t_{2i-1}-t_{2i-2})}}-n_{2i-1}^{’},i=1,2$$

(4)

n is the total gas amount inside the chamber, in mol; t denotes time, s; a_i and b_i are undetermined coefficients, in mol/s² and mol/s, respectively, and i denotes the first and second venting processes. In most cases, b₂ denotes the maximum gas release rate during TR.

3.2.3. Venting Characteristic Parameters

The primary venting characteristic parameters employed in this study encompass the maximum gas release rate, the total gas amount, and the venting duration.

Table 3. Critical points under various ambient pressures.

Parameters	P0 = 20 kPa	P0 = 100 kPa	P0 = 260 kPa
t0 (s)	0	0	0
n0 (mol)	2.28	9.75	24.54
t1 (s)	66	57	51
n1 (mol)	2.42	9.89	24.63
n1′ (mol/s)	0.00018	0.0011	0.0024
t2 (s)	250	219	94
n2 (mol)	2.48	9.99	24.69
t3 (s)	289	269	148
n3 (mol)	4.09	11.58	26.42
n3′ (mol/s)	0.0075	0.015	0.016
n4 (mol)	4.43	12.17	27.18

lists the critical points employed in Section 3.2.2 under various ambient pressures, defined as Figure 6. The safety valve of LIBs opened at t = t₀, and the initial gas amount was mainly decided by the initial pressure inside the chamber. Therefore, the real-time gas amount could be calibrated to describe and predict venting behaviors.

Figure 7 displays the comparison of the gas amount during each venting process and the predicted maximum gas release rate under different ambient pressures. For the first venting event, the total gas amount was around 0.20 ± 0.05 mol regardless of the dramatically changed ambient pressures. The amount of vaporized electrolyte and other produced gases was small, and the gases released in the first venting event over several minutes were limited. In comparison, the total gas amount during the second venting process was relatively large and presented a positive correlation with ambient pressure. Therefore, the second venting event is the dominant gas release process. The maximum gas release rate depends on the fitting method to a degree, and the predicted maximum gas releasing rate is 0.075 mol/s for the case of 20 kPa in this study, obviously larger than that of other two cases. The maximum gas release rate should appear at t = t₂ for most cases. Although the gas amount increases with ambient pressure, the maximum gas release rate exhibits an inverse trend.

Another important venting characteristic parameter is the venting duration, as shown in

Table 4. Venting duration under different ambient pressures.

Parameters	P0 = 20 kPa	P0 = 100 kPa	P0 = 260 kPa
t50 (s)	264	243	124
t50,1 (s)	32	24	45
t50,2 (s)	16	28	33

. Free cooling of the examined battery remains with the elevated temperature requires a long period, and gas-producing reactions continuously proceed; therefore, it is challenging to precisely define the venting duration for the whole venting process. In this research, t₅₀, t_50,1, and t_50,2, which are defined as the time intervals to release 50% of the total gas amount for the whole venting process, first venting event, and second venting event, respectively, are utilized to describe the venting duration and represent the average venting rate. t_50,2 is counted from t = t₂. Once the safety valve opens, all vented gases release from the examined battery more violently against ambient pressure in terms of t₅₀. The first venting event contributes slightly to the total gas amount, and t_50,1 is hardly influenced by TR reactions; hence, it is not discussed in detail in this research. As for the second venting event, the average venting rate on the basis of t_50,2 is much faster at a low ambient pressure. Although the duration of the entire failure process is longer at low pressure, the venting intensity induced by TR is more violent. Different venting modes should be applied under various ambient pressures.

3.3. Gas Analysis

3.3.1. Compositions of Vented Gases

The gas generation inside the battery due to the evolution of TR is one of the most essential processes to determine the venting behaviors. The identification of vented gases is of great importance to reveal the intrinsic mechanism of TR reactions and provide reference for hazard prevention. Both the proportion and content of the vented gases are crucial parameters to describe the venting behaviors.

Figure 8 gives the compositions and molar amount of the vented gases under different ambient pressures. In total, the six most common gas species, accounting for more than 1% by volume, were detected in all cases, including CO, CO₂, H₂, C₂H₄, CH₄, and C₃H₆, according to the order of proportion. As for the proportion by volume of the main vented gases, CO and C₃H₆ are negatively correlated with ambient pressure, while CO₂, H₂, C₂H₄, and CH₄ present a positive correlation with ambient pressure. There are significant changes in the gas proportion of CO and H₂. The molar amount which can be expressed by the product of the gas proportion and the total gas amount reflects the absolute content of each gas species, showing a consistent trend with the gas proportion. The content of CO₂ presents a strongly positive correlation with ambient pressure as the total gas amount increases. Meanwhile, there is a drop in the ratio of CO to CO₂, implying that the redox reactions among battery materials during TR are more complete under a higher ambient pressure. This can be attributed to the increased ejections of reactants under a lower pressure. On the whole, by adjusting ambient pressure, the main components of the vented gases can hardly change, while the relative proportion and absolute content of these gases significantly dissimilate. For instance, the concentration of CO drops by 12.6% (0.15 mol), while that of H₂ increases by 5.6% (0.24 mol).

The vented emissions are mainly composed of gases, particles, and liquid electrolyte or its reaction products (if any), which contributes to the mass loss together during the process of TR. In this study, gaseous emissions were mainly focused on. The total gas amount, the total mass loss and the total elemental contents of C, H and O are the overall parameters to characterize the gas venting event. Maximum battery temperature as a variable was recalculated by the average of the battery temperature, and 747.4 °C, 774.7 °C, and 808.1 °C were obtained for 20 kPa, 100 kPa, and 260 kPa, respectively, showing a strongly positive relation with ambient pressure. Figure 9 illustrates the influence of maximum battery temperature on the above parameters. The horizontal axis represents the average of the maximum temperatures for both sides of the tested battery. The mass loss varies inversely with the maximum battery temperature, while the total gas amount increases obviously with the maximum battery temperature from 2.15, 2.43 to 2.64 mol. Figure 3 points out that there are two ejections during the process of TR for each tested battery. Each ejection is made up of a fast ejection and a subsequently continuous stream. Under a lower ambient pressure, the earlier opening of the safety valve allows more liquid electrolyte to be ejected out of the tested battery during the first ejection before the TR. Under 260 kPa, the duration of the first ejection was relatively insufficient, which could explain the decrease in the mass loss ratio. Meanwhile, the total gas amount can describe the intensity of TR to some extent, satisfying the positive relation. All the most common gas species are composed of C, H and O, and the molar amount of these elements shown in Figure 9 was calculated by the detected compositions in Figure 8. Element H significantly increases with ambient pressure, potentially implying that more decomposition reactions occur under an elevated ambient pressure because of a relatively long period prior to the onset of the safety valve. Furthermore, C and O show a weakly positive trend with the maximum battery temperature as a consequence of the increasing total gas amount. The average molar weight of the vented gases, defined as the proportion of the gas mass to the gas molar amount, also represents the degree of decomposition during the whole TR process, 26.30, 26.18, and 24.64 g/mol, respectively, with the increase in the maximum battery temperature, signifying more thorough reactions during the TR process. In other words, under a lower ambient pressure or for a lower burst pressure of the safety valve, some reactants are released from the tested battery at the initial stage prior to the TR; hence, less energy is released in a series of exothermic reactions, mitigating the intensity of the TR. However, more attention should be paid to the remaining battery materials and the vented emissions in order to prevent further energy release.

3.3.2. Flammability Characteristics of Vented Gases

Flammability characteristics, including the upper flammability limit (UFL), lower flammability limit (LFL), flammability range (R_F), and flammability hazard index (H_F), can effectively represent the flammability risk of vented gases to a great extent [25]. Equation (5) provides the calculation method for flammability limits of mixed combustible gases. When an inert gas, such as CO₂, is present in the mixture, CO and CO₂ are treated as a new synthetic gas in this study. This synthetic gas is considered as a new component and substituted into Equation (5). R_F and H_F are expressed by Equations (6) and (7). The larger the R_F and H_F are, the greater the probability and hazards of combustion are. As mentioned in Section 3.3.1, CO, CO₂, H₂, C₂H₄, CH₄, and C₃H₆ are the most common species and normalized to calculate flammability characteristic parameters.

Table 5. Flammability characteristics of the vented gases.

Parameters	P0 = 20 kPa	P0 = 100 kPa	P0 = 260 kPa
UFL (%)	53.41	52.37	50.81
LFL (%)	8.53	8.20	7.49
RF (%)	44.88	44.17	43.33
HF	5.26	5.39	5.79

shows the results of flammability characteristics. Both UFL and LFL decrease slightly with ambient pressure. Based on R_F and H_F, although the probability of combustion reduces, the potential hazards increase significantly once a fire or explosion occurs.

$${\mathrm{XFL}}_{\mathrm{mix}}={\displaystyle \frac{1}{{\displaystyle \sum _i\frac{{\phi }_i}{{\mathrm{XFL}}_i}}}}\times \mathrm{100\%}$$

(5)

$$R_{\mathrm{F}}=\mathrm{UFL}-\mathrm{LFL}$$

(6)

$$H_{\mathrm{F}}={\displaystyle \frac{\mathrm{UFL}-\mathrm{LFL}}{\mathrm{LFL}}}$$

(7)

XFL denotes UFL or LFL and ${\phi }_i$ is the volume fraction of species i.

4. Discussion

Ambient pressure and burst pressure of the safety valve influence gas venting behaviors by altering the occurrence of venting events. Gas venting behaviors could be characterized by numerous parameters. In this study, the main characteristic parameters include maximum battery temperature, jet temperature, pressure shock, total gas amount, gas release rate, venting duration, early warning, and flammability risk, as summarized in

Table 6. Summary of main gas venting behaviors.

Parameters	P0 = 20 kPa	P0 = 100 kPa	P0 = 260 kPa
Maximum battery temperature	+	o	−
Jet temperature	−	o	+
Pressure shock	+	o	−
Total gas amount	+	o	−
Gas releasing rate	−	o	+
Venting duration	+	o	−
Early warning	+	o	−
Flammability risk	+	o	−

. The ‘+’ symbol indicates beneficial effects for safety design of battery systems, while the ‘−’ symbol denotes poor performance. Additionally, the ‘o’ symbol represents reference values under normal pressure conditions. Overall, lowering ambient pressure or burst pressure of the safety valve leads to a relatively safe condition.

5. Conclusions

In this study, TR tests of 50 Ah prismatic NCM523 batteries with a 75% SOC were carried out using a 400 W heating plate in a sealed chamber filled with nitrogen under different ambient pressures. Thermal parameters, venting characteristics, gas compositions, and flammability characteristics were investigated to elucidate the gas venting behaviors. The vented gases were analyzed by GC and compared in terms of the composition and molar number of main elements. By comparing the characteristics of venting behaviors under different ambient pressures, the mechanism of external pressure on the process of TR was explored. The research can assist in optimizing the burst pressure of safety valves for prismatic batteries and provide guidance for battery safety evaluation under various operating pressures.

The primary conclusions are as follows:

(1)

Different ambient pressures significantly affect the TR process and gas venting behaviors.

(2)

In this study, lower ambient pressure leads to an earlier venting event, mitigates the intensity of TR in terms of maximum battery temperature, and reserves more response time for hazard prevention. Simultaneously, lower ambient pressure reduces the total gas amount and increases the mass loss ratio.

(3)

Pressure effects and shocks are remarkable at elevated ambient pressures, particularly during the second venting event. The first venting event contributes minimally to the overall venting process, except for providing early warnings for TR.

(4)

The six most common gas species above 1% by volume were detected under different pressures, including CO, CO₂, H₂, C₂H₄, CH₄, and C₃H₆, accounting for over 97% of all vented gases by volume. Lower ambient pressure implies more incomplete reactions during the process of TR and releases less CO₂.

(5)

Among the main vented gases, element H increases significantly with ambient pressure, indicating that electrolytes and separators contribute more to gas generation during TR under higher pressures.

(6)

Reducing ambient pressure or burst pressure of the safety valve enhances the safety design of battery systems.