The widespread adoption of lithium-ion batteries (LIBs) in portable electronics, electric vehicles (EVs), and large-scale energy storage systems is a testament to their superior energy density, long cycle life, and declining cost. As the global push for electrification intensifies, the demand for higher energy density LIBs continues to grow to alleviate range anxiety in EVs and to store intermittent renewable energy more effectively. However, this relentless pursuit of higher energy density often comes at the expense of safety. The risk of thermal runaway—a violent, self-accelerating chain of exothermic reactions within a cell—increases with energy density, posing significant fire and explosion hazards. These safety incidents not only endanger lives and property but also erode public confidence, acting as a major barrier to the further penetration of LIBs in critical applications.
Conventional battery management systems (BMS) primarily rely on monitoring surface temperature and terminal voltage to assess the state of health and safety of a lithium-ion battery pack. While effective for managing normal operation, this external monitoring approach has a critical flaw: significant latency. The multilayer, compact structure of a lithium-ion battery, combined with the poor thermal conductivity of its internal components (electrodes, separators), creates a substantial thermal gradient between the core and the surface. During the onset of thermal abuse, heat and gases generated internally are trapped, causing the internal temperature and pressure to rise dramatically before any significant change is detected on the surface. Studies have shown that the peak internal temperature of a lithium-ion battery during thermal runaway can exceed the surface temperature by over 500°C. Relying solely on surface signals means the BMS may trigger countermeasures too late, after the irreversible chain reactions have already gained unstoppable momentum.
Therefore, the quest for timely and accurate early warning of thermal runaway has shifted focus inward. Direct, in-situ monitoring of internal signals—such as temperature, pressure, strain, and gas evolution—offers a paradigm shift in safety management. These internal parameters are the most direct and rapid indicators of the complex physicochemical changes preceding catastrophic failure in a lithium-ion battery. By detecting anomalies at their source, within the cell, safety systems can be activated at a much earlier stage, potentially preventing propagation and minimizing damage. This article provides a comprehensive overview of the mechanisms leading to thermal runaway in a lithium-ion battery and critically examines the emerging techniques, particularly embedded sensor technologies, for monitoring the vital internal signals that herald this dangerous event.
Understanding Thermal Runaway: The Internal Sequence of Failure
Thermal runaway in a lithium-ion battery is not a single event but a cascading sequence of interdependent exothermic reactions, typically triggered by mechanical, electrical, or thermal abuse. The process can be conceptually divided into three progressive stages, each characterized by specific internal chemical reactions and corresponding changes in key internal signals.
Stage 1: Initial Exothermic Decomposition (80°C – 140°C)
The first line of defense inside a lithium-ion battery is the Solid Electrolyte Interphase (SEI) layer, a passivating film on the anode surface. As the internal temperature rises above approximately 80°C, this metastable layer begins to decompose. This decomposition is an exothermic process that releases gases like CO2, C2H4, and small amounts of O2. Once the SEI layer is compromised (fully decomposed by ~120-140°C), the lithiated graphite anode becomes directly exposed to the organic electrolyte. A vigorous reaction ensues between the anode and the electrolyte solvents (e.g., ethylene carbonate, dimethyl carbonate), generating substantial heat and producing flammable hydrocarbon gases such as CH4, C2H6, and C3H6. At this stage, the internal temperature and pressure of the lithium-ion battery begin to climb steadily, but the cell voltage may still appear normal.
Stage 2: Accelerated Reactions and Cell Breach (140°C – 250°C+)
The heat generated in Stage 1 pushes the lithium-ion battery into a more dangerous regime. Several critical failure points are reached in rapid succession:
- Separator Meltdown: The polyolefin-based separator, with a melting point between 135°C and 165°C, begins to shrink and melt. This leads to large-scale internal short circuits between the cathode and anode, releasing the stored electrical energy as a massive, instantaneous joule heat pulse. This event dramatically accelerates the temperature rise.
- Cathode Material Decomposition: Unstable cathode materials (e.g., layered LiNixCoyMnzO2 or NCA) start undergoing exothermic decomposition reactions, often releasing oxygen. For example, a common reaction can be simplified as:
$$ \text{LiMO}_2 \rightarrow \text{Li}_{1-x}\text{MO}_2 + x\text{Li}^+ + x e^- + \frac{x}{2}\text{O}_2 $$
This released O2 is a potent oxidant for subsequent reactions. - Electrolyte Decomposition and Combustion: At temperatures above 200°C, the electrolyte solvent vigorously decomposes. The oxygen released from the cathode fuels the combustion of the electrolyte and the lithiated anode, producing large amounts of heat, CO, CO2, and other gases. The internal pressure of the lithium-ion battery rises precipitously, often leading to cell swelling and the eventual rupture of a safety vent or cell casing.
- Binder Reactions: At even higher temperatures (~260°C+), polyvinylidene fluoride (PVDF) binders can react with the lithiated anode, releasing hydrogen gas (H2), which further contributes to pressure buildup and creates an explosive mixture.
The sequence of key exothermic reactions is summarized in the table below:
| Stage | Approx. Temperature Range | Key Internal Reaction | Primary Internal Signal Changes |
|---|---|---|---|
| 1. Initial | 80°C – 140°C | SEI decomposition; Anode-Electrolyte reaction | Temperature rise; Pressure increase (CO2, C2H4, light hydrocarbons). |
| 2. Accelerated | 140°C – 250°C+ | Separator melt; Cathode decomposition; Electrolyte combustion | Rapid temp./pressure spike; O2, CO, CO2 release; Voltage drop/intern. short. |
| 3. Runaway | > 500°C | Complete decomposition of components; Combustion of ejected gases | Extreme temperature; Catastrophic pressure release; Fire/explosion. |
Stage 3: Thermal Runaway and Catastrophe (>500°C)
In this final, uncontrollable stage, all remaining cell components participate in violent exothermic reactions. The internal temperature of the lithium-ion battery can skyrocket beyond 500°C. The accumulated high pressure leads to the violent ejection of hot gases and electrode materials, which can ignite upon contact with air, leading to fire or explosion. The cell voltage collapses to zero. The heat from this single failing lithium-ion battery cell can propagate to neighboring cells in a module, causing a cascading failure.
The overall heat generation rate ($\dot{Q}_{gen}$) during thermal runaway can be modeled as the sum of the heat from individual reactions, often following an Arrhenius dependence on temperature:
$$ \dot{Q}_{gen} = \sum_i A_i C_i^n \exp\left(-\frac{E_{a,i}}{RT}\right) $$
where $A_i$ is the pre-exponential factor, $C_i$ is the concentration of reactants, $n$ is the reaction order, $E_{a,i}$ is the activation energy, $R$ is the gas constant, and $T$ is the absolute temperature inside the lithium-ion battery. This equation highlights the exponential relationship between temperature and reaction rate, underscoring why early detection of internal temperature rise is critical.
Techniques for Internal Signal Monitoring
Given the clear limitations of external monitoring, researchers are developing methods to probe the internal state of a lithium-ion battery directly. These techniques can be broadly classified into non-invasive electrochemical methods and invasive embedded sensor approaches.
Electrochemical Impedance Spectroscopy (EIS) Analysis
EIS is a powerful, non-destructive technique that probes the internal state of a lithium-ion battery by applying a small sinusoidal current or voltage perturbation over a wide frequency range and measuring the cell’s impedance response. The impedance spectrum contains information about various kinetic and transport processes, including charge transfer at electrodes, Li+ diffusion, and ohmic resistances, all of which are temperature-dependent.
The core premise for safety monitoring is that the internal temperature of a lithium-ion battery strongly influences the impedance, particularly the phase shift at a fixed high frequency (e.g., >100 Hz). The relationship between a characteristic impedance parameter $Z_{char}$ (like phase or magnitude at a specific frequency) and internal temperature $T_{int}$ can be empirically calibrated:
$$ Z_{char} = f(T_{int}) $$
During operation, real-time EIS measurements can thus be used to inversely estimate $T_{int}$. Studies have demonstrated that EIS-based internal temperature estimation can detect the onset of thermal runaway much earlier than surface-mounted thermocouples. The main advantage is its non-invasive nature; it uses the existing current collectors as “sensing electrodes.”
However, significant challenges hinder its widespread application in real-time BMS. Standard EIS equipment is bulky and expensive. While simplified, single-frequency phase detection systems have been proposed, they require careful calibration for each lithium-ion battery chemistry and are susceptible to interference from cell aging, state of charge (SOC), and the dynamic load profiles typical in EV applications.
Embedded Sensor Monitoring: The Direct Approach
The most direct strategy for internal signal acquisition is the physical integration of micro-sensors into the lithium-ion battery cell during its assembly. This approach aims to place the sensor in intimate contact with the critical components (electrodes, separator) to measure parameters like temperature, pressure, strain, and gas composition in real-time. While promising, embedding sensors into the highly reactive and constrained environment of a lithium-ion battery presents formidable challenges:
- Chemical and Electrochemical Compatibility: Sensors must be inert to the corrosive electrolyte and withstand potentials up to ~4.5V vs. Li/Li+.
- Electrical Insulation: The sensor and its leads must be perfectly insulated to prevent internal short circuits.
- Minimal Impact on Cell Performance: The sensor should not impede ionic/electronic conduction, reduce active material, or cause mechanical stress that accelerates degradation.
- Robustness and Long-Term Stability: Sensors must survive long-term cycling, swelling, and the extreme conditions of thermal runaway.
- Manufacturing Compatibility: The integration process must be scalable and compatible with high-speed lithium-ion battery manufacturing.
1. Embedded Temperature Sensors
Internal temperature is the most sought-after signal. Three primary sensor types have been investigated for embedding within a lithium-ion battery.
a) Resistance Temperature Detectors (RTDs): These sensors, often made of platinum (Pt) or nickel (Ni) thin films, work on the principle of the predictable change in electrical resistance with temperature ($R(T) = R_0[1 + \alpha(T – T_0)]$). Researchers have fabricated RTDs on flexible polyimide substrates, which are then inserted between electrode layers or attached to current collectors. A key innovation is the use of 3D-printed polymer scaffolds or creating cavities in the electrode coating to house the sensor, minimizing performance impact. Studies show embedded RTDs can detect the rapid internal temperature rise during an internal short circuit 7-10 times faster than external sensors.
b) Micro-Thermocouples (TCs): TCs generate a voltage proportional to the temperature difference between a measurement junction and a reference junction (Seebeck effect). Their small size, wide temperature range (beyond 1000°C), and simplicity make them attractive. They have been threaded into cylindrical 18650 cells or laminated between layers in pouch cells to map radial or in-plane temperature gradients. However, their low output voltage requires sensitive instrumentation, and the embedding process (e.g., drilling holes in cans) can compromise cell integrity and safety.
c) Fiber Optic Sensors: This is a highly promising category due to the sensors’ inherent advantages: immunity to electromagnetic interference, chemical inertness, small size (diameter ~125-250 µm), and ability to measure multiple parameters (temperature, strain, refractive index) along a single fiber. Fiber Bragg Gratings (FBGs) are particularly useful. An FBG acts as a wavelength-specific mirror inscribed in the fiber core. The reflected Bragg wavelength ($\lambda_B$) shifts linearly with temperature ($\Delta T$) and strain ($\Delta \epsilon$):
$$ \Delta \lambda_B = \lambda_B (\alpha_\Lambda + \alpha_n)\Delta T + \lambda_B (1 – p_e)\Delta \epsilon $$
where $\alpha_\Lambda$ is the thermal expansion coefficient, $\alpha_n$ is the thermo-optic coefficient, and $p_e$ is the photo-elastic constant. By using specialized fibers or packaging, the strain effect can be decoupled or minimized, making the FBG a highly sensitive, multiplexable internal thermometer. Researchers have successfully embedded FBGs into the core of 18650 cells, the electrode stack of pouch cells, and even demonstrated their survival through a full thermal runaway event, providing continuous internal temperature and pressure data.
The table below compares the key embedded temperature sensing technologies for lithium-ion battery applications:
| Sensor Type | Principle | Key Advantages for LIBs | Major Challenges |
|---|---|---|---|
| Thin-Film RTD | Temperature-dependent resistance | Fast response; Can be fabricated on flexible substrates for minimal intrusion; Good linearity. | Requires lead wires; Risk of short circuit if insulation fails; Limited high-temp. range. |
| Micro-Thermocouple | Seebeck effect (voltage generation) | Very small junction size; Wide temperature range; Inexpensive. | Low signal voltage; Need for cold-junction compensation; Embedding often invasive. |
| Fiber Bragg Grating (FBG) | Temperature/strain-dependent reflection wavelength | EMI immune; Chemically inert; Multiplexable (many sensors on one fiber); Can survive extreme conditions. | Complex interrogation unit; Sensitivity to both T & strain; Fiber brittleness requires careful handling. |
2. Embedded Strain and Pressure Sensors
The volume change of active materials during cycling and the gas generation during abuse lead to mechanical strain and pressure buildup inside a lithium-ion battery. Monitoring these signals can provide early warning of gas generation (pressure) and insight into mechanical degradation and SOC (strain).
Strain Sensing: FBGs are also excellent embedded strain gauges. By fixing an FBG to an electrode or the inner wall of a pouch, the strain-induced wavelength shift can be monitored. This has been used to study the expansion/contraction of silicon-based anodes and to detect the onset of abnormal swelling. Hybrid sensors combining FBGs with Fabry-Perot cavities have been used to distinguish between temperature and strain at multiple internal locations simultaneously.
Pressure Sensing: Direct internal pressure measurement is highly valuable as gas generation is a core symptom of most failure modes. Miniaturized piezoresistive or capacitive pressure sensors have been embedded in custom cells. More elegantly, an “air-gap” Fabry-Perot interferometer (FPI) can be created at the tip of an optical fiber. The length of this air cavity changes with external pressure, modulating the reflected interference spectrum. Such an FPI, fabricated alongside an FBG on a single fiber, has been used to monitor both internal temperature and pressure in a commercial 18650 lithium-ion battery during thermal runaway, clearly capturing the pressure spikes associated with venting and runaway.
3. Embedded Gas Sensors
Gas evolution is arguably the earliest and most specific indicator of internal chemical degradation in a lithium-ion battery. Different failure modes produce characteristic gas signatures (e.g., CO2 and light hydrocarbons from SEI/anode breakdown; CO from electrolyte combustion; O2 from cathode decomposition; H2 from binder reaction). Therefore, an internal gas sensor could provide the earliest possible warning.
The concept involves embedding a miniaturized gas-sensitive element, such as a metal-oxide semiconductor (MOS) or an electrochemical cell, inside the cell’s headspace or within a permeable cavity. For instance, a micromachined SnO2-based sensor with a Pt catalyst layer on a polyimide film has been developed to detect H2 within a cell. The core challenge is immense: the sensor must operate reliably in a saturated electrolyte vapor environment, discriminate between multiple reducing gases, and not be poisoned over time. Furthermore, placing any additional active electronic component inside a lithium-ion battery raises significant safety and reliability concerns. As a result, embedded gas sensing remains a largely nascent area of research compared to temperature and pressure sensing.
Conclusion and Future Perspectives
The safety of lithium-ion batteries is paramount for their sustained dominance in the energy storage landscape. Moving beyond traditional surface monitoring to directly probe the internal state of a cell represents the most effective path toward achieving early and reliable thermal runaway warning. This review has outlined the catastrophic internal reaction cascade of thermal runaway and detailed the promising techniques, especially embedded sensors, designed to capture its earliest signals.
While significant progress has been made, particularly with embedded fiber optic sensors for temperature and pressure, the field is still transitioning from laboratory proof-of-concepts to industrially viable solutions. Key future research directions should focus on:
- Multi-Parameter, Data-Fusion Approaches: No single internal signal provides a complete picture. Future smart lithium-ion batteries will likely employ a suite of co-embedded sensors (e.g., an FBG for temperature/strain, an FPI for pressure, and perhaps a selective gas sensor). Advanced BMS algorithms using artificial intelligence and machine learning can then fuse this multi-modal internal data with external signals (voltage, current) to dramatically improve prediction accuracy and reduce false alarms.
- Advanced Materials and Integration Schemes: Developing new sensor materials and designs that are inherently compatible with the harsh internal environment of a lithium-ion battery is crucial. This includes work on ultra-thin, flexible, and hermetically sealed sensor packages. Integration methods that are minimally invasive and compatible with gigawatt-hour-scale manufacturing processes must be developed.
- Wireless and Passive Sensor Solutions: Eliminating physical wire connections that pierce the cell casing is a major goal to ensure long-term hermetic seal integrity. Research into passive sensors that can be read wirelessly (e.g., through RF, ultrasound, or optical means) is of great interest.
- Standardization of Testing and Validation: As these internal sensing technologies mature, standardized abuse tests and validation protocols are needed to objectively compare their performance (e.g., detection lead time, reliability, impact on cycle life) under identical conditions.
The ultimate goal is to create an “intelligent” lithium-ion battery with a self-aware nervous system—a distributed network of internal sensors that provides a real-time health diagnosis. This paradigm shift from reactive to predictive and preventative safety management will be foundational for unlocking the full potential of high-energy-density lithium-ion batteries in applications where safety is non-negotiable.