Correlative acoustic time-of-flight spectroscopy and X-ray imaging to investigate gas-induced delamination in lithium-ion pouch cells during thermal runaway

Highlights

• Cell defects and failure investigated with X-ray imaging and acoustic spectroscopy.

• Acoustic spectrograms provide indicators of gas-induced delamination.

• Correlative X-ray imaging complements high-speed pulse-echo ultrasonic probing.

• Cell defects are identifiable, including electrode non-uniformity and gas phases.

Abstract

It remains difficult to detect internal mechanical deformation and gas-induced degradation in lithium-ion batteries, especially outside specialized diagnostics laboratories. In this work, we demonstrate that electrochemical acoustic time-of-flight (EA-ToF) spectroscopy can be used as an insightful and field-deployable diagnostic/prognostic technique to sense the onset of failure. A 210 mAh commercial lithium-ion cell undergoing thermal abuse testing is probed with in situ and operando EA-ToF spectroscopy, together with simultaneous fractional thermal runaway calorimetry (FTRC) and synchrotron X-ray imaging. The combination of X-ray imaging and EA-ToF analysis provides new understanding into the through-plane mechanical deformation in lithium-ion batteries through direct visualisation and the acoustic ToF response. Internal structural changes, such as gas-induced delamination, are identified using EA-ToF spectroscopy due to variations in the attenuation and signal peak shifts. This is corroborated using X-ray imaging, demonstrating EA-ToF spectroscopy as a promising technique for detecting onset of battery failure.

Introduction

Lithium-ion batteries have been the underpinning power supply technology in modern consumer electronics and responsible for the growth of electric vehicles (EVs) in the transport sector. As a result of this, record forecasts for lithium-ion battery production have been chronicled in depth [1]. Moreover, growth predictions are continually revised as efforts to reduce climate impact are made via the development of renewable energy sources and the integration of EVs with a smart electrical grid network [[2], [3], [4]]. Despite their widespread application, concerns regarding the safe deployment of batteries across a range of challenging applications persist.

Lithium-ion batteries can generate significant amounts of heat associated with both electrochemical reactions and ohmic losses. When the rate of heat generation exceeds the rate of heat dissipation, the cell begins to increase in temperature and at a critical temperature, highly exothermic decomposition reactions occur. This process, referred to as thermal runaway, rapidly degrades the anode, cathode, electrolyte, separator and solid electrolyte interphase (SEI) [[5], [6], [7], [8], [9], [10]] and ultimately results in catastrophic failure of the cell. Therefore, understanding the mechanism and onset of thermal runaway is essential for the design of safer lithium-ion batteries. Feng et al. [11] utilised the ‘heat-wait-seek’ method with extended volume accelerated rate calorimetry to deconstruct the sequence of events during thermally induced failure at increasing temperature. This can be grouped into three significant temperature ranges: at ca. 80 ᵒC, the SEI starts to exothermically decompose. As the anode loses this protective layer, a loss of capacity occurs via de-intercalation of lithium ions from the anode and the anode reacts with the electrolyte producing heat. At ca. 130 ᵒC; the rate of temperature increase slows, and the separator begins to melt (the majority of polyolefin commercial separators consists of polyethylene or polypropylene, which have melting points of ca. 105 ᵒC – 180 ᵒC depending on the grade). As the separator melts, this causes micro ‘soft’ short circuits of the anode and cathode, thereby considerably increasing the rate of reaction, current flow and energy delivered to the regions of short-circuit. These local regions experience greater heating, causing further degradation of the separator. Additionally, less thermally stable cathode materials may begin to decompose at this stage as the temperature continues to increase. At ca. 240 ᵒC, widespread gas generation and decomposition of the cathode, electrolyte and polyvinylidene fluoride binder (PVDF) occurs. These events rapidly produce heat and gas until the build-up of pressure can cause rupture of the cell casing, which is itself a highly destructive and dangerous event [12]. During this sequence of events, it has been reported that up to 2 L of gas can be generated per amp hour of capacity [13,14].

In this work, thermal abuse fractional thermal runaway calorimetry (FTRC) is used to analyse the thermal evolutions as thermal runaway progresses with electrochemical acoustic time-of-flight (EA-ToF) probing and co-incident X-ray imaging for internal mechanical changes. FTRC has previously been used for cylindrical cell geometry calorimetry to quantify total energy released and to also discern the heat output of ejected and non-ejected material during thermal runaway [11,[15], [16], [17]]. FTRC data has also been used to support statistical assessments which describe the event-to-event variability in thermal runaway responses for a given cell format [15]. Understanding these processes is crucial in both the battery module design and in the development of improved battery management systems (BMS), to minimize excess parasitic mass while ensuring that there is adequate isolation of the characteristic ejection of heat from each cell. From a safety perspective, this second concern is vital; individual cell failure cannot be allowed to trigger cascading thermal runaway throughout an entire module. From a BMS perspective, understanding this process may prevent thermal runaway altogether, by electrically isolating the cell upon detection of signature EA-ToF markers indicative of gas-induced delamination, to prevent further degradation.

Abuse testing methods for lithium-ion batteries typically fall into three broad categories; thermal, mechanical and electrical abuse (external or internal short-circuiting or excessive charge or discharge). The majority of failure testing utilises the first two [17]. A disadvantage of these techniques is that they are unable to reliably simulate the worst-case scenarios, particularly when all the energy is ejected in a single direction [11,[17], [18], [19], [20]]; furthermore, it is difficult to determine exactly when the thermal runaway will initiate, especially in situations of thermal abuse. Other abuse methods, such as oven tests or accelerating rate calorimetry, have limitations in the time taken to achieve the critical temperature to initiate thermal runaway, compared to FTRC which was used in this work. These insufficiently fast failure tests may cause the electrolyte to evaporate due to initial venting, thus resulting in a dry electrode assembly, possibly yielding inaccurate thermal failure insights [11]. Surface temperatures in excess of 600 °C are common during failures, which can occur within a few seconds [16,17,19]. Therefore, in batteries, the ability to forecast and understand ruptures and failures will provide key insight into module safety and performance [17].

In-depth insight with direct non-destructive imaging during thermal runaway has been demonstrated by Finegan et al. [5,16,19] and Robinson et al. [21] with synchrotron X-ray and thermal imaging, respectively. Electrochemical and structural changes have also been investigated by several authors [[22], [23], [24], [25]] using techniques such as neutron imaging and electrochemical impedance spectroscopy (EIS). EIS has been used by Fernandez et al. [26] to quantify the degradation mechanisms (conductivity loss, loss of active material and loss of lithium inventory) within a BMS; however, limitations in accuracy exist which are dependent upon the measurements and models [27]. Moreover, EIS measurements are affected by environmental and operating conditions, including temperature, C-rate, state-of-charge (SoC), depth-of-discharge (DoD) and cabling configuration. Furthermore, models can require complex computation for a limited operating range of conditions.

EA-ToF spectroscopy is applicable when materials are not stressed in tension or compression beyond their elastic limit. In this work, an ultrasonic wave is used to apply a force into component materials within a pouch cell within their elastic limit, ensuring a repeatable and valid characteristic spectroscopy. The relationship between the speed of sound (c) of the ultrasonic wave used in EA-ToF spectroscopy, is dependent upon elastic modulus (Ε) and density (ρ), as defined by the Newton-Laplace equation (Equation (1)).

$$c=\sqrt{\frac{E}{\rho }}$$

The difference in time-of-flight (ToF) between interfaces is also dependent on the thickness of each layer, with a thicker layer producing a proportionately delayed ToF. Hsieh et al. [28] have previously discussed the importance of material density on acoustic wave propagation in electrode assembly materials. Robinson et al. [29] inferred changes in the ToF, and thereby the location of the electrode from the transducer as a result of changes in the acoustic signal at the same SoC. Further EA-ToF work conducted by Robinson et al. [30] characterised behaviours induced during cycling of lithium-ion pouch cells at a range of C-rates, such as potential stresses in the electrode. In addition to ToF shifts, another important change in EA-ToF spectroscopy is the attenuation in acoustic intensity. When the ultrasonic signal is transmitted through each layer of material, the wave is scattered, reflected and absorbed to different extents. The acoustic impedance of a system, ΔZ, as a function of the change in density, Δρ, and the elastic modulus, E, is given by:

$$\Delta Z=\sqrt{\left|\Delta \rho \right|{\rm E}}$$

Gas exhibits significantly higher attenuation than liquids or solids as it is very poor at transmitting ultrasonic signals. This is shown as a reduction in acoustic intensity or loss of reflected signal during EA-ToF analysis. Galushkin et al. [31,32] investigated the generation of gases within an electrode assembly during cycling including their contributions to thermal runaway, these gases and phases changes within an electrode assembly would be identifiable with acoustic spectroscopy. As the lack of propagation of the ultrasonic wave pulsed by the transducer and the change in density at the interfaces of different phases would exhibit high attenuation and delay in ToF.

Hsieh et al. [28] have demonstrated that EA-ToF has the capability to determine SoC and state-of-health (SoH) of pouch cells at low C-rates (0.2 C, 0.1 C, 0.05 C), given prior acoustic measurements and voltage correlation to deduce relative ToF shift. Models developed by Davies et al. [33] displayed robustness in predicting SoC to 1% accuracy for both intact and damaged cells. Additional models by Davies et al. [33] evaluated SoH by comparison of EA-ToF measurements at full charge after 20 cycles, also displayed similar accuracy. Spatially resolved acoustic ToF characterisation by Robinson et al. [29] has shown promise in probing the internal structure as a function of SoC. Other acoustic techniques using guided waves of an order of magnitude lower in frequency have probed the SoC and SoH of cells [34,35]. Ladpli et al. [35] used algorithms to decompose the acoustic waveforms into simpler constituents to analyse the acoustic-electrochemistry phenomena in order to predict SoC and SoH. Gold et al. [34] have also utilised ultrasonic waves at 200 kHz to resolve SoC determination over one cycle. The use of the lower frequency signals facilitated the characterisation of a graphite electrode by analysis of arrival times of the slower, compressional waves at different SoC. Previous work conducted by the authors [30] has demonstrated temporally resolved EA-ToF spectroscopy as a correlative proxy for determining the effects of high-rate cycling and identifying electrochemical stiffness effects in batteries. However, there remain uncertainties regarding the mechanisms which cause structural degradation; an indication of the onset of failure, which have not to date been characterised by EA-ToF analysis.

In this paper, electrode delamination and gas formation in lithium-ion batteries is investigated using coupled EA-ToF spectroscopy and X-ray imaging. These events are known causes of cell degradation and precursors to the onset of thermal runaway. Gas-induced delamination in a planar electrode assembly is prominently visible via EA-ToF spectroscopy due to the significant changes in the measured attenuation and shift in the waveform peak location. Laboratory and synchrotron X-ray imaging with X-ray CT are used to highlight the internal phenomena which occur during defect driven gas-induced delamination and thermal abuse FTRC and to provide correlative validation. The paper is split into two parts, the first demonstrates the application of laboratory X-ray imaging with EA-ToF spectroscopy to identify gas formation in a cell cycling under normal conditions; the second utilises EA-ToF spectroscopy with high-speed synchrotron imaging to observe and characterise thermal runaway.

This is the first time to the authors' knowledge where acoustic spectroscopy and X-ray imaging have been effectively combined to explore the performance and failure of Li-ion batteries. The techniques used in concert provide unique insight into the gas induced delamination and are subsequently used to explore the nucleation and propagation of failure. Understanding, and improving the safety of Li-ion batteries remains a critical challenge, and therefore we anticipate this work will have a significant impact across the battery community; providing new fundamental insight into the failure process, and a platform of new capabilities for future studies of thermal runaway behaviour.