- Email: [email protected]
On methods to measure the energetics of a lithium ion battery in thermal runaway
Journal Pre-proof On methods to measure the energetics of a lithium ion battery in thermal runaway J.G. Quintiere PII:
S0379-7112(19)30410-2
DOI:
https://doi.org/10.1016/j.firesaf.2019.102911
Reference:
FISJ 102911
To appear in:
Fire Safety Journal
Received Date: 2 August 2019 Revised Date:
28 September 2019
Accepted Date: 5 November 2019
Please cite this article as: J.G. Quintiere, On methods to measure the energetics of a lithium ion battery in thermal runaway, Fire Safety Journal (2019), doi: https://doi.org/10.1016/j.firesaf.2019.102911. This is a PDF file of an article that has undergone enhancements after acceptance, such as the addition of a cover page and metadata, and formatting for readability, but it is not yet the definitive version of record. This version will undergo additional copyediting, typesetting and review before it is published in its final form, but we are providing this version to give early visibility of the article. Please note that, during the production process, errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain. © 2019 Published by Elsevier Ltd.
On Methods to Measure the Energetics of a Lithium Ion Battery in Thermal Runaway Quintiere J. G. Prof. Emeritus, University of Maryland, Department of Fire Protection Engineering, College Park, MD, USA. [email protected] ABSTRACT Methods are described for measuring the energy released by a Li-ion battery in thermal runaway.. A calorimetry technique is described to measure the exothermic energy released during thermal runaway. The technique uses the battery as a calorimeter with temperature and mass loss measurements to analyze the energetics. Runaway is induced by heating of the battery. Only one battery is investigated over a range of heating power and state of charge (SOC). The dynamics of the battery are investigated including time events, temperature, mass lost and energies. The total energy in runaway is manifested by the internal energy stored in the battery and the enthalpy of the ejected mass. Combustion of the ejected gases is described by a technique using the Cone Calorimeter. Results are reported. Both of these techniques are considered less desireable because of the rapid release of energy therefore two new techniques are proposed. The technique of Lyon&Walters using a standard Bomb Calorimeter with nitrogen instead of oxygen to measure the runaway exothermic energy, and measurement of the resulting ejected gas composition to measure the combustion energy. KEYWORDS: Batteries, calorimeter, Li-ion, thermal runaway.
NOMENCLATURE
c
specific heat
H m Mi p P Q
enthalpy mass molecular weight, species i pressure heater power heat or energy rate of heat or energy
R T U V t
universal gas constant temperature internal energy volume time
Abbreviations HF heat flux HOC heat of combustion SOC state of charge
Greek β constant, Eq. (2) Subscripts b battery Cu copper f final g gas h heater i initial loss energy losses w water
1. INTRODUCTION Lithium ion batteries have a high electrical energy density. These batteries are used extensively in commercial products from electronics to powering automobiles. As with any new technology they can cause unanticipated safety issues. It has become well known that such batteries are prone to “thermal runaway” that can lead to fire and explosion hazards. H. Webster [1,2] has produced a series of reports to document the consequences of thermal runaway with Li-batteries in shipping configurations. There have been review articles that summarize the processes, e.g. [3]. Detailed models that include the kinetics of the decomposition reactions and the transport of heat and electrical current have been presented [4,5,6]. A device that has often been used to study the energetics of batteries before runaway is the Accelerating Rate Calorimeter (ARC) [7]. Such a device is based on adiabaticity and cannot track the rapid process of runaway. In general, the term “thermal runaway” applies to an exothermic chemical process that is accelerated due to temperature. It involves a feedback mechanism in which the exothermic energy produces an increase in temperature, and the temperature causes an increase in the rate of energy generated. In runaway of a Li-ion battery, many exothermic decomposition reactions are triggered among its components, and modeling all of the electrical and chemical processes in runaway is very complex [4-6]. In battery thermal runaway the destruction of the battery occurs due to electric discharge, heat transfer and several chemical reactions. Oxygen can be generated in heating by the Li-metal-oxide cathode that then reacts with the combustible electrolyte. An electrical discharge can follow with Joulian heating along with the destruction of the regulating plastic membrane. Afterwards several exothermic decomposition reactions can occur from the battery componets. This process is very fast, as heat feedback encourages the reactions that increase with temperature. Energy is liberated that can heat the battery to very high temperatures, release decomposition products that are flammable, and therefore transfer heat to its surroundings. A primary hazard from the runaway of a battery is the transfer of this energy to other batteries causing a possible chain reaction among them. This can lead to fire of other items and possible explosion in a confined system. The energy dynamics of runaway are key in the design of safety mitigation systems to prevent the consequences from runaway. A typical Li-ion battery consists of a cathode of a lithium metal oxide on aluminum and a graphite anode on copper. The electrolyte is an organic solvent and is typically combustible. A plastic membrane separator, within the electrolyte, limits the ion transfer between the anode and cathode. The separator is a key failure point in the initiation of runaway. At the onset of runaway oxygen released from the cathode oxide can react with the electrolyte. Failure to runaway can be initiated in several ways: (1) a manufacturing defect that produces dendritic growth breaking the membrane, (2) heating to cause the membrane to melt, (3) direct accidental puncture, and (4) over-charging. Failure of the membrane leads to an internal short circuit and thermal runaway. The internal energy generated within the battery in runaway is composed of resistive heating and the sum of several exothermic decomposition reactions involving the cathode, anode and electrolyte [6]. Gases produced in early reactions pressurize the battery and are usually first intentionally vented by a pressure-relief device. This is followed by a more catastrophic ejection of mass during runaway. The energy output of a battery in thermal runaway is composed of (1) the internal energy from resistive heating and decomposition, (2) the thermal energy of the vented materials, and (3) potentially combustion energy of the vented gases. The vented materials can include gases composed of CO2, CO, H2, CH4, other hydrocarbons and possibly a little oxygen, and solids
composed of copper, graphite and molten aluminum [3]. These three energy forms comprise the potential fire hazard to the surroundings. Measurement techniques have been used to measure the thermodynamic energy and the rate of energy from batteries during runaway. A commercially available technique is an adiabatic Accelerating Rate Calorimeter (ARC) but it is limited to temperature ranges well below the temperatures after the onset of runaway. In Roth et al [7] an ARC measures energy rates up to about 160 ºC. Another technique using the battery itself as a calorimeter was first presented at the FAA 6th Triennial International Fire & Cabin Safety Research Conference by Quintiere and Crowley [8] and also by Liu and Stoliarov. Liu expands upon this method in his thesis [9]. Results will be described herein from a report by Quintiere et al [10] that included the measurement of the exothermic decomposition energy by a “battery calorimeter” and the combustion energy as measured in a combustion calorimeter [11]. It will be shown that the battery calorimeter technique and the use of a combustion calorimeter such as the Cone Calorimeter [11] both have drawbacks. While these techniques are capable of measuring the rate of exothermic and combustion energy, their accuracy is questionable. This is due the very rapid rate of energy release occurring over as little as several seconds. It will be shown for fully charged batteries this release time is so small to warrant its resolution. Instead the total energy released in exothermic decomposition and subsequent combustion are more appropriate to measure. An accurate technique to measure the former has been developed by Walters and Lyon [12] and Lyon and Walters [13] using a standard oxygen bomb device, but with an atmosphere of nitrogen. A method to measure the combustion energy can work off this “nitrogen bomb” technique if the gas mixture composition is measured as demonstrated by Malloney [14]. These measurement methods will be theoretically developed, and results from the “battery calorimeter” and Cone Calorimeter will be presented. Primarily a Tenergy 18650 LiCoO2 battery will be mostly discussed. 2. BATTERY CALORIMETER TO MEASURE EXOTHERMIC DECOMPOSITION ENERGY OF RUNAWAY The calorimeter design is based on an energy balance for the battery. An 18650 battery (18 mm in diameter and 65 mm in length) was considered without its plastic covering. Although the battery is a cylinder, its internal battery element consists of a thin rectangular battery pouch wound into a cylindrical shape. A measurement of the battery temperature and its mass over time in runaway is used to compute the various energy components. The battery is heated by a wrapped powered Nichrome wire to induce runaway. An electric arc is used to ignite the escaping gases at the battery’s vented end. A holder for the battery consisted of a tight thin copper sleeve (0.6 mm), covered by thin ceramic paper with a winding of Nichrome heating wire around its outside. Power (P) to the wire was measured by current and voltage. The entire battery and holder assembly is continuously weighed; and temperature is measured by a flattened thermocouple sandwiched between the copper sleeve and the outer steel shell of the battery. The entire battery and holder is wrapped in thick low density (48 kg/m3) ceramic blanket and jacketed by a steel 70 mm diameter cylindrical shell. The components of the assembly are illustrated in Figure 1. A schematic is shown in Figure 2.
Fig. 1. Calorimeter components.
Fig. 2. Schematic of calorimeter.
2.1 Calibration of calorimeter The battery calorimeter is calibrated by using an energy balance with a solid aluminum cylinder identical in size to the 18650 batteries. The specific heat of the aluminum is taken as 1.05 kJ/kgK at 125 ºC. A control volume (CV) is selected to encompass the aluminum cylinder, copper and ceramic paper sleeves, and the Nichrome heating wires. Heat is lost from the control volume to surroundings comprising the ceramic blanket wool and the exiting electrical wires for power. The aluminum is inert so the mass of the CV-contents is constant and taken as that of copper (21.5 g) and aluminum (47 g). The masses of the ceramic paper and the resistance wire are neglected because they are small. Applying the conservation of energy to the CV with the following assumptions: (1) the battery and copper cylinder are considered thermally thin at the same homogeneous temperature, and (2) the heat loss rate ( ) is by conduction into a semi-infinite media, gives (1) The heat loss rate to the ceramic insulation and from the wires is assumed to be of the form (2) The specific heat of aluminum was derived from the calorimeter analysis and compared to its literature value. The analysis used β =1.4 W-s1/2/K and a specific heat for copper of 0.39 kJ/kgK in two repeated tests each with a power input of 26.4 W. Figure 3 shows the computed specific heat for aluminum. For early times and for corresponding low temperatures, the thermally thin assumption is not appropriate. However, as the heat loss is small in this early period, the aluminum calorimeter appears to give a good calculation for its specific heat.
Fig. 3. Calibration of the calorimeter using an aluminum battery surrogate.
2.2 Battery In this study using the battery calorimeter only one battery was tested. It was a Li-ion 18650 battery with a LiCoO2 cathode having a nominal 3.7 V and 2200 mA-hr rating. This gives a nominal maximum stored energy of 29.3 kJ (3.7v x 2.2A x 3600s). The battery mass was 44.2 +/- 0.1 g over the series tested in which the power input to initiate runaway was varied from about 10 to 74 W, and the State of Charge (SOC) was varied from about 0 to 100 %. Here the SOC is found by a discharge of the battery from its rated capacity. 2.3 Temperature Uniformity of Battery Several tests were made to investigate the assumption of the battery and copper sleeve as thermal thin and uniform in temperature at any time. The internal battery temperature was measured at its axial and lateral center along with the normal calorimeter temperature measured on the outer center of the battery between it and the tight fitting copper sleeve. Figure 4a shows the results. Also Figure 4b shows the results for the outer center and end of the battery in another test. The results indicate that the battery temperature is reasonably homogeneous within about 20 ºC, however there is an anomaly for the center-outer temperature after runaway. From Figure 4a it can be seen that there is a delay in response to the outer battery shell of about 20 s and about 60 s for the interior of the battery. So the thermally thin assumption for the system is reasonable after 60 s.
Fig. 4a. Temperature within battery and outer shell.
Fig. 4b. Outer center and end temperatures.
2.4 Specific Heat of Battery The specific heat of the battery was determined for each thermal runaway experiment during the inert battery phase before runaway. This inert period was primarily below 200ºC for this battery. Equation (1), with the battery substituted for the aluminum, was used to reduce the data. Figure 5 shows a typical determination for the battery specific heat. The inaccuracy below about 100 ºC is due to the delayed temperature response of the battery. The specific heat did not significantly vary over the range of testing this battery and was found to be 0.95 +/- 0.03 kJ/kg-K. This value was taken as constant for all analyses in the tests, in spite of mass lost from the battery during runaway.
Fig. 5. Typical determination of the battery specific heat before runaway.
2.5 Governing equation for calorimeter In these tests, runaway was induced by heating of the outer battery. The heating power was varied in the tests as well as the SOC of the battery. The heating likely melted of the plastic membrane separator and caused an internal electrical discharge and runaway. This electrical and chemical energy is manifested within the battery as a generation term , and outside of the battery by an ejection term of enthalpy from the battery. The conservation of mass for the system CV yields that the rate of mass ejected is equal to the rate of change for the battery mass, m. As shown in [15], the conservation of energy states that the rate of internal energy plus the rate of ejected enthalpy is
equal to the sum of the heating power (P) and the generation term ( ceramic wool (
) minus the heat loss into the
) given by Eq. (2). Rearranging to solve for the internal generation term gives (3)
The values for the constants in Eq. (3) are: (mc)Cu = 8.4 J/K, cb = 0.95 J/g-K, and cg is assumed as 1.05 J/g-K. The value of the specific heat for the ejected materials is selected for expedience as their components are not known and can comprise gases of hydrogen, carbon dioxide, methane and others, as well as possibly molten aluminum. At 1000 K, the specific heats of hydrogen, carbon dioxide and methane are respectively, 15, 1.2 and 4.5 J/g-K. Therefore the uncertainty of the specific heat for the ejected mixture as well as ignoring any possible melting leads to an underestimation for this ejected enthalpy rate term given as .
(4)
This term will be relatively small compared to the exothermic battery energy.
It will be shown that the battery undergoes very rapid changes during runaway. As a consequence the computation of time derivatives from the temperature and mass measurements can be problematic. Moreover the practical manner to assess the hazard of the battery in runaway is not to have the precise rate of energy released, but to have the total energy, realizing this is a potential transfer to the surroundings in a short time. Therefore, Eq. (3) is integrated to obtain the energy released over time. (5) with the ejected energy as Qg ≡ cg
∫
m mi
T dm . Equation (5) will be used to reduce the data for a series
of tests on the battery varying SOC and P. In all the tests the heating power was initiated at the start and kept constant throughout the test. An electric arc was manually initiated at the start of venting to attempt to ignite the gases. This was not always successful due to timing and location, and at times auto ignition could occur before the arc was struck. 2.6 Typical data and analyses Test 17 using a heating power of 25.6 W and a SOC of 100 % is examined to show the typical results for this battery. The raw data of battery temperature and mass are shown in Figure 6. There is a nearly linear rise in temperature until about 500 s when the pressure safety vent opens and releases white vapors. In general these vapors could be ignited but ignition was not achieved in this test at the first vent. These vapors continue to vent until about 550 s where thermal runaway occurs. The feedback between battery temperature and decomposition takes place over several seconds. This is seen as a nearly vertical jump in temperature and drop in battery mass. Runaway is a violent
Fig. 6. Battery temperature and mass at 100 % SOC and heating power at 25.6 W.
process with a high velocity jet from the pressure safety end of the battery. This ejection is about 35 % of the original battery mass. In several seconds the battery temperature jumped from about 200 ºC to 800 ºC. Runaway is a dramatic event. The runaway event is examined more closely over the jump period (551.5 to 554.5 s). Figure 7 shows the temperature jump and Figure 8 shows the rate of internal energy computed, . The dramatic jump in 3 seconds shows the characteristic of thermal runaway in a battery. While the rate of internal energy generation can be computed, its distribution over time may not be as significant as the total energy dumped in this process. However ejected energy from the battery commenced at the first vent (~ 500 s) at about 190 ºC; then most of this category of energy was released during the jump after the second vent (~ 550 s). Consequently most of the battery energy converted from its original electrochemical energy is primarily transferred in this jump period, and will be represented by Qb and Qg over the entire heating period of the battery as given in Eqns. (5) and (6).
Fig. 7. Temperature at jump.
Fig. 8. Internal generation rate of energy.
Figure 9 shows the specific energy terms in Eq. (5) individually computed for Test 17. As Qb is computed from the algebraic sum of the RHS terms some inaccuracy can be noted as its value is about –1.5 kJ before the jump instead of zero. However after the jump this internal energy computed is 30 kJ. Qg reaches about 18 kJ. Therefore the total energy of the battery released during runaway is about 48 kJ, compared to its rated energy at 100 % SOC of 29.3 kJ. Of course this does not include the combustion energy that can be released if the vented gases are ignited.
Fig. 9. Battery energy terms over time.
3. CALORIMETER DECOMPOSITION RESULTS About 25 tests were made with the same battery to assess the effects of the heating power the SOC. It might be expected that energy in runaway would be related to the electrical energy stored, and that the heating power should affect the time to runaway. The results are shown in the remainder of this section and the trend lines for the data are based on a linear or smooth fit with judgement to indicate a linear invariance. 3.1 Effect of Heating Power In this series the battery SOC was maintained at 80 % and the heating power varied from about 10 to 74 W for the duration of each test. The heating power reduces the time for runaway, as the safety vent operation and for the second venting at thermal runaway both decrease with heating (Figure 10). In the configuration of the battery in the calorimeter, all heating powers were able to initiate runaway. The duration of the runaway event varied from 3 to 12 s without a definite trend with the heating power. Although the times for venting and runaway varied significantly, the associated temperatures only varied slightly. Figure 11 shows these slight trends where the trigger temperature for runaway is about 220 to 300 ºC. During runaway the temperatures rise to over 750 ºC in 3 to 12 s. This rapid rise is indicative of the battery hazard in thermal runaway. Figure 12 shows the mass loss fractions of the battery after the first safety vent and the runaway vent occur. The mass loss fractions are not significantly dependent on the heat power. About 10 % of the original mass is ejected during the
safety venting and up to 40 % is lost after runaway. This indicates that more hot stuff is ejected following runaway, and much of this is combustible.
Fig. 10. Vent times as a function of heating.
Fig. 11. Temperature at onset and after runaway.
Fig. 12. Battery mass lost at venting. Fig. 13. Battery runaway energies vs heating rate. Figure 13 shows a slight effect of heating power on the internal energy and on the ejected energy. The former is about 30 to 35 kJ for this 80 % SOC and the latter is about 10 to 18 kJ. Hence heating power significantly effects the time for venting and runaway, but not significantly the vent and runaway temperatures, and mass ejected and energy produced.
3.2 Effect of State of Charge The effect of SOC is now examined. Here the heating power is essentially constant for all the tests at about 25 +/- 2 W. Figure 14 shows the times to safety vent (1st) and runaway (2nd) are nearly invariant with SOC. Liu [9] has also seen this invariance. The first vent time is likely invariant with the SOC because it occurs due to a designed pressure relief for the battery, and due to the constant internal volume of the battery the pressure is directly related to temperature. Once this temperature occurs it likely kicks off the kinetics of the runaway exothermic reaction leading to full
development of the reaction by the second vent time. The time interval between the two vent events is roughly about 20 % of the first vent time. The temperatures at of the 1st vent and the onset runaway are also invariant with SOC as the relief pressure controls the 1st, and the second might be considered an ignition temperature or the onset of the full exothermic reaction. However, the final temperature of the battery increases with the SOC (Figure 15). This is an indication of the initial electrochemical energy of the battery corresponding to its SOC. The 1st vent temperature is indicative of the designed pressure relief vent configuration, and is due solely to the heating power. The 2nd vent temperature is indicative of the onset of thermal runaway that depends on the battery and its heat loss environment. The temperature after runaway depends on the internal energy remaining in the battery and its ejected energy.
Fig. 14. Vent times vs SOC.
Fig. 15. Temperature vs SOC.
Fig. 16. Battery mass vs SOC.
Fig. 17. Battery energies vs SOC.
Fig. 18. Runaway time period vs SOC.
Fig. 19. Average runaway power vs runaway period.
Figures 16 and 17 show that, at runaway, the mass ejected and the energies released (internal and ejected) are directly related to the SOC. If runaway is perceived as an internal short circuit, then a portion of the original battery energy would be represented as resistance heating. As the remaining portion of the energy is due to decomposition dependent on temperature then it too will increase with SOC as Figure 15 shows the temperature after runaway increases with SOC. As the mass ejected increases after runaway, so does the combustion potential hazard of these gases. Here the “runaway period” is defined as the time increment from just after the onset of runaway to just before the end of runaway. The onset of runaway is indicated by the time of the 2nd vent. The end of runaway is indicated by the abrupt decrease in battery temperature. The runaway period decreases dramatically with SOC – from nearly 100 s at 20 % to as low as 0.5 s at 80% at which the runaway temperature nearly reached 1400 ºC (Figs. 15 and 18). The average runaway power in this period can be computed from the total energy released. Figure 19 shows how this varies with the runaway period. Nearly 100 kW can be released in ½ second! Figure 20 shows a comparison between the current results and a similar battery calorimeter developed by Liu [9]. He only computed the internal energy Qb, not the additional vented energy, Qg. Figure 20 shows good agreement with his results for Qb and the present.
Fig. 20. Comparison of runaway battery energy by battery calorimeter measurement 4. COMBUSTION ENERGY BY CONE CALORIMETER The Cone Calorimeter is a fire based analytical instrument designed to measure the combustion energy of the burned fire product gases [11]. In reference [10] its sample tray was fabricated to ensure a reduction of the jet momentum of ejected battery matter, and avoid propulsion of the battery out of the Cone test space. The electric arc ignition source was removed and the radiant heater intitiated runaway at a given heat flux. The ejected gases rose through the cone heater and were auto-ignited. The standard method for the oxygen depletion of the burning gases was used to measure the rate of heat release. Sean Crowley conducted Cone combustion experiments for a number of batteries [10]. The energy release period was very short in time, and the normal data reduction process of the Cone Calorimeter was not fast enough to compute the rate of energy released by combustion. Consequently a de-convolution algorithm developed by Richard Lyon was used to obtain an accurate rate from the time lag of the measuring process [10]. An example is shown in Figure 21.
Fig. 21. Panasonic LIB tested at 50 kW/m2 external heat flux showing raw and de-convoluted data in Cone Calorimeter (taken from [10].
The SOC charge had the most influence on the combustion energy. Figures 22 and 23. Here only mass loss and rate of combustion energy released were recorded as a function of heat flux to the battery (10 to 75 kW/m2), and as a function of the state of charge (SOC) for one battery at a heat flux of 50 kW/m2. Figure 22a shows the total combustion energy and mass lost for each battery test over a range of SOC. Figure 22b shows the effective heat of combustion of the vented battery gases. Both the Energy in Combustion and the effective heat of combustion decrease with SOC. It is not clear why they decrease. As there is some small amount of oxygen released in the breakdown of the cathode lithium oxide, it is possible that some of the combustion energy is from within the battery, not by combustion with ambient air. The cone will not record the internal combustion; moreover the high velocity jet associated with the combustion could affect the capture of the exhasut gases by the cone calorimeter. Other studies have also found this decrease with SOC for other Li-ion batteries [9,10,16]. Overall effective heat of combustion ranged from about 2 to 26 kJ/g-lost for 9 different batteries [10]. With the SOC at 90%, the same battery exhibited limited effect of heat flux above a threshold of 20 kW/m2 as shown in Figure 23. 120
20
Combustion Data from Cone
LG Heat Flux 50 kW/m
100
Energy/Mass lost, Eff. HoC kJ/g
Energy in Combustion kJ and Mass Lost g
Combustion Data from Cone 2
80
60 Tot HR kJ 40 Mass loss g 20
LG Heat Flux 50 kW/m
2
15
10
5
0
0 0
20
40
60
80
100
0
20
40
60
80
100
SOC %
SOC %
(a)
(b) 2
Fig. 22. LG Li-ion battery at a heating rate of 50 kW/m : Energy of combustion, mass lost, and effective HOC [10] 12
Energy/Mass lost, Eff. HoC kJ/g
100
80
60
Energy kJ
No 2nd vent
Energy in Combustion kJ and Mass Lost g
Combustion Data from Cone LG SOC 90%
40
Mass Loss g
20
0
Combustion Data from Cone LG SOC 90% 10
8
6
No 2nd vent
120
4
2
0 0
10
20
30
40
50 2
HF kW/m
(a)
60
70
80
0
10
20
30
40
50 2
60
70
80
HF kW/m
(b)
Fig. 23. LG Li-ion battery combustion energy, mass loss, and effective heat of combustion as a function of heating rate [10]
5. ALTERNATIVE METHODS FOR MEASURING EXOTHERMIC AND COMBUSTION ENERGY 5.1 Exothermic Decomposition Energy An alternative technique to measure the overall energy released by the runaway energy was developed by Lyon and Walters [12,13]. They modified an Oxygen Bomb apparatus using instead an atmosphere of nitrogen. The Oxygen Bomb with correct calibration is an accurate standard method for measuring the heat of combustion. Therefore, its use with nitrogen surrounding a battery in thermal runaway will measure all of its energy released. A comparison of their results with the total energy (Qb + Qg) measured by the battery calorimeter for nearly the same battery is shown in Figure 24. The battery tested by Lyon and Walters used a newer version of the Tenergy 18650 with a slightly higher capacity of 2600 mA-hr over 2200. Recall that a weakness of the current battery calorimeter method is the uncertainty of the specific heat of the material ejected, and there is no consideration in the analysis for melting. The “Nitrogen Bomb” of Lyon and Walters is believed to be more accurate in capturing the total energy released, especially above 80 % SOC where the calorimeter energy is lower. The variation in the results for a given SOC could be attributed to the slight differences in the batteries or simply the chaotic nature of runaway. The total runaway energy is approximately two times higher than the corresponding rated energy at each SOC. It is not uncommon for batteries in runaway to release several times their original electrical energy capacities.
Fig. 24. Current method with nitrogen bomb [12,13].
5.2 Governing Equations for the Nitrogen Bomb Calorimeter Equations are formulated here to define the procedure for the Nitrogen Bomb to measure the runaway energy with attention to calibration. This is not been completely done before. The conservation of energy is applied to a closed system comprised of the Oxygen Bomb apparatus containing a battery with a NiChrome wire heating assembly. The system includes the bomb, the water bath, and the insulated walls of the outer container. The water bath experiences normal ambient pressure while the pressure varies in the sealed bomb. A schematic is shown in Figure 25.
Fig. 25. Schematic of Nitrogen Bomb system
The conservation of energy is applied to the entire closed system in which the final internal energy is equal to the initial plus that from the heater. As a closed system, the change in internal energy of all the components is equal to the intentional heat added by an electrical heater around the battery, Q h.
U f = U i + Qh
(6)
Introduce enthalpy H = U + pV,
( )
∆H − ∆ pV = Qh
(7)
This enthapy change can be broken down into the various system components. The pressure work term is present for the gases enclosed in the bomb. ∆H = ∆H b,battery+s,structure+w,water + ∆H g ,batterygaes+nitrogen and ∆ pV = ∆ pV g (8)
( )
( )
Now let us represent the gases generated by the battery in runaway to globally include the chemical term representing the total energy in decomposition and electrical heating in runaway. Its enthapy change per unit mass can be expressed as a chemical/electrical term of runaway and its specific heat component. This is an approximent representation for the energy of runaway.
(
∆hg = −∆hd + cg T − Ti
)
Specifically representing all the terms, cb(mb,fTf – mb,iTi) is the change in energy of the solid battery components,
(Cs + mwcw)(Tf – Ti) is the change in energy of the “bomb” structure and the water, cgmN2(Tf – Ti) is the change in enthalpy of the nitrogen added to the “bomb”, mg∆hg is the change in enthalpy association with the electro chemical reaction, and
(9)
(pg,f – pg,i)V is the pressure work term. Then the energy equation becomes cb(mb,fTf – mb,iTi) + (Cs + mwcw)(Tf – Ti) + cgmN2(Tf – Ti) + mg[(-∆hd +cg(Tf – Ti) - (pg,f – pg,i)V = Qh
(10) The effective heat of decomposition (involving both exothermic chemical reactions and electrical dissipation) given off from the battery runaway is
(
(
Qd = mg ∆hd = Cs + mw cw + cg mN 2 + mg
(
)
)) (T
f
) (
− Ti + cb mb,, f T f − mb,iTi
)
− pg, f − p,g ,i V - Qh (11)
Lyon and Walters [13] did not include the heat capacity terms of the gas contained in the bomb free space. They considered the total energy released by the battery as the first four terms on the righthand-side. In the execution of the analysis it can be shown that the dominate terms are the first and the heat addition, Qh. To make this analysis work the Bomb must be calibrated. Standard calibration with oxygen to measure the heat of combustion normally will give the Cs + mwcw. Here the Bomb can be calibrated with a know heat input and a spent battery after a runaway test. Based on the conservation of mass
mb,i = mb, f + mg
.
(12)
Then the calibration involves measuring
(C + m c s
w w
)(
) (
)
(
+ cb mb,, f T f − Ti = pg , f − p,g ,i V + Qh − cg mN 2 T f − Ti
) (13)
From a standard calibration Cs is found. The mass and specific heat of water is known in the experiment. In the calibration with the battery cbmb,f can be found. Note this includes any solid or condensed material in the ejection. The Equation of State allows for the computation of the nitrogen and combustion gas mixture masses.
mN 2 =
and
piVM N 2 RTi
initially
(14)
mN + mg ≈ 2
p f VM N RT f
2
mN 2 + mg =
p f VM N 2 RT f finally. (15)
Lyon and Walters [12,13] executed such an analysis by calibrating with a cyliinder of aluminum. An example of their results is shown in Figure 24. 5.3 Combustion Energy Here the technique developed by Maloney [14] is adapted to estimate the combustion energy. Following the execution of the procedure to measure the decomposition runaway energy, the gas concentrations contained in the bomb can be measured. Maloney conducted battery runaway experiments in a nitrogen closed system and measured the composition of the combustion gases. These gases included hydrogen, carbon dioxide, carbon monoxide, and several hydrocarbons. He presents the partial volumes of these compositions at 10 psia, 27 ºC and for various SOC levels. To illustrate the technique, his results for a Tenergy LiCoO2 18650 (2600 mA-hr) battery at 100 % SOC is considered. His results are tablulated in Table 1 along with properties needed to compute the total available combustion energy. He did not compute the combustion, so it is now done for one of his experiments to investigate its feasibility and accuracy against combustion results from the Cone Calorimeter. The mass of each component gas can be computed from the Equation of State using the partial volumes at 10 psia and 27 ºC. In general the equation becomes
mi (g) = 0.02764Vi (L)M i (g/mol) .
(16)
The theoretical combustion energy from each component is determined by their complete heat of combustion. Table 1 shows two total values based on assuming the properties of the THC, and assuming the THC are composed of only methane and ethylene. This gives, respectively, 63.8 and 49.7 kJ. Results from the Cone Calorimeter at the same SOC (100%) for the 2200 mA-hr battery was found to be 48 and 55 kJ. It is felt that the gas composition method and Cone Calorimeter values are within reasonable aggreement with the expected level of accuracy and the slight variations in the batteries. Table 1. Vented gas composition Tenergy LiCoO2 18650 (2600 mA-hr) battery at 100 % SOC [14] Gas
Partial Volume*, L
Molecular weight, g/mol
Mass, g
Heat of combustion, kJ/g
Combustion energy, kJ
Hydrogen, H2
1.5
2
0.0829
141.6
11.7
Carbon monoxide, CO
2.0
28
1.55
10.1
15.7
Total hydrocarbons, THC
0.7
~ 40
0.774
~ 47
36.4
Methane, CH4
0.35 (assumed)
16
0.155
55.5
8.59
Ethylene, C2H4
0.35 (assumed)
28
0.271
50.3
13.62
Total with THC
63.8
Total with CH4&C2H4
49.7
* partial volumes at 27 ºC and 10 psia 6. CONCLUSIONS Methods have been presented to measure the energy liberated by a Li ion battery in runaway due to is electro-exothermic decomposition and its potential combustion energy from ejected gases in runaway. The battery calorimeter developed here is capable for measuring the rate and total energy of decomposition for a battery in runaway. The Cone Calorimeter is also capable of measuring the rate of combustion of the gases ejected in runaway. Both forms of energy pose a potential hazard to the battery surroundings. It is shown that the runaway decomposition energy increases with the SOC, and combustion energy is also high. As the energy release occurs over a very short time, the hazard is more aptly described in terms of the total energy released. In this regard, more accurate thermodynamic measurement techniques are found to be preferred for their simplicity and accuracy. This follows the technique developed by Lyon and Walters using an Oxygen Bomb with nitrogen to measure the runaway decomposition energy, and subsequent measurement of the combustion gases to determine the combustion. This paper has shown the comparability of these thermodynamic techniques, and recommends them for further consideration. 7. ACKNOWLEDGEMENTS This work was supported by the FAA Technical Center, where the experimental support of Sean Crowley was essential, and discussions with Richard Walters, Richard Lyon and Thomas Maloney were invaluable. This paper has been expanded from a previous work [16], and partially supported by the Pipeline and Hazardous Materials Safety Administration, PHMSA # 693JK3193C000007. 8. REFERENCES [1] H. Webster, Flammability Assessment of Bulk-Packed, Rechargeable Lithium-Ion Cells in Transport Category Aircraft, Report No. DOT/FAA/AR-06/38, Federal Aviation Administration, 2006. [2] H. Webster, Fire Protection for the Shipment of Lithium Batteries in Aircraft Cargo Compartments, Report No. DOT/FAA/AR-10/31, Federal Aviation Administration, 2010.
[3] C. Mikolajczak, M. Kahn, K. White, R.T. Long, Lithium-ion Batteries Hazard and Use Assessment, Report No. 1100034.000/A0F0/0711/CM01, The Fire Protection Research Foundation, 2011. [4] D.H. Doughty, P.C. Butler, R.G. Jungst, E.P. Roth, Lithium Battery Thermal Models, J. Power Sources 110 (2002) 357-363. [5] R. Spotnitz, J. Franklin, Abuse Behavior of High-power, Lithium-ion Cells, J. Power Sources 113 (2003) 81-100. [6] Q. Wang, P. Ping, X. Zhao, G. Chu, J. Sun, C. Chen, Thermal runaway caused fire and explosion of lithium ion battery, Journal of Power Sources, 208 (2012) 210–224. [7] E.P. Roth, D.H. Doughty, Thermal Abuse of High-power 18650 Li-ion Cells, J. Power Sources, 128 (2004) 308-318. [8] J.G. Quintiere and S. Crowley, The Sixth Triennial International Fire & Cabn Safety Research Conference, FAA Technical Center, Philadelphia Marriott Downtown, Dec 2-5, 2013. [9] X. Liu, Comprehensive calorimetry and modeling of the thermally-induced failure of a lithium ion battery, PhD Thesis, Department of Fire Protection Engineering, University of Maryland, College Park MD, 2016. [10] J.G. Quintiere, S. Crowley, R.N. Walters, R.E. Lyon, D. Blake, Fire Hazards of Lithium Batteries, DOT/FAA/TC-TN15/17, Federal Aviation Admin. William J. Hughes Technical Center, Atlantic City Int. Airport, NJ 08405, February 2016. [11] ASTM E7354-04a, Standard Test Method for Heat and Visible Smoke Release Rates for Materials and Products Using an Oxygen Consumption Calorimeter, ASTM International, West Conshohocken, 2004. [12] R.N. Walters, R.E. Lyon, Measuring Energy Release of Lithium-ion Battery Failure using a Bomb Calorimeter, Report No. DOT/FAA/TC-TN15/40, Federal Aviation Administration, March 2016. [13] R.E. Lyon, R.N. Walters, Energy Release by Rechargeable Lithium-Ion Batteries in Thermal Runaway, Report No. DOT/FAA/TC-TN16/22, Federal Aviation Administration, April 2016. [14] T. Maloney, Lithium Battery Thermal Runaway Vent Gas Analysis Report No. DOT/FAA/TC-15/59, Federal Aviation Administration, November 2016. [15] J.G. Quintiere, Fundamentals of Fire Phenomena, John Wiley and Sons, Ltd., Chichester, 2006, pp 66-67. [16] J.G. Quintiere, Measuring the Energetics of a Lithium Ion Battery in Thermal Runaway, Proceedings of the Ninth International Seminar on Fire and Explosion Hazards (ISFEH9), Edited by Bradley D., Chaumeix N., Liu N. A., Molkov V., Snegirev A. and Tamanini F.Saint-Petersburg Polytechnic University Press, 2019.
AUTHOR DECLARATION I wish to confirm that there are no known conflicts of interest associated with this Publication, and there has been no significant financial support for this work that could have influenced its outcome. I understand that the Corresponding Author is the sole contact for the Editorial process and is responsible for submissions of revisions and final approval of proofs. I confirm that I have provided a current, correct email address. [email protected] Signed
August 2, 2019
S0379-7112(19)30410-2
DOI:
https://doi.org/10.1016/j.firesaf.2019.102911
Reference:
FISJ 102911
To appear in:
Fire Safety Journal
Received Date: 2 August 2019 Revised Date:
28 September 2019
Accepted Date: 5 November 2019
Please cite this article as: J.G. Quintiere, On methods to measure the energetics of a lithium ion battery in thermal runaway, Fire Safety Journal (2019), doi: https://doi.org/10.1016/j.firesaf.2019.102911. This is a PDF file of an article that has undergone enhancements after acceptance, such as the addition of a cover page and metadata, and formatting for readability, but it is not yet the definitive version of record. This version will undergo additional copyediting, typesetting and review before it is published in its final form, but we are providing this version to give early visibility of the article. Please note that, during the production process, errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain. © 2019 Published by Elsevier Ltd.
On Methods to Measure the Energetics of a Lithium Ion Battery in Thermal Runaway Quintiere J. G. Prof. Emeritus, University of Maryland, Department of Fire Protection Engineering, College Park, MD, USA. [email protected] ABSTRACT Methods are described for measuring the energy released by a Li-ion battery in thermal runaway.. A calorimetry technique is described to measure the exothermic energy released during thermal runaway. The technique uses the battery as a calorimeter with temperature and mass loss measurements to analyze the energetics. Runaway is induced by heating of the battery. Only one battery is investigated over a range of heating power and state of charge (SOC). The dynamics of the battery are investigated including time events, temperature, mass lost and energies. The total energy in runaway is manifested by the internal energy stored in the battery and the enthalpy of the ejected mass. Combustion of the ejected gases is described by a technique using the Cone Calorimeter. Results are reported. Both of these techniques are considered less desireable because of the rapid release of energy therefore two new techniques are proposed. The technique of Lyon&Walters using a standard Bomb Calorimeter with nitrogen instead of oxygen to measure the runaway exothermic energy, and measurement of the resulting ejected gas composition to measure the combustion energy. KEYWORDS: Batteries, calorimeter, Li-ion, thermal runaway.
NOMENCLATURE
c
specific heat
H m Mi p P Q
enthalpy mass molecular weight, species i pressure heater power heat or energy rate of heat or energy
R T U V t
universal gas constant temperature internal energy volume time
Abbreviations HF heat flux HOC heat of combustion SOC state of charge
Greek β constant, Eq. (2) Subscripts b battery Cu copper f final g gas h heater i initial loss energy losses w water
1. INTRODUCTION Lithium ion batteries have a high electrical energy density. These batteries are used extensively in commercial products from electronics to powering automobiles. As with any new technology they can cause unanticipated safety issues. It has become well known that such batteries are prone to “thermal runaway” that can lead to fire and explosion hazards. H. Webster [1,2] has produced a series of reports to document the consequences of thermal runaway with Li-batteries in shipping configurations. There have been review articles that summarize the processes, e.g. [3]. Detailed models that include the kinetics of the decomposition reactions and the transport of heat and electrical current have been presented [4,5,6]. A device that has often been used to study the energetics of batteries before runaway is the Accelerating Rate Calorimeter (ARC) [7]. Such a device is based on adiabaticity and cannot track the rapid process of runaway. In general, the term “thermal runaway” applies to an exothermic chemical process that is accelerated due to temperature. It involves a feedback mechanism in which the exothermic energy produces an increase in temperature, and the temperature causes an increase in the rate of energy generated. In runaway of a Li-ion battery, many exothermic decomposition reactions are triggered among its components, and modeling all of the electrical and chemical processes in runaway is very complex [4-6]. In battery thermal runaway the destruction of the battery occurs due to electric discharge, heat transfer and several chemical reactions. Oxygen can be generated in heating by the Li-metal-oxide cathode that then reacts with the combustible electrolyte. An electrical discharge can follow with Joulian heating along with the destruction of the regulating plastic membrane. Afterwards several exothermic decomposition reactions can occur from the battery componets. This process is very fast, as heat feedback encourages the reactions that increase with temperature. Energy is liberated that can heat the battery to very high temperatures, release decomposition products that are flammable, and therefore transfer heat to its surroundings. A primary hazard from the runaway of a battery is the transfer of this energy to other batteries causing a possible chain reaction among them. This can lead to fire of other items and possible explosion in a confined system. The energy dynamics of runaway are key in the design of safety mitigation systems to prevent the consequences from runaway. A typical Li-ion battery consists of a cathode of a lithium metal oxide on aluminum and a graphite anode on copper. The electrolyte is an organic solvent and is typically combustible. A plastic membrane separator, within the electrolyte, limits the ion transfer between the anode and cathode. The separator is a key failure point in the initiation of runaway. At the onset of runaway oxygen released from the cathode oxide can react with the electrolyte. Failure to runaway can be initiated in several ways: (1) a manufacturing defect that produces dendritic growth breaking the membrane, (2) heating to cause the membrane to melt, (3) direct accidental puncture, and (4) over-charging. Failure of the membrane leads to an internal short circuit and thermal runaway. The internal energy generated within the battery in runaway is composed of resistive heating and the sum of several exothermic decomposition reactions involving the cathode, anode and electrolyte [6]. Gases produced in early reactions pressurize the battery and are usually first intentionally vented by a pressure-relief device. This is followed by a more catastrophic ejection of mass during runaway. The energy output of a battery in thermal runaway is composed of (1) the internal energy from resistive heating and decomposition, (2) the thermal energy of the vented materials, and (3) potentially combustion energy of the vented gases. The vented materials can include gases composed of CO2, CO, H2, CH4, other hydrocarbons and possibly a little oxygen, and solids
composed of copper, graphite and molten aluminum [3]. These three energy forms comprise the potential fire hazard to the surroundings. Measurement techniques have been used to measure the thermodynamic energy and the rate of energy from batteries during runaway. A commercially available technique is an adiabatic Accelerating Rate Calorimeter (ARC) but it is limited to temperature ranges well below the temperatures after the onset of runaway. In Roth et al [7] an ARC measures energy rates up to about 160 ºC. Another technique using the battery itself as a calorimeter was first presented at the FAA 6th Triennial International Fire & Cabin Safety Research Conference by Quintiere and Crowley [8] and also by Liu and Stoliarov. Liu expands upon this method in his thesis [9]. Results will be described herein from a report by Quintiere et al [10] that included the measurement of the exothermic decomposition energy by a “battery calorimeter” and the combustion energy as measured in a combustion calorimeter [11]. It will be shown that the battery calorimeter technique and the use of a combustion calorimeter such as the Cone Calorimeter [11] both have drawbacks. While these techniques are capable of measuring the rate of exothermic and combustion energy, their accuracy is questionable. This is due the very rapid rate of energy release occurring over as little as several seconds. It will be shown for fully charged batteries this release time is so small to warrant its resolution. Instead the total energy released in exothermic decomposition and subsequent combustion are more appropriate to measure. An accurate technique to measure the former has been developed by Walters and Lyon [12] and Lyon and Walters [13] using a standard oxygen bomb device, but with an atmosphere of nitrogen. A method to measure the combustion energy can work off this “nitrogen bomb” technique if the gas mixture composition is measured as demonstrated by Malloney [14]. These measurement methods will be theoretically developed, and results from the “battery calorimeter” and Cone Calorimeter will be presented. Primarily a Tenergy 18650 LiCoO2 battery will be mostly discussed. 2. BATTERY CALORIMETER TO MEASURE EXOTHERMIC DECOMPOSITION ENERGY OF RUNAWAY The calorimeter design is based on an energy balance for the battery. An 18650 battery (18 mm in diameter and 65 mm in length) was considered without its plastic covering. Although the battery is a cylinder, its internal battery element consists of a thin rectangular battery pouch wound into a cylindrical shape. A measurement of the battery temperature and its mass over time in runaway is used to compute the various energy components. The battery is heated by a wrapped powered Nichrome wire to induce runaway. An electric arc is used to ignite the escaping gases at the battery’s vented end. A holder for the battery consisted of a tight thin copper sleeve (0.6 mm), covered by thin ceramic paper with a winding of Nichrome heating wire around its outside. Power (P) to the wire was measured by current and voltage. The entire battery and holder assembly is continuously weighed; and temperature is measured by a flattened thermocouple sandwiched between the copper sleeve and the outer steel shell of the battery. The entire battery and holder is wrapped in thick low density (48 kg/m3) ceramic blanket and jacketed by a steel 70 mm diameter cylindrical shell. The components of the assembly are illustrated in Figure 1. A schematic is shown in Figure 2.
Fig. 1. Calorimeter components.
Fig. 2. Schematic of calorimeter.
2.1 Calibration of calorimeter The battery calorimeter is calibrated by using an energy balance with a solid aluminum cylinder identical in size to the 18650 batteries. The specific heat of the aluminum is taken as 1.05 kJ/kgK at 125 ºC. A control volume (CV) is selected to encompass the aluminum cylinder, copper and ceramic paper sleeves, and the Nichrome heating wires. Heat is lost from the control volume to surroundings comprising the ceramic blanket wool and the exiting electrical wires for power. The aluminum is inert so the mass of the CV-contents is constant and taken as that of copper (21.5 g) and aluminum (47 g). The masses of the ceramic paper and the resistance wire are neglected because they are small. Applying the conservation of energy to the CV with the following assumptions: (1) the battery and copper cylinder are considered thermally thin at the same homogeneous temperature, and (2) the heat loss rate ( ) is by conduction into a semi-infinite media, gives (1) The heat loss rate to the ceramic insulation and from the wires is assumed to be of the form (2) The specific heat of aluminum was derived from the calorimeter analysis and compared to its literature value. The analysis used β =1.4 W-s1/2/K and a specific heat for copper of 0.39 kJ/kgK in two repeated tests each with a power input of 26.4 W. Figure 3 shows the computed specific heat for aluminum. For early times and for corresponding low temperatures, the thermally thin assumption is not appropriate. However, as the heat loss is small in this early period, the aluminum calorimeter appears to give a good calculation for its specific heat.
Fig. 3. Calibration of the calorimeter using an aluminum battery surrogate.
2.2 Battery In this study using the battery calorimeter only one battery was tested. It was a Li-ion 18650 battery with a LiCoO2 cathode having a nominal 3.7 V and 2200 mA-hr rating. This gives a nominal maximum stored energy of 29.3 kJ (3.7v x 2.2A x 3600s). The battery mass was 44.2 +/- 0.1 g over the series tested in which the power input to initiate runaway was varied from about 10 to 74 W, and the State of Charge (SOC) was varied from about 0 to 100 %. Here the SOC is found by a discharge of the battery from its rated capacity. 2.3 Temperature Uniformity of Battery Several tests were made to investigate the assumption of the battery and copper sleeve as thermal thin and uniform in temperature at any time. The internal battery temperature was measured at its axial and lateral center along with the normal calorimeter temperature measured on the outer center of the battery between it and the tight fitting copper sleeve. Figure 4a shows the results. Also Figure 4b shows the results for the outer center and end of the battery in another test. The results indicate that the battery temperature is reasonably homogeneous within about 20 ºC, however there is an anomaly for the center-outer temperature after runaway. From Figure 4a it can be seen that there is a delay in response to the outer battery shell of about 20 s and about 60 s for the interior of the battery. So the thermally thin assumption for the system is reasonable after 60 s.
Fig. 4a. Temperature within battery and outer shell.
Fig. 4b. Outer center and end temperatures.
2.4 Specific Heat of Battery The specific heat of the battery was determined for each thermal runaway experiment during the inert battery phase before runaway. This inert period was primarily below 200ºC for this battery. Equation (1), with the battery substituted for the aluminum, was used to reduce the data. Figure 5 shows a typical determination for the battery specific heat. The inaccuracy below about 100 ºC is due to the delayed temperature response of the battery. The specific heat did not significantly vary over the range of testing this battery and was found to be 0.95 +/- 0.03 kJ/kg-K. This value was taken as constant for all analyses in the tests, in spite of mass lost from the battery during runaway.
Fig. 5. Typical determination of the battery specific heat before runaway.
2.5 Governing equation for calorimeter In these tests, runaway was induced by heating of the outer battery. The heating power was varied in the tests as well as the SOC of the battery. The heating likely melted of the plastic membrane separator and caused an internal electrical discharge and runaway. This electrical and chemical energy is manifested within the battery as a generation term , and outside of the battery by an ejection term of enthalpy from the battery. The conservation of mass for the system CV yields that the rate of mass ejected is equal to the rate of change for the battery mass, m. As shown in [15], the conservation of energy states that the rate of internal energy plus the rate of ejected enthalpy is
equal to the sum of the heating power (P) and the generation term ( ceramic wool (
) minus the heat loss into the
) given by Eq. (2). Rearranging to solve for the internal generation term gives (3)
The values for the constants in Eq. (3) are: (mc)Cu = 8.4 J/K, cb = 0.95 J/g-K, and cg is assumed as 1.05 J/g-K. The value of the specific heat for the ejected materials is selected for expedience as their components are not known and can comprise gases of hydrogen, carbon dioxide, methane and others, as well as possibly molten aluminum. At 1000 K, the specific heats of hydrogen, carbon dioxide and methane are respectively, 15, 1.2 and 4.5 J/g-K. Therefore the uncertainty of the specific heat for the ejected mixture as well as ignoring any possible melting leads to an underestimation for this ejected enthalpy rate term given as .
(4)
This term will be relatively small compared to the exothermic battery energy.
It will be shown that the battery undergoes very rapid changes during runaway. As a consequence the computation of time derivatives from the temperature and mass measurements can be problematic. Moreover the practical manner to assess the hazard of the battery in runaway is not to have the precise rate of energy released, but to have the total energy, realizing this is a potential transfer to the surroundings in a short time. Therefore, Eq. (3) is integrated to obtain the energy released over time. (5) with the ejected energy as Qg ≡ cg
∫
m mi
T dm . Equation (5) will be used to reduce the data for a series
of tests on the battery varying SOC and P. In all the tests the heating power was initiated at the start and kept constant throughout the test. An electric arc was manually initiated at the start of venting to attempt to ignite the gases. This was not always successful due to timing and location, and at times auto ignition could occur before the arc was struck. 2.6 Typical data and analyses Test 17 using a heating power of 25.6 W and a SOC of 100 % is examined to show the typical results for this battery. The raw data of battery temperature and mass are shown in Figure 6. There is a nearly linear rise in temperature until about 500 s when the pressure safety vent opens and releases white vapors. In general these vapors could be ignited but ignition was not achieved in this test at the first vent. These vapors continue to vent until about 550 s where thermal runaway occurs. The feedback between battery temperature and decomposition takes place over several seconds. This is seen as a nearly vertical jump in temperature and drop in battery mass. Runaway is a violent
Fig. 6. Battery temperature and mass at 100 % SOC and heating power at 25.6 W.
process with a high velocity jet from the pressure safety end of the battery. This ejection is about 35 % of the original battery mass. In several seconds the battery temperature jumped from about 200 ºC to 800 ºC. Runaway is a dramatic event. The runaway event is examined more closely over the jump period (551.5 to 554.5 s). Figure 7 shows the temperature jump and Figure 8 shows the rate of internal energy computed, . The dramatic jump in 3 seconds shows the characteristic of thermal runaway in a battery. While the rate of internal energy generation can be computed, its distribution over time may not be as significant as the total energy dumped in this process. However ejected energy from the battery commenced at the first vent (~ 500 s) at about 190 ºC; then most of this category of energy was released during the jump after the second vent (~ 550 s). Consequently most of the battery energy converted from its original electrochemical energy is primarily transferred in this jump period, and will be represented by Qb and Qg over the entire heating period of the battery as given in Eqns. (5) and (6).
Fig. 7. Temperature at jump.
Fig. 8. Internal generation rate of energy.
Figure 9 shows the specific energy terms in Eq. (5) individually computed for Test 17. As Qb is computed from the algebraic sum of the RHS terms some inaccuracy can be noted as its value is about –1.5 kJ before the jump instead of zero. However after the jump this internal energy computed is 30 kJ. Qg reaches about 18 kJ. Therefore the total energy of the battery released during runaway is about 48 kJ, compared to its rated energy at 100 % SOC of 29.3 kJ. Of course this does not include the combustion energy that can be released if the vented gases are ignited.
Fig. 9. Battery energy terms over time.
3. CALORIMETER DECOMPOSITION RESULTS About 25 tests were made with the same battery to assess the effects of the heating power the SOC. It might be expected that energy in runaway would be related to the electrical energy stored, and that the heating power should affect the time to runaway. The results are shown in the remainder of this section and the trend lines for the data are based on a linear or smooth fit with judgement to indicate a linear invariance. 3.1 Effect of Heating Power In this series the battery SOC was maintained at 80 % and the heating power varied from about 10 to 74 W for the duration of each test. The heating power reduces the time for runaway, as the safety vent operation and for the second venting at thermal runaway both decrease with heating (Figure 10). In the configuration of the battery in the calorimeter, all heating powers were able to initiate runaway. The duration of the runaway event varied from 3 to 12 s without a definite trend with the heating power. Although the times for venting and runaway varied significantly, the associated temperatures only varied slightly. Figure 11 shows these slight trends where the trigger temperature for runaway is about 220 to 300 ºC. During runaway the temperatures rise to over 750 ºC in 3 to 12 s. This rapid rise is indicative of the battery hazard in thermal runaway. Figure 12 shows the mass loss fractions of the battery after the first safety vent and the runaway vent occur. The mass loss fractions are not significantly dependent on the heat power. About 10 % of the original mass is ejected during the
safety venting and up to 40 % is lost after runaway. This indicates that more hot stuff is ejected following runaway, and much of this is combustible.
Fig. 10. Vent times as a function of heating.
Fig. 11. Temperature at onset and after runaway.
Fig. 12. Battery mass lost at venting. Fig. 13. Battery runaway energies vs heating rate. Figure 13 shows a slight effect of heating power on the internal energy and on the ejected energy. The former is about 30 to 35 kJ for this 80 % SOC and the latter is about 10 to 18 kJ. Hence heating power significantly effects the time for venting and runaway, but not significantly the vent and runaway temperatures, and mass ejected and energy produced.
3.2 Effect of State of Charge The effect of SOC is now examined. Here the heating power is essentially constant for all the tests at about 25 +/- 2 W. Figure 14 shows the times to safety vent (1st) and runaway (2nd) are nearly invariant with SOC. Liu [9] has also seen this invariance. The first vent time is likely invariant with the SOC because it occurs due to a designed pressure relief for the battery, and due to the constant internal volume of the battery the pressure is directly related to temperature. Once this temperature occurs it likely kicks off the kinetics of the runaway exothermic reaction leading to full
development of the reaction by the second vent time. The time interval between the two vent events is roughly about 20 % of the first vent time. The temperatures at of the 1st vent and the onset runaway are also invariant with SOC as the relief pressure controls the 1st, and the second might be considered an ignition temperature or the onset of the full exothermic reaction. However, the final temperature of the battery increases with the SOC (Figure 15). This is an indication of the initial electrochemical energy of the battery corresponding to its SOC. The 1st vent temperature is indicative of the designed pressure relief vent configuration, and is due solely to the heating power. The 2nd vent temperature is indicative of the onset of thermal runaway that depends on the battery and its heat loss environment. The temperature after runaway depends on the internal energy remaining in the battery and its ejected energy.
Fig. 14. Vent times vs SOC.
Fig. 15. Temperature vs SOC.
Fig. 16. Battery mass vs SOC.
Fig. 17. Battery energies vs SOC.
Fig. 18. Runaway time period vs SOC.
Fig. 19. Average runaway power vs runaway period.
Figures 16 and 17 show that, at runaway, the mass ejected and the energies released (internal and ejected) are directly related to the SOC. If runaway is perceived as an internal short circuit, then a portion of the original battery energy would be represented as resistance heating. As the remaining portion of the energy is due to decomposition dependent on temperature then it too will increase with SOC as Figure 15 shows the temperature after runaway increases with SOC. As the mass ejected increases after runaway, so does the combustion potential hazard of these gases. Here the “runaway period” is defined as the time increment from just after the onset of runaway to just before the end of runaway. The onset of runaway is indicated by the time of the 2nd vent. The end of runaway is indicated by the abrupt decrease in battery temperature. The runaway period decreases dramatically with SOC – from nearly 100 s at 20 % to as low as 0.5 s at 80% at which the runaway temperature nearly reached 1400 ºC (Figs. 15 and 18). The average runaway power in this period can be computed from the total energy released. Figure 19 shows how this varies with the runaway period. Nearly 100 kW can be released in ½ second! Figure 20 shows a comparison between the current results and a similar battery calorimeter developed by Liu [9]. He only computed the internal energy Qb, not the additional vented energy, Qg. Figure 20 shows good agreement with his results for Qb and the present.
Fig. 20. Comparison of runaway battery energy by battery calorimeter measurement 4. COMBUSTION ENERGY BY CONE CALORIMETER The Cone Calorimeter is a fire based analytical instrument designed to measure the combustion energy of the burned fire product gases [11]. In reference [10] its sample tray was fabricated to ensure a reduction of the jet momentum of ejected battery matter, and avoid propulsion of the battery out of the Cone test space. The electric arc ignition source was removed and the radiant heater intitiated runaway at a given heat flux. The ejected gases rose through the cone heater and were auto-ignited. The standard method for the oxygen depletion of the burning gases was used to measure the rate of heat release. Sean Crowley conducted Cone combustion experiments for a number of batteries [10]. The energy release period was very short in time, and the normal data reduction process of the Cone Calorimeter was not fast enough to compute the rate of energy released by combustion. Consequently a de-convolution algorithm developed by Richard Lyon was used to obtain an accurate rate from the time lag of the measuring process [10]. An example is shown in Figure 21.
Fig. 21. Panasonic LIB tested at 50 kW/m2 external heat flux showing raw and de-convoluted data in Cone Calorimeter (taken from [10].
The SOC charge had the most influence on the combustion energy. Figures 22 and 23. Here only mass loss and rate of combustion energy released were recorded as a function of heat flux to the battery (10 to 75 kW/m2), and as a function of the state of charge (SOC) for one battery at a heat flux of 50 kW/m2. Figure 22a shows the total combustion energy and mass lost for each battery test over a range of SOC. Figure 22b shows the effective heat of combustion of the vented battery gases. Both the Energy in Combustion and the effective heat of combustion decrease with SOC. It is not clear why they decrease. As there is some small amount of oxygen released in the breakdown of the cathode lithium oxide, it is possible that some of the combustion energy is from within the battery, not by combustion with ambient air. The cone will not record the internal combustion; moreover the high velocity jet associated with the combustion could affect the capture of the exhasut gases by the cone calorimeter. Other studies have also found this decrease with SOC for other Li-ion batteries [9,10,16]. Overall effective heat of combustion ranged from about 2 to 26 kJ/g-lost for 9 different batteries [10]. With the SOC at 90%, the same battery exhibited limited effect of heat flux above a threshold of 20 kW/m2 as shown in Figure 23. 120
20
Combustion Data from Cone
LG Heat Flux 50 kW/m
100
Energy/Mass lost, Eff. HoC kJ/g
Energy in Combustion kJ and Mass Lost g
Combustion Data from Cone 2
80
60 Tot HR kJ 40 Mass loss g 20
LG Heat Flux 50 kW/m
2
15
10
5
0
0 0
20
40
60
80
100
0
20
40
60
80
100
SOC %
SOC %
(a)
(b) 2
Fig. 22. LG Li-ion battery at a heating rate of 50 kW/m : Energy of combustion, mass lost, and effective HOC [10] 12
Energy/Mass lost, Eff. HoC kJ/g
100
80
60
Energy kJ
No 2nd vent
Energy in Combustion kJ and Mass Lost g
Combustion Data from Cone LG SOC 90%
40
Mass Loss g
20
0
Combustion Data from Cone LG SOC 90% 10
8
6
No 2nd vent
120
4
2
0 0
10
20
30
40
50 2
HF kW/m
(a)
60
70
80
0
10
20
30
40
50 2
60
70
80
HF kW/m
(b)
Fig. 23. LG Li-ion battery combustion energy, mass loss, and effective heat of combustion as a function of heating rate [10]
5. ALTERNATIVE METHODS FOR MEASURING EXOTHERMIC AND COMBUSTION ENERGY 5.1 Exothermic Decomposition Energy An alternative technique to measure the overall energy released by the runaway energy was developed by Lyon and Walters [12,13]. They modified an Oxygen Bomb apparatus using instead an atmosphere of nitrogen. The Oxygen Bomb with correct calibration is an accurate standard method for measuring the heat of combustion. Therefore, its use with nitrogen surrounding a battery in thermal runaway will measure all of its energy released. A comparison of their results with the total energy (Qb + Qg) measured by the battery calorimeter for nearly the same battery is shown in Figure 24. The battery tested by Lyon and Walters used a newer version of the Tenergy 18650 with a slightly higher capacity of 2600 mA-hr over 2200. Recall that a weakness of the current battery calorimeter method is the uncertainty of the specific heat of the material ejected, and there is no consideration in the analysis for melting. The “Nitrogen Bomb” of Lyon and Walters is believed to be more accurate in capturing the total energy released, especially above 80 % SOC where the calorimeter energy is lower. The variation in the results for a given SOC could be attributed to the slight differences in the batteries or simply the chaotic nature of runaway. The total runaway energy is approximately two times higher than the corresponding rated energy at each SOC. It is not uncommon for batteries in runaway to release several times their original electrical energy capacities.
Fig. 24. Current method with nitrogen bomb [12,13].
5.2 Governing Equations for the Nitrogen Bomb Calorimeter Equations are formulated here to define the procedure for the Nitrogen Bomb to measure the runaway energy with attention to calibration. This is not been completely done before. The conservation of energy is applied to a closed system comprised of the Oxygen Bomb apparatus containing a battery with a NiChrome wire heating assembly. The system includes the bomb, the water bath, and the insulated walls of the outer container. The water bath experiences normal ambient pressure while the pressure varies in the sealed bomb. A schematic is shown in Figure 25.
Fig. 25. Schematic of Nitrogen Bomb system
The conservation of energy is applied to the entire closed system in which the final internal energy is equal to the initial plus that from the heater. As a closed system, the change in internal energy of all the components is equal to the intentional heat added by an electrical heater around the battery, Q h.
U f = U i + Qh
(6)
Introduce enthalpy H = U + pV,
( )
∆H − ∆ pV = Qh
(7)
This enthapy change can be broken down into the various system components. The pressure work term is present for the gases enclosed in the bomb. ∆H = ∆H b,battery+s,structure+w,water + ∆H g ,batterygaes+nitrogen and ∆ pV = ∆ pV g (8)
( )
( )
Now let us represent the gases generated by the battery in runaway to globally include the chemical term representing the total energy in decomposition and electrical heating in runaway. Its enthapy change per unit mass can be expressed as a chemical/electrical term of runaway and its specific heat component. This is an approximent representation for the energy of runaway.
(
∆hg = −∆hd + cg T − Ti
)
Specifically representing all the terms, cb(mb,fTf – mb,iTi) is the change in energy of the solid battery components,
(Cs + mwcw)(Tf – Ti) is the change in energy of the “bomb” structure and the water, cgmN2(Tf – Ti) is the change in enthalpy of the nitrogen added to the “bomb”, mg∆hg is the change in enthalpy association with the electro chemical reaction, and
(9)
(pg,f – pg,i)V is the pressure work term. Then the energy equation becomes cb(mb,fTf – mb,iTi) + (Cs + mwcw)(Tf – Ti) + cgmN2(Tf – Ti) + mg[(-∆hd +cg(Tf – Ti) - (pg,f – pg,i)V = Qh
(10) The effective heat of decomposition (involving both exothermic chemical reactions and electrical dissipation) given off from the battery runaway is
(
(
Qd = mg ∆hd = Cs + mw cw + cg mN 2 + mg
(
)
)) (T
f
) (
− Ti + cb mb,, f T f − mb,iTi
)
− pg, f − p,g ,i V - Qh (11)
Lyon and Walters [13] did not include the heat capacity terms of the gas contained in the bomb free space. They considered the total energy released by the battery as the first four terms on the righthand-side. In the execution of the analysis it can be shown that the dominate terms are the first and the heat addition, Qh. To make this analysis work the Bomb must be calibrated. Standard calibration with oxygen to measure the heat of combustion normally will give the Cs + mwcw. Here the Bomb can be calibrated with a know heat input and a spent battery after a runaway test. Based on the conservation of mass
mb,i = mb, f + mg
.
(12)
Then the calibration involves measuring
(C + m c s
w w
)(
) (
)
(
+ cb mb,, f T f − Ti = pg , f − p,g ,i V + Qh − cg mN 2 T f − Ti
) (13)
From a standard calibration Cs is found. The mass and specific heat of water is known in the experiment. In the calibration with the battery cbmb,f can be found. Note this includes any solid or condensed material in the ejection. The Equation of State allows for the computation of the nitrogen and combustion gas mixture masses.
mN 2 =
and
piVM N 2 RTi
initially
(14)
mN + mg ≈ 2
p f VM N RT f
2
mN 2 + mg =
p f VM N 2 RT f finally. (15)
Lyon and Walters [12,13] executed such an analysis by calibrating with a cyliinder of aluminum. An example of their results is shown in Figure 24. 5.3 Combustion Energy Here the technique developed by Maloney [14] is adapted to estimate the combustion energy. Following the execution of the procedure to measure the decomposition runaway energy, the gas concentrations contained in the bomb can be measured. Maloney conducted battery runaway experiments in a nitrogen closed system and measured the composition of the combustion gases. These gases included hydrogen, carbon dioxide, carbon monoxide, and several hydrocarbons. He presents the partial volumes of these compositions at 10 psia, 27 ºC and for various SOC levels. To illustrate the technique, his results for a Tenergy LiCoO2 18650 (2600 mA-hr) battery at 100 % SOC is considered. His results are tablulated in Table 1 along with properties needed to compute the total available combustion energy. He did not compute the combustion, so it is now done for one of his experiments to investigate its feasibility and accuracy against combustion results from the Cone Calorimeter. The mass of each component gas can be computed from the Equation of State using the partial volumes at 10 psia and 27 ºC. In general the equation becomes
mi (g) = 0.02764Vi (L)M i (g/mol) .
(16)
The theoretical combustion energy from each component is determined by their complete heat of combustion. Table 1 shows two total values based on assuming the properties of the THC, and assuming the THC are composed of only methane and ethylene. This gives, respectively, 63.8 and 49.7 kJ. Results from the Cone Calorimeter at the same SOC (100%) for the 2200 mA-hr battery was found to be 48 and 55 kJ. It is felt that the gas composition method and Cone Calorimeter values are within reasonable aggreement with the expected level of accuracy and the slight variations in the batteries. Table 1. Vented gas composition Tenergy LiCoO2 18650 (2600 mA-hr) battery at 100 % SOC [14] Gas
Partial Volume*, L
Molecular weight, g/mol
Mass, g
Heat of combustion, kJ/g
Combustion energy, kJ
Hydrogen, H2
1.5
2
0.0829
141.6
11.7
Carbon monoxide, CO
2.0
28
1.55
10.1
15.7
Total hydrocarbons, THC
0.7
~ 40
0.774
~ 47
36.4
Methane, CH4
0.35 (assumed)
16
0.155
55.5
8.59
Ethylene, C2H4
0.35 (assumed)
28
0.271
50.3
13.62
Total with THC
63.8
Total with CH4&C2H4
49.7
* partial volumes at 27 ºC and 10 psia 6. CONCLUSIONS Methods have been presented to measure the energy liberated by a Li ion battery in runaway due to is electro-exothermic decomposition and its potential combustion energy from ejected gases in runaway. The battery calorimeter developed here is capable for measuring the rate and total energy of decomposition for a battery in runaway. The Cone Calorimeter is also capable of measuring the rate of combustion of the gases ejected in runaway. Both forms of energy pose a potential hazard to the battery surroundings. It is shown that the runaway decomposition energy increases with the SOC, and combustion energy is also high. As the energy release occurs over a very short time, the hazard is more aptly described in terms of the total energy released. In this regard, more accurate thermodynamic measurement techniques are found to be preferred for their simplicity and accuracy. This follows the technique developed by Lyon and Walters using an Oxygen Bomb with nitrogen to measure the runaway decomposition energy, and subsequent measurement of the combustion gases to determine the combustion. This paper has shown the comparability of these thermodynamic techniques, and recommends them for further consideration. 7. ACKNOWLEDGEMENTS This work was supported by the FAA Technical Center, where the experimental support of Sean Crowley was essential, and discussions with Richard Walters, Richard Lyon and Thomas Maloney were invaluable. This paper has been expanded from a previous work [16], and partially supported by the Pipeline and Hazardous Materials Safety Administration, PHMSA # 693JK3193C000007. 8. REFERENCES [1] H. Webster, Flammability Assessment of Bulk-Packed, Rechargeable Lithium-Ion Cells in Transport Category Aircraft, Report No. DOT/FAA/AR-06/38, Federal Aviation Administration, 2006. [2] H. Webster, Fire Protection for the Shipment of Lithium Batteries in Aircraft Cargo Compartments, Report No. DOT/FAA/AR-10/31, Federal Aviation Administration, 2010.
[3] C. Mikolajczak, M. Kahn, K. White, R.T. Long, Lithium-ion Batteries Hazard and Use Assessment, Report No. 1100034.000/A0F0/0711/CM01, The Fire Protection Research Foundation, 2011. [4] D.H. Doughty, P.C. Butler, R.G. Jungst, E.P. Roth, Lithium Battery Thermal Models, J. Power Sources 110 (2002) 357-363. [5] R. Spotnitz, J. Franklin, Abuse Behavior of High-power, Lithium-ion Cells, J. Power Sources 113 (2003) 81-100. [6] Q. Wang, P. Ping, X. Zhao, G. Chu, J. Sun, C. Chen, Thermal runaway caused fire and explosion of lithium ion battery, Journal of Power Sources, 208 (2012) 210–224. [7] E.P. Roth, D.H. Doughty, Thermal Abuse of High-power 18650 Li-ion Cells, J. Power Sources, 128 (2004) 308-318. [8] J.G. Quintiere and S. Crowley, The Sixth Triennial International Fire & Cabn Safety Research Conference, FAA Technical Center, Philadelphia Marriott Downtown, Dec 2-5, 2013. [9] X. Liu, Comprehensive calorimetry and modeling of the thermally-induced failure of a lithium ion battery, PhD Thesis, Department of Fire Protection Engineering, University of Maryland, College Park MD, 2016. [10] J.G. Quintiere, S. Crowley, R.N. Walters, R.E. Lyon, D. Blake, Fire Hazards of Lithium Batteries, DOT/FAA/TC-TN15/17, Federal Aviation Admin. William J. Hughes Technical Center, Atlantic City Int. Airport, NJ 08405, February 2016. [11] ASTM E7354-04a, Standard Test Method for Heat and Visible Smoke Release Rates for Materials and Products Using an Oxygen Consumption Calorimeter, ASTM International, West Conshohocken, 2004. [12] R.N. Walters, R.E. Lyon, Measuring Energy Release of Lithium-ion Battery Failure using a Bomb Calorimeter, Report No. DOT/FAA/TC-TN15/40, Federal Aviation Administration, March 2016. [13] R.E. Lyon, R.N. Walters, Energy Release by Rechargeable Lithium-Ion Batteries in Thermal Runaway, Report No. DOT/FAA/TC-TN16/22, Federal Aviation Administration, April 2016. [14] T. Maloney, Lithium Battery Thermal Runaway Vent Gas Analysis Report No. DOT/FAA/TC-15/59, Federal Aviation Administration, November 2016. [15] J.G. Quintiere, Fundamentals of Fire Phenomena, John Wiley and Sons, Ltd., Chichester, 2006, pp 66-67. [16] J.G. Quintiere, Measuring the Energetics of a Lithium Ion Battery in Thermal Runaway, Proceedings of the Ninth International Seminar on Fire and Explosion Hazards (ISFEH9), Edited by Bradley D., Chaumeix N., Liu N. A., Molkov V., Snegirev A. and Tamanini F.Saint-Petersburg Polytechnic University Press, 2019.
AUTHOR DECLARATION I wish to confirm that there are no known conflicts of interest associated with this Publication, and there has been no significant financial support for this work that could have influenced its outcome. I understand that the Corresponding Author is the sole contact for the Editorial process and is responsible for submissions of revisions and final approval of proofs. I confirm that I have provided a current, correct email address. [email protected] Signed
August 2, 2019