In Situ Gas Analysis and Fire Characterization of Lithium-Ion Cells During Thermal Runaway Using an Environmental Chamber

Summary

Here, we describe a test procedure developed to characterize thermal runaway and fires in lithium-ion cells through in situ measurements of various parameters in an environmental chamber.

Abstract

An experimental apparatus and a standard operating procedure (SOP) are developed to collect time-resolved data on the gas compositions and fire characteristics during and post-thermal runaway of lithium-ion battery (LIB) cells. A 18650 cylindrical cell is conditioned to a desired state-of-charge (SOC; 30%, 50%, 75%, and 100%) before each experiment. The conditioned cell is forced into a thermal runaway by an electrical heating tape at a constant heating rate (10 °C/min) in an environmental chamber (volume: ~600 L). The chamber is connected to a Fourier transform infrared (FTIR) gas analyzer for real-time concentration measurements. Two camcorders are used to record major events, such as cell venting, thermal runaway, and the subsequent burning process. The conditions of the cell, such as surface temperature, mass loss, and voltage, are also recorded. With the data obtained, cell pseudo-properties, venting gas compositions, and venting mass rate can be deduced as functions of cell temperature and cell SOC. While the test procedure is developed for a single cylindrical cell, it can be readily extended to test different cell formats and study fire propagation between multiple cells. The collected experimental data can also be used for the development of numerical models for LIB fires.

Introduction

In the last few decades, lithium-ion batteries (LIBs) have gained popularity and benefited from tremendous technological advancements. Owing to various advantages (e.g., high energy density, low maintenance, low self-discharge and charge times, and long lifespan), the LIB has been considered a promising energy storage technology and extensively used in various applications, such as large energy storage systems (ESSs), electric vehicles (EVs), and portable electronic devices. While the global demand for LIB cells is expected to double from 725 GWh in 2020 to 1,500 GWh in 2030¹, there has been a substantial increase in fires and explosions related to LIBs in recent years². These accidents demonstrate the high risks associated with LIBs, raising concerns regarding their large-scale utilization. To mitigate these concerns, it is crucial to gain a thorough understanding of the process of LIB thermal runaway leading to fires.

Previous accidents have revealed that LIB cells fail when the cell electrochemistry is disrupted by overheating in abnormal operating circumstances (such as external short circuit, rapid discharge, overcharging, and physical damage) or due to manufacturing defects and poor design^2,3,4. These events lead to the decomposition of the solid-electrolyte interface (SEI), stimulating highly exothermic chemical reactions between electrode materials and electrolytes. When the heat produced in these reactions exceeds that being dissipated, it results in rapid self-heating of the cells, also known as thermal runaway. Internal temperature and pressure can continue rising until built-up pressure causes the battery to rupture and release flammable, toxic gases at high speed. In a multi-cell battery configuration, a thermal runaway in a single cell, if not controlled, can lead to thermal runaway propagation to other cells and incidents of fire and explosion at catastrophic levels, especially in enclosed spaces with limited ventilation. This poses significant threats to human safety and structures.

In the past few decades, a number of studies have been carried out to investigate the thermal runaway reactions of LIBs leading to the combustion of organic electrolytes inside the battery and the release of flammable gases under different heating conditions^{2,5,6,7,8,9,10,11,12}. For example, Jhu et al.¹⁰ demonstrated the hazardous nature of charged cylindrical LIBs compared to uncharged ones using an adiabatic calorimeter. Many other studies focused on the thermal runaway behavior of LIBs at different state-of-charges (SOCs). For example, Joshi et al.¹³ investigated the thermal runaway of various types of commercial LIBs (cylindrical and pouch) at different SOCs. It was noticed that cells at higher SOCs had a higher chance of undergoing thermal runaway compared to those at lower SOCs. In addition, the minimum SOC for a thermal runaway to occur varied with the cell formats and chemistries. Roth et al.¹¹ tested cylindrical LIBs in an accelerating rate calorimeter (ARC) and observed that, as the SOC increased, the onset temperature of thermal runaway decreased and the acceleration rate increased. Golubkov et al.¹² developed a custom-designed test stand and showed that the maximum surface temperature of cylindrical LIBs could be as high as 850 °C. Ribière et al.¹⁴ used a fire propagation apparatus to investigate the fire-induced hazards of pouch LIBs and noticed that the heat release rate (HRR) and toxic gas production varied significantly with the cell SOC. Chen et al.¹⁵ studied the fire behaviors of two different 18650 LIBs (LiCoO₂ and LiFePO₄) at different SOCs, using a custom-made in situ calorimeter. HRR, mass loss, and maximum surface temperature were found to increase with SOC. It was also demonstrated that the risk of explosion was higher for a fully charged lithium cobalt oxide (LiCoO₂) cathode 18650 cell compared to a lithium iron phosphate (LiFePO₂) cathode 18650 cell. Fu et al.¹⁶ and Quang et al.¹⁷ conducted fire experiments on LIBs (at 0%-100% SOCs) using a cone calorimeter. It was observed that LIBs at a higher SOC resulted in higher fire hazards due to shorter lengths of time to ignition and explosion, higher HRR, higher surface temperature, and higher CO and CO₂ emissions.

To summarize, previous studies using different calorimeters^18,19 (ARC, adiabatic calorimetry, C80 calorimetry, and modified bomb calorimetry) have provided abundant data on the electrochemical and thermal processes associated with LIB thermal runaway and fires (e.g., HRR, compositions of the vented gases) and their dependencies on the SOC, battery chemistry, and incident heat flux^2,3,7,20. However, most of these methods were designed originally for conventional solid combustibles (e.g., cellulose samples, plastic) and provide limited information when applied to LIB fires. While some previous tests measured the HRR and the total energy generated from chemical reactions, the kinetics aspects of post-thermal runaway fires were not fully addressed.

The severity of hazards during thermal runaway is mainly dependent on the nature and composition of the gases released^2,5. Therefore, it is important to characterize the released gases, the venting rate, and their dependence on the SOC. Some previous studies measured the vent gas compositions of LIB thermal runaway in an inert environment (e.g., in nitrogen or argon)^12,21,22; the fire component during the thermal runaway was excluded. In addition, these measurements were mostly performed post-experiments (instead of in situ). Evolutions of vent gas composition during and post-thermal runaway, especially those involving fires and toxic gases, remained under-explored.

It is known that thermal runaway disrupts the electrochemistry of the battery and impacts the cell voltage and temperature. A comprehensive test to characterize the thermal runaway process of the LIB should, therefore, provide simultaneous measurement of the temperature, mass, voltage, and vented gases (rate and composition). This has not been achieved in a single setup in the previous studies. In this study, a new apparatus and test protocol are developed to collect time-resolved data on the cell information, gas compositions, and fire characteristics during and post-thermal runaway of LIB cells²³. The test apparatus is shown in Figure 1A. A large (~600 L) environmental chamber is used to confine the thermal runaway event. The chamber is equipped with a pressure relief valve (with a set gauge pressure at 0.5 psig) to prevent pressure rise in the chamber. A Fourier transform infrared (FTIR) gas analyzer is connected to the chamber for in situ gas sampling throughout the test. It detects 21 gas species (H₂O, CO₂, CO, NO, NO₂, N₂O, SO₂, HCl, HCN, HBr, HF, NH₃, C₂H₄, C₂H₆, C₃H₈, C₆H₁₄, CH₄, HCHO, C₆H₆O, C₃H₄O, and COF₂). The FTIR sampling rate is 0.25 Hz. In addition, a standalone hydrogen sensor is installed inside the chamber near the FTIR sampling port to record the H₂ concentration. Two pumps (a 1.3 cfm chemical-resistant diaphragm pump and a 0.5 hp vacuum pump) are installed in the chamber exhaust line. After each experiment, a chamber clean-up procedure is followed to filter and pump the chamber gas directly to the building exhaust line.

In each experiment, the cell is set up inside the chamber in a sample holder (Figure 1B). Thermal runaway is triggered by a proportional-integral-derivative (PID)-controlled electric heating tape at a constant heating rate of 10 °C/min. Cell surface temperatures are recorded by thermocouples in three different locations along the length of the cell. The mass loss of the cell is measured by a mass balance. The chamber pressure is monitored by a pressure transducer. The cell voltage and the power input (voltage and current) to the heating tape are also recorded. All sensor readings (thermocouples, mass loss, cell voltage, heating tape current, and voltage) are collected by a custom data acquisition program at a rate of 2 Hz. Lastly, two camcorders (1920 pixel x 1080 pixel resolution) are used to record the entire process of the experiments from two different angles.

The objective of developing this new test method is twofold: 1) to characterize the smoke and fire behaviors associated with LIB thermal runaway and 2) to provide time-resolved experimental data that enables the development of high-validity numerical models for battery fires. The long-term goal is to advance the understanding of how thermal runaway propagates between cells in a battery pack and how a battery fire scales up when going from single cells to multi-cell batteries. Ultimately, this will help improve guidelines and protocols for storing and transporting LIBs safely.

Protocol

1. Startup of the FTIR gas analyzer

NOTE: The procedures can be different for different brands and models of the FTIR gas analyzer. The following procedure is for the specific gas analyzer used in this work.

1. Install a new filter or a clean filter (i.e., one that has been cleaned in an ultrasonic bath) in the filter/valve unit (see Figure 1 and Figure 2).
2. Open the valve of the nitrogen cylinder connected to the gas analyzer (see Figure 2). Adjust the nitrogen flow rate to 150-250 cc/min.
   NOTE: This is to prepare for the N₂ purging during the pre/post-test cleaning of the gas analyzer.
3. Follow the FTIR startup procedure described in the manufacturer manual, "FTIR and PAS Pro for the FTT Smoke Density Chamber Standard Operating Procedure"²⁴, Version 3.1.
   ​NOTE: While the FTIR is running, the gas line between the FTIR and the chamber (see Figure 2) is maintained at 180 °C to prevent gas condensation. Be careful not to touch the heated line and the filter/valve unit.

2. Cell preparation

1. Record the date, time, SOC, test participants, test number, cell manufacturer, cell format, and cell model number on an experiment log sheet.
2. Measure and record the initial voltage and mass of the cell (with a precision of 0.01 g) on the experiment log sheet.
3. Attach heating tape (1 in x 2 in, 20 W/in²) to the center of the cell and take a picture of the cell with the heating tape. Ensure that the heating tape wires point toward the negative side of the cell (see Figure 3).
4. Attach three thermocouples (K-type with a probe diameter of 0.02 in, length of 12 in) to the cell surface using high-temperature resistant tape, one near the positive terminal, one in the middle, and one at the bottom near the negative terminal of the cell, all located 5 mm away from the edge of the heating tape (see Figure 3A). Use the thermocouple near the positive terminal to control the heating rate through the PID. After installing the thermocouples, take a picture of the cell with a ruler to confirm the distance from the heating tape.
5. Spot weld nickel tabs (0.1 mm in thickness, 5 mm in width, and 100 mm in length) to the positive and negative terminals of the cell for the cell voltage measurement. Ensure that the nickel tabs are oriented in different directions to prevent them from touching each other, resulting in an external short circuit (Figure 3B).
6. Load the cell on the cell holder, as shown in Figure 3C.
7. Confirm that all wires of the voltage measurement and the thermocouples are routed toward the negative terminal of the cell to avoid the vent ports on the positive terminal of the cell.

3. Test chamber setup

1. Turn on the light emitting diode (LED) light in the chamber.
2. Place the cell and the cell holder on the mass balance in the chamber (see Figure 4). Connect the thermocouple connectors, heating tape, and nickel tabs to chamber-feedthrough plugs and wires.
3. Turn on the mass balance. Tare the balance.
4. Turn on the power supply for the hydrogen sensor.
5. Turn on the PID controller for the heating tape. Set up the heating profile (temperature: 200 °C; ramp time: 17 min). Connect the cables for the PID controller, data acquisition, and the mass balance to a laptop and start the data acquisition program on the laptop.
6. Make sure all the sensor readings shown in the data acquisition program are reasonable: cell voltage close to the value measured in step 2.2, voltage and current input to the heating tape close to zero (since the power is not on yet), thermocouple readings close to room temperature (~25 °C), chamber pressure ~1 atm, and mass reading ~0 g. After checking the measurements, turn off the data acquisition program.
7. Adjust the front- and side-view camcorder settings: manual white balance (calibrated initially using a white paper), manual focus (fixed on the cell surface near the positive terminal), auto exposure, auto IRIS, and auto shutter speed. Make sure that the camcorder battery is full.
8. Position the front-view camcorder on a tripod outside of the chamber (see Figure 4). Start recording on the side-view camcorder and place it inside a protection box in the chamber. Check the side-view camcorder angle and view. Lock the protection box.
9. Double-check if there are any hazardous or unnecessary items inside the chamber and if any steps listed above have been skipped.
10. Close the chamber and ensure all the screws on the cover plates are tightly fastened (e.g., using an impact wrench).
11. Use the vacuum or diaphragm pump to conduct a leak check. Double-check that all the valves, cover plates, and observation windows are securely fastened.
    NOTE: If the pressure decreases slowly or does not drop, there are leakages somewhere.
12. Change the FTIR intake from ambient air to the chamber.
13. Connect the FTIR return line to the chamber (see Figure 2).

4. Thermal runaway and fire experiment

1. Set the PID controller to ramp-soak mode.
2. Turn off the light in the room and the LED light in the chamber.
3. Start the front-view camcorder recording. Use the camera to record the actions in steps 4.4 and 4.5 for time synchronization of all the collected data (sensor data, FTIR readings, and videos) after the experiments.
4. Start the data recording in the data acquisition program on the laptop.
5. Start the PID ramp-soak mode at 10 s on the data acquisition program timer. Turn on the chamber LED light. Start the FTIR recording.
6. Position the front-view camcorder on the tripod and continue recording the experiment.
7. Move to a different room and continue monitoring the data acquisition panel on the laptop through a remote-controlled desktop program. Note that this step is taken for extra precaution and is not required. Since the experiments are fully confined in the environmental chamber, the risk to the surrounding personnel is minimal.
8. If present in the same room as the chamber, wear appropriate personal protective equipment (PPE) during the entire test period (e.g., gloves, P100 respirator, safety goggles, and fire-resistant lab coat).

5. Termination of the experiment

1. When thermal runaway occurs (i.e., thermocouple readings show sudden spikes) or after the PID controller has maintained the cell temperature at 200 °C for 60 min (whichever occurs first), turn off the power to the heating tape and set the PID controller to standby mode.
2. Wait for all the thermocouple readings to drop to room temperature (<50 °C). Note that the cooling process for a single cell can take about 30 min.
3. Stop the data acquisition program on the laptop, FTIR measurement, and video recording.

6. Turning off the FTIR gas analyzer

1. Follow the FTIR turn-off procedure documented in the manufacturer manual, "FTIR and PAS Pro for the FTT Smoke Density Chamber Standard Operating Procedure", Version 3.1.
2. Purge the FTIR gas analyzer with nitrogen to clean the tube and analyzer for ~15 min. Ensure that the flow rate of N₂ to the FTIR gas analyzer is 150-250 cc/min.
3. While purging the gas analyzer, transfer the FTIR results to a USB memory stick.
4. After purging, turn off the gas analyzer.
5. Wear appropriate PPE, including a pair of heat-insulating gloves, and remove the filter in the heated filter/valve unit. Be extremely careful, as the filter/valve unit can be very hot.
6. Clean the removed filter with an ultrasonic bath of a cleaning solution.

7. Chamber clean-up and data collection

1. Before the chamber-cleanup vacuuming procedure, check if the FTIR sampling (intake) line (which is connected to the chamber) is closed or open to the ambient air. For the gas analyzer model presented in this study, select Ambient Air on the PAS Pro software or shut down the FTIR entirely. Failure to do this causes damage to the FTIR.
2. Make sure a carbon filter is installed between the chemical-resistant diaphragm pump (Pump 1 in Figure 2) and the chamber. Mark the number of uses on the filter and replace it with a new one every ~10-15 tests.
3. Open Valve 1 to prepare for partially vacuuming the chamber using the chemical-resistant diaphragm pump.
4. Run the diaphragm pump until the chamber pressure drops to P₁ = 9.7 psia (i.e., -5 gauge pressure).
5. Turn off the diaphragm pump and close Valve 1.
6. Open Valve 3 (see Figure 4) to fill the chamber with ambient air.
7. Close Valve 3 when the chamber pressure recovers to the ambient pressure, P_∞.
8. Repeat the partial vacuuming procedure (steps 7.3-7.7) five times. Through this, the exhaust gas percentage in the chamber should drop to (P₁/P_∞)⁵ = 12.5%.
9. Open Valve 2 to prepare for fully vacuuming the chamber using the vacuum pump (Pump 2 in Figure 2).
10. Run the vacuum pump until the chamber pressure drops to P₂ = 4.7 psia (or -10 psia gauge pressure).
11. Turn off the pump and close Valve 2.
12. Open Valve 3 to fill the chamber with ambient air until the chamber pressure recovers to the ambient pressure, P_∞.
13. Repeat the complete vacuuming procedure (steps 7.9-7.12) two times.
    NOTE: After the partial and complete vacuuming procedures, the exhaust gas percentage in the chamber should be lower than 1.3%.
14. Open the chamber and retrieve the camcorder and the cell.
15. Turn off the mass balance.
16. Use a wet paper towel to clean the interior of the chamber (e.g., remove all debris and wipe the inner walls of the chamber).
17. Take photos before, during, and after taking the cell off the cell holder.
18. Weigh the cell and record the post-test mass of the cell.
19. Retrieve all the recorded data (thermocouple readings, cell voltage, heating tape voltage, current, chamber pressure, and cell mass measurement) from the laptop and the video recordings from the two camcorders.
20. Combine the collected videos using a video editing software. Record the onset time of the major events, such as cell venting, thermal runaway, and fire. Save the combined video in a desired format (e.g., mp4 or avi).
21. Post-process the collected data and generate plots to visualize the time evolution of all measurements.

Representative Results

Videos representing typical thermal runaway processes with and without fires are included in Supplementary File 1 and Supplementary File 2, respectively. Key events are depicted in Figure 5. As the cell temperature is raised (to ~110-130 °C), the cell starts swelling, indicating the buildup of the internal pressure (caused by the vaporization of electrolytes and the thermal expansion of gases inside the cell²). This is followed by the opening of the venting port and the release of the vent gas (Figures 5A and 5B, respectively). The gradual venting process continues for a few minutes. After that, the cell starts to vent profusely (Figure 5C), and a thermal runaway occurs (Figure 5D). These happen regardless of the SOC. At higher SOCs (e.g., 75% and 100%), sparks (Figure 5D), fire (Figure 5E), and cell content expulsion (see post-test pictures in Figure 5F,G) are also observed during and post-thermal runaway. At lower SOCs (e.g., 30% and 50%), the eruption of electrolytes with heavy smoke is observed without sparks or fires. Note that, depending on the phenomena of interest, the camera/camcorder settings and the background LED light needs to be carefully chosen. In Figure 5A, the camcorder is focused on the venting port, and the bright white background light is chosen to capture the electrolyte boiling phenomenon at the onset of the venting process. If the interest is in the gaseous fire, auto-adjusted camcorder settings, a dimmer green LED light, and a dark background are recommended.

Representative measurements are plotted in Figure 6, with key events marked by vertical dash lines. These plots are for a test where fire occurs (at 75% SOC, shown in Supplementary File 1). Figure 6A shows that the cell temperature is higher at the mid-location than at the top (near the positive terminal) and bottom (near the negative terminal) locations. The reading of the top-location thermocouple (which is used for the PID control) confirms that the cell heating rate is at the intended value (i.e., ~10 °C/min or 0.167 °C/s). Note that the temperature readings show a momentary dip at the onset of cell venting (Event 3). This is due to the sudden heat loss due to the release of gases through the vent.When thermal runaway occurs, the cell temperature shows a sudden spike. After the thermal runaway, especially for cases where fire and cell content expulsions occur, the thermocouples may detach from the cell surface and hence read the gas temperatures instead of the battery surface temperatures. Special caution needs to be taken when interpreting the data. Further, special attention should be paid in confirming that the thermocouples do not get detached during the test.

Also, the cell voltage drops to zero (Event 2) before thermal runaway occurs (minutes before the cell starts to vent in the representative case shown in Figure 6A). It is known that the decomposition of the solid-electrolyte interphase (SEI) layer starts at ~80-120 °C and the separator starts melting at 135-166 °C². The breakdown of these components leads to an internal short circuit (ISC) between the two electrodes, accompanied by electrolyte decomposition, and then eventually, the thermal runaway of a LIB cell. The cell voltage drop is the first signal of the LIB failure event. Depending on the chemistry, format, and design of the cell, each failure event (e.g., voltage drop, venting, thermal runaway) may occur at different times and at different cell temperatures.

The mass loss rate can be deduced from the mass loss data obtained in the test procedure. The mass loss (shown in Figure 6B) indicates two distinct gas release periods, one during cell venting and the other during thermal runaway. The mass loss during the venting period is similar (~3-4 g) at all considered SOCs, while the mass loss at thermal runaway increases with the SOC. Also, the mass loss at thermal runaway accounts not only for the vented gas, but also for the ejected cell content and components that burn away.

The concentrations of major hydrocarbon and toxic gas species are shown in Figure 6C-E. Different compositions are observed during the venting period and thermal runaway. As the vent gas disperses across the chamber after fire extinction, the concentration of each species converges to a stable value.

The recorded current (I) and voltage (V) supplied to the heating tape (shown in Figure 7A) can be used to calculate the power input to the cell. Cumulated energy input and heating power are calculated as follows:

(1)

(2)

In the representative test, the cumulative energy curve (E in Eq. 1; solid black line in Figure 7B) can be fit by the second-order polynomial regression line (solid blue line in Figure 7B). Using this regression line, the power input (dE/dt in Eq. 2) to the cell is found to increase linearly with time (blue dash line in Figure 7B).

Figure 1: Experimental apparatus and schematics. (A) The experimental apparatus for LIB thermal runaway experiments. (B) The schematics of the setup inside the chamber. Please click here to view a larger version of this figure.

Figure 2: Schematic of the flow system for the apparatus. Please click here to view a larger version of this figure.

Figure 3: The preparation of a 18650 cell. (A) Step 2.4. (B) Step 2.5. (C) Step 2.6. Please click here to view a larger version of this figure.

Figure 4: The installation of the LIB cell inside the chamber with data acquisition. (A) Step 3.2. (B) Step 3.5. (C-E) Step 3.8. Please click here to view a larger version of this figure.

Figure 5: Key events during a typical thermal runaway process. (A) Opening of the venting port and boiling of the electrolyte. (B) Gradual release of the vent gas. (C) Intense release of the vent gas before thermal runaway. (D) Onset of the thermal runaway. (E) Fire. (F-G) Ejected cell contents observed during the post-test inspection. Please click here to view a larger version of this figure.

Figure 6: Representative data obtained for an 18650 cylindrical cell at 75% SOC. (A) Cell temperature. (B) Mass loss. (C-E) Concentrations of major hydrocarbon and toxic gas species. Please click here to view a larger version of this figure.

Figure 7: Representative data for the heating tape power input. (A) Voltage and current supplied to the heating tape. (B) Calculated energy and power supplied to the heating tape. Please click here to view a larger version of this figure.

Supplementary File 1: Video of the thermal runaway process of the 18650 cell at 75% SOC. Please click here to download this File.

Supplementary File 2: Video of the thermal runaway process of the 18650 cell at 50% SOC. Please click here to download this File.

Discussion

The most critical steps in the protocol are those concerning the toxic gases released in the LIB thermal runaway. The leak test in step 3.11 needs to be carefully performed to ensure that the toxic gases are confined in the chamber during the experiments. The chamber gas clean-up procedures (steps 7.1-7.14) must also be properly done to mitigate the hazard from the toxic gases. Toxic gases may constitute only a small fraction of the vent gas during LIB thermal runaway. However, even very low concentrations of some toxic gases pose a great threat to human health. Occupational 8 hour exposure limits of acrolein and formaldehyde imposed by the Occupational Safety and Health Administration (OSHA) are 0.1 and 0.75 ppm, respectively, which are significantly lower than the measured values using the 600 L chamber (see Figure 6E). This emphasizes the importance of having a sealed chamber and wearing a suitable mask during the entire test. This also further highlights the need to have a test method, such as the one presented here, for characterizing toxic gas release for LIBs.

Other critical steps regard the time synchronization between sensor measurements, FTIR readings, and camcorder videos. In protocol steps 4.3-4.5, video recording and the onset of the LED light provide a mean to synchronize all data. Unless alternative synchronization methods are used, these steps need to be carefully followed. Only with synchronized data can the vent gas species and fire characteristics be correlated to the cell conditions (e.g., temperature, mass loss, voltage) and to different events of the thermal runaway.

Limitations exist for the presented test method. First, it is limited to thermal runaway caused by external thermal abuse. The results may not represent the thermal runaway process caused by other battery failure modes (e.g., mechanical abuse, internal short circuit). Second, the vent gas mass release rate is not directly measured. Instead, it is deduced from the recorded mass loss of the cell. During the venting process prior to thermal runaway, the cell mass loss rate can be interpreted as the mass release rate of the vent gas. However, during thermal runaway, the cell mass loss accounts not only for the vented gas, but also for the ejected cell content and components that burn away. In addition, this test method does not characterize pressure rise in the chamber during and post-LIB thermal runway. On the other hand, the chamber gauge pressure is limited by a pressure relief valve for safety concerns (see Figure 2)

The presented experimental method provides a framework for characterizing thermal runaway and fires of lithium-ion batteries through the in situ measurement of various parameters in one single test. The detailed time-resolved data also provides empirical parameters for the development of numerical models. For example, the vent gas mass release rate deduced from the cell mass reading and the FTIR gas species readings can be implemented into a computational fluid dynamics (CFD) model as boundary conditions. This removes the need to simulate the electrochemistry of the cell and allows fewer assumptions to be made, resulting in a more general, numerically cost-effective, and precise model for battery fires.

While only the test procedure for a cylindrical cell is presented in the current study, this procedure can be applied to cells of different formats (e.g., pouch or prismatic) and can be readily extended to test thermal runaway propagation between multiple cells in a battery. Also, it is worth noting that the gas concentrations obtained during the thermal runaway process include not only the vent gas but also the combustion products during the battery fire. If the interest is on the vent gas generated before and during the thermal runaway, an inert chamber environment (e.g., argon or nitrogen) should be considered.

Materials

Name	Company	Catalog Number	Comments
Balance	A&D	EJ-6100
Carbon filter	Whatman	WHA67041500
Current transducer	NK Technologies	AT1-010-000-FT
Front camera	Sony	FDR-AX53
FTIR gas analyzer	Fire Testing Technology	Protea atmosFIR AFS-A-15
Heating tape (1.00" x 2.00")	Birk Manufacturing, Inc.	BK3512-19.6-L24-03
High-temperature resistant tape	Kapton
Hydrogen sensor	Amphenol	AX220135
K-type, thermocouple	Omega	KMQSS-020U-12
LabVIEW	National Instruments
Matlab	MathWorks
NI-9213	National Instruments	NI-9213
NI-9219	National Instruments	NI-9219
NI-cDAQ-9174	National Instruments	NI-cDAQ-9174
NI-USB-6009	National Instruments	NI-USB-6009
PID controller	Omega	CN8200
PILOT5000 Chemical Resistant Diaphragm Vacuum Pump	The Lab Depot	TLD5000
Pressure relief valve	Straval	RVL20-10T-N4675
Pressure Transmitter	Keller	0308.01601.081303.02
Pure Nickel Strip (0.1x5x100mm 99.6% Nickel)	U.S. Solid Product
Respirator	McMaster	55865T52
Respirator Cartridge	Honeywell	75Scp100L
Rotary vane vacuum pump (0.5 hp)	Alcatel	Pascal 2010
Side camera	Sony	HDR-CX110
Spot Welder	SUNKKO	737G+
TeamViewer	TeamViewer
Voltage transducer	CR Magnetics Inc.	CR4510-50