Introduction

Utilization of Li-ion batteries (LIBs) has witnessed a remarkable surge in recent years, driven by their applications in various sectors, including portable electronics, electric vehicles (EVs), and renewable energy storage systems. The LIB market is expanding at an annual rate of over 30%, projected to reach $400 billion with a market size of 4.7 TWh by 20301. Simultaneously, the global adoption of environmentally friendly policies and the industry’s shift towards electric mobility featuring extended driving ranges have spiked the demand for batteries with larger capacity and higher power. However, this surge in demand for high-energy LIBs has raised concerns regarding safety and potential risks. Even though the incidence rate of EV fires is 60 times lower than that of gas vehicles2, EV fires can still pose significant risks of thermal runway, causing uncontrolled temperature spikes and dangerous chemical leaks. To prioritize the driver’s safety, automobile manufacturers are investing heavily in recalls to mitigate potential battery defects in LIBs. GM has allocated $2 billion, while Hyundai Group has committed $900 million to tackle these concerns. Therefore, the rapid increase in LIB usage highlights the urgent need to address the significant safety challenges particularly posed by large-scale applications.

Although commercial LIBs are already equipped with external safety apparatuses, such as pressure-limited valves, cell-to-cell fire extinguishers, and high thermal insulating materials, they often face delays and failures in preventing fire accidents3. Furthermore, the adoption of the cell-to-pack or cell-to-chassis designs, which forgo modules in battery configuration, necessitates reinforced safety measures for individual cells. For improved safety management, an autonomous system is required to compensate for the large discrepancy between the internal and external temperatures during the initial battery fire. In this context, there are two primary strategies for implementing thermo-responsive characteristics4 in batteries: one involves modifying electrolytes5,6,7 or separators8,9,10 with thermo-responsive polymers, thereby suppressing ionic conduction within the battery. The other strategy employs positive thermal coefficient (PTC) materials to interrupt electronic conduction when heated. PTC materials can exhibit rapid thermal response through processes like ion doping/de-doping, making them a preferable option in terms of response rate compared to materials undergoing bulk phase changes. In this aspect, there have been active attempts4,11,12 to integrate PTC materials, including conductive particle/plastic composites13,14,15,16,17,18,19 and p-type conductive polymers20,21,22,23,24, into LIBs. Their integration has been achieved through direct mixing with the slurry21,25 or by creating an interlayer between the current collector and electrode13,14,15,16,17,20,26,27. PTC materials are designed to interrupt intensive current flow by transitioning from conductors to insulators in response to overheating caused by internal short circuits. Despite successful proof-of-concept studies with PTC materials since the early 2010s, their industrial adoption remained elusive due to concerns about their compatibility with standard battery manufacturing processes, potential adverse effects on the energy/power densities of LIBs, and the lack of extensive demonstrations with Ah-level batteries3,4,12. To effectively incorporate the PTC effect as an internal safety feature in commercial batteries, it is crucial for the materials and fabrication techniques to be fully compatible with high-volume production processes.

Here, we introduce a scalable approach to fabricating the safety reinforced layer (SRL), designed to provide LIBs with an immediate shutdown capability in the event of internal short circuits, thereby minimizing the risk of overheating and explosions. The side chains of conductive polymers within the SRL have been tailored to trigger the PTC transition around 100 °C, while also ensuring the polymer’s high solubility in a commercially viable solvent, toluene, as opposed to chloroform. We uncover an unexplored phenomenon that incorporating carbon additives into the SRL facilitates the doping/de-doping kinetics of the conductive polymer, maintaining the high conductivity of SRL under standard battery operation. Despite constituting <0.5% of the total cathode weight and thus preserving the battery’s power and energy densities, the SRL efficiently interrupts current flow when a short circuit occurs, or the temperature surpasses 100 °C by elevating the resistance fourfold. Utilizing a 500 L reactor for the mass production of the polymer and a roll-to-roll (R2R) coating system designed for 1700 mm-wide current collectors, we have achieved a production capacity of 5 km per day. Impact tests on 3.4-Ah pouch cells have statistically shown that the SRL decreases the rate of battery explosions by 53%.

Results and discussion

Thermal runaway prevention with SRL

The SRL is positioned between the aluminum (Al) current collector and the cathode within the electrochemical cell. Under normal battery operating conditions, the SRL exhibits low resistance and, therefore, has a negligible influence on the battery performance. However, during destructive events such as an internal short circuit or a sudden temperature surge in batteries, the conductivity of the SRL drops, resulting in an instantaneous cessation of current flow. This prompt response can interrupt internal short circuits, preventing further current passage through the cell. Thus, when incorporated into batteries, the SRL provides a shutdown mechanism, reducing the risk of overheating and battery explosions. We acknowledge that PTC materials can be directly incorporated into the composite electrode by adding them to the slurry21,25. However, this approach has the potential to alter the stability and physical properties of the slurry, presenting challenges for its widespread adoption in commercial manufacturing. Therefore, we opted to create a PTC interlayer to improve the practical feasibility of this method.

One of the conductive polymers, Polythiophene (PTh), possesses physical and chemical properties adequate for serving as an SRL (Fig. 1). In addition to polythiophene, there are various conductive polymers available, such as polyaniline and polypyrrole. However, we selected polythiophene due to its favorable chemical modification and polymerization processability, as well as its cost-effectiveness. When doped with anions, the oxidized PTh demonstrates electrical conductive28, yet its conductivity diminishes rapidly in response to external stimulation that induces de-doping. Applying a voltage below the reduction potential of PTh can increase its resistance through de-doping, establishing a shutdown mechanism in the event of an internal short circuit in LIBs24. In parallel, by introducing ethoxy or alkyl side chains at the third position of PTh, a temperature increase readily triggers de-doping, leading to an elevation in electrical resistance, known as the PTC effect. This occurs because the vibrational energy of the side chains intensifies at elevated temperatures, causing anions to detach from PTh. Thus, in an ideal scenario for the PTh-based SRL, the oxidized PTh maintains its p-doped state within the operational voltage/temperature range of LIBs and promptly undergoes de-doping in response to an internal short circuit or a sudden temperature surge.

Fig. 1: A schematic diagram of a thermal runaway prevention mechanism in practical batteries employing a PTh-based safety reinforced layer (SRL).
figure 1

The SRL, positioned between the cathode and aluminum current collector, maintains high conductivity during normal battery operation. In response to an internal short circuit, the PTh in the SRL undergoes de-doping, halting further current flow and averting overheating within the cell.

Full size image

Molecular tailoring of PTh

To optimize the PTC effect of SRL, we tailored the molecular structure of PTh through side-chain engineering. Shorter side chains reduce the intermolecular distance of PTh, resulting in improved electrical conductivity. However, this can potentially dilute the PTC effect due to the lower vibration energy of shorter side chains29,30. On the other hand, longer side chains can lower the transition temperature of the PTC effect, requiring careful tuning to avoid disruption of the conventional battery operation at elevated temperatures (approximately 50 °C). Table 1 presents PTh variations for SRL considered in this study (Supplementary Fig. 1). While P3DDT exhibits an onset temperature of the PTC effect too close to the conventional battery operation range, P3HT with shorter chains can exhibit a higher point (120 °C) than P3DDT. Aiming to enhance the solution processability of PTh and adjusting the onset point to around 100 °C, we introduced triethylene glycol groups to the P3DDT and P3HT copolymer, resulting in the final structure of poly(3-dodecylthiophene-co-3-hexylthiophene-co-3-triethylene glycol thiophene) (PDDHEO). This polymer was synthesized by oxidative copolymerization of 3-dodecylthiophene, 3-hexylthiophene, and 3-triethylene glycol thiophene, in a ratio of 49.75:49.75:0.5. Characterizations using NMR (Supplementary Fig. 2) and GPC (Mn: 40k, Mw: 136k, PDI: 3.4) confirmed its structure. GIXRD results indicated that the intermolecular distance of PDDHEO was 2.3 nm (Supplementary Fig. 3), corresponding to the average of P3HT and P3DDT homopolymers. Following our design strategy, PDDHEO exhibited an onset temperature of the PTC effect at 95 °C and was nearly nine times more soluble in our processing solvent, toluene, than P3HT. Nevertheless, PDDHEO still demonstrated a similar impedance to P3DDT and P3HT, indicating comparable electrical conductivities when doped with anions at 4 V (vs. Li/Li+).

Table 1 Polythiophene characteristic via side chain differences
Full size table

The physical properties of PTh become crucial when considering its application in a large-scale roll-to-roll coating process. Typically, PTh exhibits poor solubility, limiting the choice of suitable solvents. Although chloroform is commonly used at the laboratory scale15,20,24, its toxicity and significant vapor pressure present safety challenges in scaled manufacturing (Hazard class: 6, recommended exposure level (REL): 2 ppm, Vapor pressure: 200 mmHg). Consequently, we explored alternative, safer solvent options for the PDDHEO copolymer and identified toluene as a viable choice (Hazard class: 3, REL: 100 ppm, Vapor pressure: 22 mmHg). This enabled us to successfully demonstrate scaled SRL production, providing a contrast to previous lab-scale research using PTh (Supplementary Table 1).

Leveraging the high solubility of PDDHEO against toluene, we successfully optimized the SRL coating solution. At a concentration of 3 wt% (optimized concentration for micro gravure coating), P3HT and P3DDT were not fully soluble, with viscosities of 15 and 401 cps, respectively. In contrast, PDDHEO exhibited excellent fluidity with a viscosity of 4.5 cps at the same concentration, deemed suitable for the micro gravure process (Supplementary Fig. 4). Both solution concentration and micro gravure roll speed (rpm) play pivotal roles in determining the quality and thickness of the SRL during the large-scale coating process. The high solubility of PDDHEO allows a wide range of concentration changes without inducing precipitation, ensuring a homogenous and conformal SRL coating.

Role of carbon additives in SRL

Given that the ohmic resistance of the SRL can degrade battery power densities, achieving an ideal SRL involves minimizing the resistance while preserving the PTC effect. In this consideration, Super C, a conducting additive commonly used in electrode preparation, was incorporated into SRL. We systematically explored the changes in the electrochemical properties and the PTC effect of PTh depending on the content of Super C. Figure 2a–c illustrate noticeable morphological changes when adding Super C at 0 wt%, 20 wt%, and 40 wt% (SC0, SC20, and SC40, respectively) to the PDDHEO-based SRL. The initially smooth and compact SRL became rough and porous with the addition of Super C, although PTh remained uniformly distributed in the SRL (Supplementary Fig. 5). Importantly, the PTC effect persisted even after the incorporation of Super C (Fig. 2d). SC40 exhibited an onset point of 105 °C, and its response rate surpassed that of SC0 (SC0: 561 ohms s−1, SC40: 1135 ohms s−1). The faster temperature response of SC40 contributes to improved battery safety by reducing the time for exothermic reactions through immediate current interruption.

Fig. 2: Morphology and electrochemical characteristics of PDDHEO and Super C hybrid layer.
figure 2

a–c SEM images of top surfaces of SC0 (a), SC20 (b), and SC40 (c). The inset shows the corresponding cross-sectional image. d Resistance changes with temperature increase for SC0 and SC40. The dashed lines show open-circuit voltage (OCV) responses of each sample at the given temperature. e, f Resistance change with applied voltage increase from 3 to 4.5 V, simulating the charging process (e), and with voltage decrease from 4.5 to 3 V, mimicking the discharging process in LIBs (f). The blue line represents conventional charge/discharge curves of LiCoO2 (LCO). g, Cyclic voltammetry (CV) curves of SC0, SC20, and SC40 at a scan rate of 0.4 mV s−1. h, i CV curves for SC0 (h) and SC40 (i) as a function of scan rate, ranging from 0.15 to 0.8 mV s−1.

Full size image

It is important to note that doping/de-doping of PTh can be facilitated by adding Super C. Figure 2e, f depict the impedance response to the applied voltage (Supplementary Fig. 6). Upon charging, the resistance of SC40 dropped sharply with the increase in applied voltage, reaching near 3.1 V, while SC0 remained insulating up to 3.6 V. Similarly, upon discharging, the resistance of SC0 started increasing near 3.6 V, whereas that of SC40 remained almost consistent above 3.2 V. It is also worth noting that increase in resistance is more significant with Super C. This, in turn, ensures reliable current flow interruption in the event of an internal short circuit when the cell voltage drops sharply below the reduction potential of PTh31. We emphasize that the presence of Super C prevents interference of the SRL in battery cycling by ensuring complete doping of PTh within the LiCoO2 (LCO) operational voltage range between 3.7 and 4.5 V, as illustrated with the blue lines in Fig. 2e,f.

The cyclic voltammetry further supports our hypothesis of facilitated doping/de-doping by carbon additives. Figure 2g shows the CV curves of SC0, SC20, and SC40 at a sweep rate of 0.4 mV s−1. Typically, PTh exhibits two oxidation peaks; the PTh in the neutral state converts to a polaron state and to a bipolaron state upon oxidation with anion doping. PTh becomes electrically conductive from the polaron state32,33. The first oxidation peak of pure PTh, or SC0, at 3.74 V shifted down to 3.56 and 3.41 V by increasing the Super C content to 20 wt% and 40 wt% in the SRL, respectively, while the second oxidation peak remained consistent at 3.83 V. Additionally, the reversibility of the first oxidation improved by adding Super C, as distinguished from the gradual increase in the corresponding reduction peak current. Even at an exceptionally low sweep rate of 0.15 mV s−1, equivalent to 0.33 C in the case of galvanostatic cycling, PTh oxidation with anion doping merely occurs above 3.65 V and barely showed any corresponding reduction peak (Fig. 2h), in contrast to the corresponding CV curve of SC40 (Fig. 2i). Note that the redox peaks were nearly consistent for multiple CV cycles, indicating high electrochemical stability of the PTh-based SRL. (Supplementary Fig. 7) Collectively, the CV results further confirm that Super C not only accelerates p-doping but also ensures reversible de-doping of PTh, in addition to simply increasing the electronic conductivity of the SRL. We note that the chemical state and intermolecular distance of the PTh was invariant with the Super C content. (Supplementary Figs. 8, 9). Thus, we hypothesize that the increased conductivity and porosity after adding Super C would promote both electron and anion transfer/transport in the SRL. Although conductive carbon has been commonly mixed with p-type polymer when using the PTC effect, our observation that the carbon facilitates the doping/de-doping response upon voltage change is unprecedented. This phenomenon is critical for the SRL to ensure both high conductivity during normal battery operation and high resistivity during internal short circuits. Thus, our results indicate that combining conductive carbon and polythiophene can improve both the performance and safety of LIBs when designing SRL.

Roll-to-roll production of SRL-coated electrode

To validate the large-scale productivity of SRL to be use in commercial LIBs, the PDDHEO synthesis was scaled up to a 500 L reactor, allowing the production of up to 300 kg of the polymer solution (3 wt%), as shown in Supplementary Fig. 10. With the substantial volume of the polymer solution, we then demonstrated the SRL production through a roll-to-roll process. While various methods such as drop casting13, blade coating16,20,24, spin coating34, and deposition27 have been adopted to form SRL in previous lab-scale demonstrations, we employed roll-to-roll coating to validate the functionality of the mass-produced SRL (Supplementary Table 1). We emphasize the importance of avoiding the use of toxic solvent, such as chloroform, as essential practice to comply with legal and environmental regulations in large-scale manufacturing. In parallel, it is crucial to minimize the thickness of SRL while ensuring complete coverage and uniformity of the layer. To achieve this goal, we employed micro gravure coating, a technique adept at managing low-viscosity solutions, which enabled the formation of a 600 nm-thick SRL layer using a toluene solution.

The roll-to-roll coating system was capable of producing SRL on Al current collectors with a maximum width of 1700 mm at a maximum velocity of 50 m min−1, as illustrated in Fig. 3a. The coating system comprised an unwinder, micro gravure coater, dryer and rewinder. Although the micro gravure roll is typically set to be wider than the coating film to ensure complete coverage, we designed the micro gravure to be narrower than the current collector, leaving a 10 mm non-SRL-coated area on both edges of the Al current collector. This design makes it adaptable for conventional pouch-cell production. We note that the micro gravure roll, featuring helical patterns (Supplementary Fig. 11), tends to generate an uneven or asymmetric side-to-side distribution of the coating. Typically, this deviation can be minimized by making the length of the roll longer than the film area, but due to the need to leave a non-coating area, some distribution was inevitable. Therefore, we used a method to minimize the left and right thickness variation by adjusting the pressure of the guide roll, as shown in the side view in Fig. 3b. As depicted in Fig. 3c, the resulting SRL was uniform, and the coating lines were distinctly visible. We further evaluated the thickness across the produced SRL and confirmed that the thickness variation was <49 nm (Fig. 3d). We also verified that the mechanical strength of the Al foil was not affected by the SRL coating, indicating no influence on subsequent processes (Supplementary Fig. 12).

Fig. 3: Roll-to-roll demonstration of large-scale SRL production.
figure 3

a, b A schematic diagram of the SRL coating system (a) and detailed illustrations of the micro gravure coater (b). c Photographs of the actual roll-to-roll manufacturing line. d Thickness profiles to assess the side-to-side deviation of SRL formed on the current collector.

Full size image

In the actual coating process, the SRL-coated Al current collector is fed to a dry zone spanning 24 meters, maintained at a temperature of 140 °C, and moving at a speed of 15 m min−1. This speed is sufficient to produce 3.6 km of both side-coated films each day. Coating an Al foil with a 1-meter width with SRL allows for the production of nearly 60,000 3-Ah pouch cells (4.5 × 10−3 m2). Thus, our demonstration effectively showcases the practical viability of SRL.

Battery performances with SRL

We assessed how SRL influenced the electrochemical performance of the cathode. The LCO electrode was slurry-coated onto the Al current collector incorporating SRL. Notably, SRL was incorporated into the electrode layer after calendaring (Fig. 4a). As a result, the final thickness of the entire cathode (61 μm) closely matched the original thickness without SRL (61 μm), showcasing the negligible impact of SRL on the volumetric energy density of the practical cell. The areal loading of SRL is below 0.14 mg cm−2, representing <0.5% of the conventional electrode weight, further confirming its marginal impact on the gravimetric energy density of the battery.

Fig. 4: Electrochemical performances of SRL-incorporated electrodes.
figure 4

a Cross-sectional SEM image of SRL-incorporated LCO cathode. b Rate capabilities of LCO cathodes having different SRL conditions (bare, SC0, SC20, SC40), cycled in a voltage range between 3 and 4.5 V. The rectangle symbols represent the coulombic efficiencies of the given condition. c Charge/discharge profiles at 0.5 C in 1st, 30th, 60th, 90th, and 120th cycles of the LCO cathodes with different SRL conditions. d The corresponding energy density retentions of the electrodes in (c). e Rate capabilities of the bare and SRL-incorporated LCO electrodes in 3-Ah pouch cells. f Capacity retention of the bare and SRL-incorporated LCO electrodes in 3-Ah pouch cells at 1 C having recovery cycles at 0.5 C after every 50 cycles.

Full size image

We compared the rate capability of the SRL electrode with the bare electrode in a galvanostatic cycling test, increasing the C-rate from 0.1 to 2 C (Fig. 4b). Despite the elevated resistivity of the current collector with SRL (Supplementary Fig. 13), SC40 electrode exhibited an initial capacity of 179.5 mAh g−1, demonstrating that the capacity loss is below 1.2%. Even at 2 C, the capacity gap between the bare and SC40 electrodes remained under 4.5%, while the SC0 electrodes had a gap of 13%. This confirms that under typical battery operation conditions, SRL minimally degrades the capacity. The difference in energy density at 2 C was also less than 2.9% due to the marginal increase in overpotential (Supplementary Fig. 14). We note that the reversible capacity of the SC40 electrode exceeds that of SC20 by 3.7% at 2 C, as expected from the increased conductivity and extended voltage range of p-doped PTh. Further, the energy density retention of SC20 and SC40 electrodes after 140 cycles were 92% and 95%, respectively, showing marginal difference with the bare electrode (Fig. 4c,d).

We then assembled 3-Ah pouch cells using SRL electrodes to assess their electrochemical performance. These cells had dimensions of 50 × 90 mm2 and were constructed with 12 stacks of positive electrodes and 13 stacks of negative electrodes (The capacity deviation among cells was within 0.4%, Supplementary Fig. 15). The capacity difference between the bare and SRL cells still remained below 1.5% when cycled from 0.1 to 2.5 C. Additionally, the capacity retention of the SRL cell was 97% in the recovery cycle after 300 cycles, coinciding with that of the bare electrode. Overall, our findings confirm that the incorporation of SRL closely preserves the gravimetric/volumetric energy and power density of practical batteries.

Accelerating rate calorimetry test (ARC) of 1-Ah pouch cell

We conducted an ARC test using fully charged 1-Ah pouch cells (graphite|LCO) made of bare and SRL-coated Al foils, respectively (Supplementary Fig. 16). For the bare cell, the onset of thermal runaway (TR) began at 173 °C (marked as T2) after 1758 min. The cell ignited for approximately 10 min, reaching a peak temperature of 418 °C in 1769 min. In contrast, for the SRL-coated cell, the TR was delayed by about 300 min, which is attributed to SRL’s effect. Additionally, the maximum temperature (T3) for the SRL-coated cell was 354 °C, approximately 15% lower than that of the bare cell, suggesting the SRL layer would reduce the fire damage by reducing its scale.

Needle penetration tests with SRL

We conducted needle penetration tests by inserting a thermocouple (TC) inside the needle to measure the temperature surge within the battery. Using a 50 mAh mono cell, we aimed to accurately monitor the localized heat generated within the battery upon needle penetration without inducing explosive thermal runaway. By modifying the needle design reported by Huang et al.35, we measured the internal temperature change of both bare and SC40 cells upon penetration of TC-integrated needles. The needle speed was set to 0.02 mm s−1 to ensure the TC remained in the heating position during the internal short circuit, enabling accurate real-time heat measurement. The internal temperature of the bare cell surged to 92 °C, while the SC40 cells remained below 57 °C. The temperature rise rates for the bare and SC40 cells were 154 and 42 °C s−1, respectively (Supplementary Fig. 17a, b). These results confirm that the addition of SRL delays the temperature rise, reducing the maximum temperature under a short circuit by preventing further current flow.

Nail penetration tests on pouch cells

After confirming the suppressed exothermic reaction by SRL with the needle test on mono cells, we further performed nail penetration tests to ensure the effectiveness of SRL in stabilizing 3-Ah stack cells (Fig. 5a). For bare cells, five out of six cells ignited, whereas none of the six SRL-applied cells ignited. According to the voltage profiles of the bare cells (Fig. 5b), rapid voltage drops down to 0 V, indicating a short circuit, were observed within approximately three seconds, leading to ignition for most cells. In contrast, all the SRL-applied cells showed immediate voltage recovery after the penetration, suggesting that the SRL layer effectively suppressed the short circuit as soon as it occurred (Fig. 5c). Consistent with our impact test results, this experiment clearly confirmed that the SRL layer effectively improves the battery safety under the mechanical abuse condition even for the high-capacity cells.

Fig. 5: Safety tests using 3-Ah pouch cells.
figure 5

a A schematic illustration of a nail penetration test. b-c, Voltage and temperature profiles upon nail penetration and photographs of the resulting pouch cells after the test using (b) bare and (c) SRL electrodes. We conducted the test using six cells for each condition to assess reproducibility. d A schematic illustration of the impact test. e Statistical pie graphs of 3-Ah cells comprising bare and SRL electrodes. The red area represents ignited cell and the blue area represents unexploded cell (safe). f Snapshots from the impact test. The majority of bare cells (upper row) combusted after the impact, whereas the SRL cells (lower row) were predominantly stable under the same abuse condition. g Disassembled cell components (cathode, anode, separator) from unexploded cells after the impact test. The components from the bare cell were severely damaged and deformed, contrasting with those from the SRL cell. h Temperature evolution of the bare and SRL cells during the impact test. Even without explosion, the temperature surge in the bare cell (blue) occurred 1.7 times faster than in the SRL cells (orange). The potential of each cell dropped immediately when the impact test started (dashed line).

Full size image

Impact tests on pouch cells

Finally, we delved into how SRL enhances battery safety under mechanical abuse conditions using 3-Ah pouch cells. The impact test (displayed Fig. 5d) was applied to both the bare and SRL-based pouch cells. To ensure statistical robustness, we replicated the test on 19 cells. Our results reveal that 12 out of 19 bare cells exploded, while 17 out of 19 SRL cells remained intact (Fig. 5e). Figure 5f illustrates representative captures for each condition over time during the impact test. The clear contrast strongly underscores the crucial role of SRL in enhancing battery safety. To further enhance battery safety, we envision that improving the adhesion and thermal stability of the SRL layer would be viable for optimizing its stability under severe mechanical abuse.

Importantly, it is worth noting that previous study developing SRL rarely conducted thermal stability assessments, and none of the existing literature supported their findings through statistical verification (Supplementary Table 1). Hence, our demonstration holds significance in providing statistical support for the SRL effect in preventing thermal runaway of practical LIBs.

Figure 5h displays the in-situ temperature variations measured using a thermocouple (TC) attached to the external surface of the pouch cell. In the event of ignition, the temperature surged at an astonishing rate of up to 59.4 °C s−1 due to cascade exothermic reactions among the battery components. Even without explosion, the bare cell experienced localized heating because structural deformation from the impact induced intense current flow. Remarkably, the temperature increase in the unexploded bare cell was 1.7 times faster compared to the corresponding SRL cells. This result aligns with the distinct structural deformation and the contrasting temperature responses observed in the impact test, depending on SRL inclusion.

To obtain direct evidence of the current flow interruption by SRL, we collected and characterized the battery components after the impact test, which physically smashes the pouch cell and inevitably causes a short circuit. In addition to the difference in the physical damage shown in Fig. 5g, we further confirmed that there were noticeable differences in Cu contents on the cathode surface (Supplementary Fig. 18). In the bare cells, significant amounts of Cu migrated from the anode to the cathode, whereas no Cu was detected on the cathode side in the SRL cells. In general, Cu diffusion to the cathode is known to occur during over-discharge36. During the impact test, intensive short circuits lead to over-discharge, which is followed by Cu diffusion. Thus, the result clearly confirmed that SRL layer successfully prevented over-discharge and the subsequent Cu diffusion upon mechanical abuse, which we attribute to the rapid increase in resistance upon shorting through PTh de-doping.

The nail penetration results with 3-Ah pouch cells provide additional support for the underlying mechanism (Fig. 5b, c). After the sudden drop in the voltage upon nail penetration, the SRL cell rapidly recovered the original OCV, which contrast to the complete OCV drop to 0 V for the bare cells. This suggests that the SRL layer effectively blocks the overflow of the current, preventing further exothermic reactions in the shorted battery.

We attribute these contrasting results to the effective de-doping of PTh in the SRL upon temperature increase and voltage drop caused by internal shorting, providing a mechanism for an instantaneous shutdown of localized current flow. Consequently, SRL can effectively prevent internal overheating, ultimately reducing the probability of a battery explosion.

In summary, we showcased the scalability and effectiveness of SRL as an autonomous shutdown system for preventing fire accidents of LIBs. Through molecular engineering, we achieved large scale PTh processing in toluene, while ensuring its swift de-doping near 100 °C. Furthermore, we uncovered that carbon additives facilitated the doping/de-doping kinetics of PTh. This enabled the SRL to maintain high conductivity across the entire operation voltage range of LCO, thus minimizing any negative impact on the power and energy densities of LIBs. Our roll-to-roll manufacturing system had the capacity to produce 3.6 km of double-sided SRL coating on Al current collectors each day, sufficient for making nearly 60,000 3-Ah pouch cells. Despite its minimal impact on the capacity and rate capability of the pouch cell, the SRL effectively reduces the battery explosion rate by 53%, as demonstrated through impact tests. The technical advancements achieved at both materials and system levels bridge the gap between lab-scale demonstrations and large-scale manufacturing, providing guidance for achieving practical breakthroughs in commercial products.

Methods

Synthesis of 3-triethylene glycol thiophene

3 g of 3-Methoxythiophene (26.28 mmol, 1 eq) and 7.03 g of triethylene glycol monomethyl ether (39.42 mmol, 1.5 eq) were dissolved in 150 ml of toluene with 500 mg of p-toluenesulfonic acid (p-TsOH) (2.63 mmol, 0.1 eq) and refluxed under a nitrogen atmosphere. The methanol produced by transetherification was removed using a type 4 Å molecular sieve charged with a Soxhlet extractor. After being refluxed for 24 h, the reaction was cooled to room temperature, quenched with water, and then extracted with ethyl acetate. The resulting mixture was washed with brine and dried over magnesium sulfate (MgSO4). The solvent was removed using a rotary evaporator, and the residue was purified by column chromatography using an eluent of methylene chloride:hexanes (2:1), resulting in 3-triethylene glycol thiophene.

Synthesis of polythiophene

215 g of FeCl3 were dispersed in 700 mL of hexane and stirred vigorously in 30 min at 30 °C. Droplets of thiophene monomers (53.3 g of 3-dodecylthiophene, 41.5 g of 3-hexylthiophene, 3-triethylene glycol thiophene) were added to the solution (5 ml min−1). The mixture was stirred for 1 h at 150 rpm. The temperature was set as 10 °C and the solution was precipitated with 2.4 L of MeOH. The precipitate was re-dissolved in 2 L of THF for 5 h at 60 °C. The dissolved solution underwent re-precipitation through the addition of 2.5 L of MeOH to eliminate Fe salt. The Fe salt was washed with 1.5 L of MeOH for 3 times. After filtration, the polythiophene powder was dissolved in toluene with a 3 wt% concentration.

Preparation of polythiophene coating solution

The polythiophene polymer was dissolved in toluene at a concentration of 3 wt%. The conductive additive (Super C) was added according to the manufacturing ratio. The resulting solution underwent sonication for 1 h and was then mixed using a mechanical stirrer. The sonication and mixing procedure were repeated four times to obtain the final coating solution.

Roll-to-roll coating process

For microgravure coating, the pan was filled with the coating solution until it reached its full capacity, and a circulation pump was employed to ensure proper circulation. To control the coating thickness, the speed of the microgravure was adjusted, and the coating was applied at a rate of 15 meters per minute. The drying temperature was gradually increased from 60 to 140 °C. After the coating process, the polymer-coated aluminum foil underwent heat treatment at 110 °C for 10 h.

Characterization of the safety reinforced layer (SRL)

The crystal structures of all samples were evaluated through X-ray diffraction (XRD) analysis, using a Bruker D8 Discover instrument featuring a Cu–K alpha source at 40 kV and 40 mA. Morphology and elemental composition analyses of the polythiophene layer were carried out using a JEOL JSM-7800F Prime scanning electron microscope (SEM) equipped with an Oxford Instruments X-Max EDS detector, operating at 15 kV and a working distance of 8 mm. Nuclear magnetic resonance (NMR) experiments were performed on a Bruker 700 MHz NMR spectrometer. The molecular weight of the polymer was determined through Gel Permeation Chromatography (GPC). The GPC analysis utilized Waters equipment, specifically the e2695 pump in conjunction with the 2414 RI detector. Shodex columns (KF801, KF802, KF803, KF804) were employed for the analysis, arranged in a series of four to enhance separation efficiency. To examine the chemical state changes of the polythiophene film, X-ray photoelectron spectroscopy (XPS) measurements were conducted using a Thermo Fisher Scientific Inc. K-Alpha+ instrument with a monochromatic Al K-alpha X-ray source (spot size: 400 μm). The collected XPS signals were fitted with respect to the C 1s hydrocarbon energy level at 284.8 eV. Dynamic mechanical analysis (DMA) experiments were conducted with a TA DMA850. The storage modulus was determined under the following conditions: strain 0.05%, frequency 1 Hz, ramp rate 2 °C min−1, temperature range 0–50 °C. The tensile strength of electrodes with different SRL conditions was determined by a universal testing machine (UTM, Zwick Roell Z030). The samples were prepared in the form of 20 mm × 100 mm pieces and tested at a speed of 1 mm min−1.

Electrochemical impedance spectroscopy (EIS) measurement depending on temperature/voltage

The half-cell was prepared for the electrochemical characterization of SRL. When assembling the half-cells, the working electrode (600 nm SRL coating on Al foil), a PE separator (20 μm), and a lithium disc (300 μm) were sandwiched in a 2032 cell. Depending on the experiment, various types of electrolytes were used.

The experiment utilized an electrolyte composed of EC:DEC in the ratio of 5:5 for measuring temperature-dependent EIS. The electrolyte composed of EC:DMC:DEC (3:3:4) was used for measuring voltage-dependent EIS. After connecting to the potentiostat (Princeton Applied Research, PARASTAT-MC), a voltage of 4.5 V was applied for 2 min during the formation process. Impedance was measured in the range of 150 kHz to 1 Hz, and the Nyquist plot was used to determine the impedance of the SRL layer. The impedance measurement for the PTC test was conducted by increasing the temperature in 10 °C intervals from 25 °C. Once the set temperature was reached, the impedance was measured, and the temperature was increased again. The temperature rise rate was approximately 5 °C min−1.

For the voltage-dependent impedance measurement, the voltage range varied from 4.5 to 3 V and inversely from 3 to 4.5 V. The experiments were carried out in a 25 °C chamber.

Cyclic voltammetry (CV) measurement

The CV measurements of the polymer layer were carried out using a potentiostat. These measurements were conducted utilizing the coin half-cell containing an SRL layer. The half-cell was prepared for the electrochemical characterization of SRL. When assembling the half-cells, the working electrode (600 nm SRL coating on Al foil), a PE separator (20 μm), and a lithium disc (300 μm) were sandwiched in a 2032 cell. The electrolyte composition was EC:DMC:DEC = 3:3:4.

Electrode fabrication

Aluminum foil coated with polythiophene was used for the cathode current collector, and bare copper foil served as the anode current collector. The cathode slurry was prepared by mixing LCO cathode material, conductive additive (Super C), and PVDF in a ratio of 97.5:1:1.5 in 1-NMP solvent. The anode slurry was prepared by mixing graphite, a conductive additive, and PVDF in a ratio of 91:1:8 in NMP solvent. After coating, both the anode and cathode were roll-pressed and subjected to vacuum drying at 110 °C for 10 h.

Half-cell test

LCO electrodes with or without having SRL were used for the coin half-cell fabrication. The cathode was designed with a loading density of 10 mg cm−2 and a porosity of 3.5 g cc−1. Half cells were assembled by sandwiching the LCO electrode, a PE separator (20 μm), and a lithium disc (300 μm) to create a 2032 cell. A 1 M LiPF6 in EC:DMC:DEC (3:3:4) was used as the electrolyte. The half cells were tested at 25 °C using a PNE SC battery cycler from WONIK PNE Co., LTD. The rate capability was tested by gradually increasing the charge C-rate (1 C = 200 mAh g−1) from 0.5 to 2 C (4.5 V CC/CV charge to 0.05 C cut-off, 0.5 C CC discharge to 3 V cut-off) after three formation cycles at 0.1 C and three 0.3 C cycles in the voltage range. To assess capacity retention, the half-cells underwent one formation cycle at 0.1 C, followed by cycling at 0.5 C for 140 cycles at a 25 °C chamber (4.5 V CC/CV charge to 0.05 C cut-off, 0.5 C CC discharge to 3 V cut-off).

Pouch-cell demonstration

1-Ah pouch cells (30 mm × 42 mm) were fabricated using the bare and SRL-coated LCO electrode. The cathode was designed with a loading density of 24.28 mg cm−2 and a porosity of 3.99 g cc−1. The anode loading density was 13.88 mg cm−2 with a porosity of 1.75 g cc−1. Six sheets of cathode and seven sheets of anode were stacked to produce an 1100 mAh stack cell. The PE separator was sandwiched between the electrodes. A 1 M LiPF6 in EC:DMC:DEC (3:3:4) was used as the electrolyte (3 µL mA h−1).

3-Ah pouch cells (50 mm × 90 mm) were fabricated using the bare and SRL-coated LCO electrode. The cathode was designed with a loading density of 24.28 mg cm−2 and a porosity of 3.99 g cc−1. The anode loading density was 13.88 mg cm−2 with a porosity of 1.75 g cc−1. 12 sheets of cathode and 13 sheets of anode were stacked to produce a 3200 mAh stack cell. The PE separator was sandwiched between the electrodes. The volumetric energy density of the stack cells was 600 Wh L−1. A 1 M LiPF6 in EC:DMC:DEC (3:3:4) was used as the electrolyte (3 µL mAh−1). The rate capability was tested at a 25 °C chamber by gradually increasing the discharge C-rate (1 C = 200 mAh g−1) from 0.1 to 2.5 C (0.7 C CC/CV charge to 4.5 V 0.05 C cut-off, CC discharge to 3 V cut-off) after three formation cycles at 0.1 C in the voltage range. To assess capacity retention, the pouch cells underwent three formation cycles at 0.1 C, followed by cycling at 1 C for 300 cycles at 25 °C chamber while including 0.5 C discharge every 49th step (0.7 C CC/CV charge to 4.47 V 0.05 C cut-off, 1 C CC discharge to 3 V cut-off, 0.5 C CC discharge to 2.75 V cut-off).

Mono-cell preparation

Mono cells (30 mm × 42 mm) were fabricated using the bare and SRL-coated LCO electrode. The cathode was designed with a loading density of 24.28 mg cm−2 and a porosity of 3.99 g cc−1. The anode loading density was 13.88 mg cm−2 with a porosity of 1.75 g cc−1. The cathode and anode were stacked to produce a mono cell. The PE separator was sandwiched between the electrodes.

Accelerated rate calorimetry (ARC) test

ARC tests were conducted to investigate the thermal stability of 1-Ah pouch cells. Prior to the ARC tests, the pouch cells were charged up to 4.5 V at a rate of 0.1 C. The cells were equipped with thermocouples and secured with insulation Kapton tape before being placed in the ARC chamber. A standard Heat-Wait-Seek (HWS) test procedure was followed, starting at an initial temperature of 50 °C. The exothermic limit was set to 0.02 °C min−1 to monitor any self-heating of the cells. Once self-heating was detected, the ARC system tracked temperature changes until thermal runaway occurred. The terminal temperature was set to 300 °C to ensure proper cooling of the ARC chamber.

Safety test

The needle test of mono cells (50 mAh, SOC 100) was conducted by penetrating a TC-embedded needle. TC was installed and welded in an 18 G (O.D.: 1.27 mm) sized needle. The needle was moved into the mono cell at the speed of 0.02 mm s−1. Voltage and temperature were recorded simultaneously35.

The nail test on stacked cells (3500 mAh, SOC 100) was conducted by penetrating a stainless steel nail (3 mm diameter, 30° tip angle) into the center of the pouch cell at a speed of 150 mm s−1 as described in the previous study37.

The impact test of pouch cells (3500 mAh, SOC 100) was conducted by dropping a 9.1 kg iron ball from a height of 61 cm. A stainless-steel rod with a diameter of 2 cm was placed at the center of the pouch cell, and the ball was released to induce a short circuit. The voltage and surface temperature of the cell were measured, and the ignition status was observed and recorded on video. Following the impact test, X-ray fluorescence (XRF) measurements were conducted using a Bruker micro-XRF apparatus. The X-ray tube was set to 5 kV and 600 mA while employing a Rh target.