As a researcher focused on fire safety engineering, the proliferation of electrochemical energy storage, particularly lithium-ion battery technology, presents both a transformative opportunity and a significant safety challenge. The pursuit of carbon neutrality has accelerated the deployment of energy storage systems, where the energy storage cell serves as the fundamental unit. Its high energy density is a double-edged sword, enabling efficient storage while also harboring the potential for catastrophic failure known as thermal runaway (TR). This violent, self-sustaining exothermic reaction, triggered by electrical, mechanical, or thermal abuse, can lead to intense fires, toxic gas emissions, and explosive hazards, directly threatening infrastructure and life. The critical question for safety engineers like myself is how to effectively intervene once this dangerous process is initiated within an energy storage cell. Traditional suppressants like water, while effective at cooling, pose risks of electrical short circuits and increased toxic gas generation in large-scale installations. Gaseous agents often fail to cool the cell core sufficiently, leading to re-ignition. This investigation explores the potential of liquid nitrogen (LN2) as a superior suppression agent, leveraging its immense latent heat of vaporization and inerting properties to halt the thermal runaway cascade in a commercial lithium iron phosphate (LFP) energy storage cell.
The inherent hazard stems from the complex chemistry within an energy storage cell. Under thermal abuse, exothermic decomposition reactions sequentially occur. The solid-electrolyte interphase (SEI) layer on the anode breaks down first, followed by reactions between the anode and electrolyte, cathode decomposition, and finally electrolyte combustion. These reactions release significant heat and flammable, toxic gases (e.g., CO, HF, H2), causing internal pressure to rise until the safety vent ruptures, ejecting jet flames and particle-laden vapor. Once the cell’s self-heating rate surpasses its heat dissipation capability, thermal runaway becomes inevitable, with temperatures exceeding 500°C. Therefore, an ideal suppression agent must perform three functions rapidly: extinguish flames, inert the atmosphere to prevent re-ignition of ejected gases, and, most crucially, extract heat from the cell to quench internal chemical reactions. Liquid nitrogen, boiling at -196°C at atmospheric pressure, is uniquely suited for this role. Its phase change from liquid to gas absorbs approximately 199 kJ/kg (its latent heat, $$r_{LN2}$$), providing profound cooling. The resulting nitrogen gas dilutes oxygen, effectively inerting the local environment. This study systematically evaluates the efficacy of LN2 injection at different stages of TR development and with varying doses on a large-format LFP energy storage cell.
The experimental platform was designed to simulate a single energy storage cell within a confined enclosure, representative of a module in a stationary battery energy storage system (BESS). The cell under test was a prismatic LFP energy storage cell with a nominal capacity of 65 Ah and 100% state of charge (SOC). Thermal runaway was induced by a constant-power heater attached to one side of the cell. The cell was instrumented with multiple thermocouples on both the heated and non-heated surfaces to capture temperature gradients and define an average surface temperature, $$T_{avg}$$. Another thermocouple monitored the environment within the sealed metal enclosure. The LN2 delivery system consisted of a pressurized Dewar, a solenoid valve, and a discharge nozzle positioned directly above the cell’s safety vent. The key experimental variables were the injection timing and the total mass of LN2 delivered, $$m_{LN2}$$.
First, a baseline test with no suppression established the TR characteristics of this specific energy storage cell. The temperature profile confirmed the classic stages of TR: external heating, self-heating, thermal runaway, and decay. Critical temperature thresholds were identified to define LN2 injection timings for subsequent suppression tests. The table below summarizes the experimental matrix designed to study the effect of injection timing.
| Test Case | Injection Trigger (Non-heated Surface Max Temp) | LN2 Dose, $$m_{LN2}$$ (kg) | Injection Duration (s) | TR Phase Description |
|---|---|---|---|---|
| 1 | 90°C | 1.2 | 40 | Pre-venting (Safety valve intact) |
| 2 | 135°C | 1.2 | 40 | Early-stage TR (After venting, during self-heating) |
| 3 | 135°C | 6.7 | 150 | Early-stage TR |
| 4 | 320°C | 7.2 | 160 | Fully developed TR (Peak reaction rates) |
The results from the baseline test were striking. The average temperature, $$T_{avg}$$, of the energy storage cell soared to a maximum of 375.8°C. The temperature change rate, $$dT_{avg}/dt$$, peaked at an alarming 10.85 °C/s during the most violent period of TR, unequivocally demonstrating the formidable energy release and rapid propagation potential within a failing energy storage cell. The safety vent opened at approximately 90°C, followed by vigorous jetting of gas and particles, which, when ignited, produced a sustained turbulent flame.
In Test Case 1, injecting 1.2 kg of LN2 prior to venting was highly effective. The environment temperature plunged to -137°C, and $$T_{avg}$$ dropped from 78.4°C to 44.5°C. Post-injection, the temperature stabilized around 58°C without re-acceleration. The cooling interrupted the decomposition sequence, preventing the energy storage cell from ever reaching the self-heating threshold for TR. This underscores the paramount importance of early detection and intervention for preventing catastrophic failure in an energy storage cell.
Test Case 2, using the same 1.2 kg dose but during early-stage TR (post-venting), told a different story. The injection provided only a transient, minimal cooling effect on a few measurement points. The average temperature did not decrease significantly, and after LN2 flow stopped, the $$dT_{avg}/dt$$ quickly rose again, peaking at 3.03 °C/s. While the final maximum temperature (≈304°C) was lower than the baseline, TR was not suppressed; it was merely slightly attenuated. This indicates that once the internal exothermic reactions in the energy storage cell have gained sufficient momentum, a small dose of LN2 is inadequate. Its heat absorption is overwhelmed by the cell’s internal heat generation rate.
This led to Test Case 3, where the dose was increased to 6.7 kg at the same early-stage trigger. The outcome was successful suppression. $$T_{avg}$$ plummeted from 132.6°C to 4.7°C, and the environment reached -117.9°C. Crucially, after injection, the temperature recovered to only 66.7°C, and the $$dT_{avg}/dt$$ remained well below the TR threshold. The massive heat extraction quenched the ongoing reactions. Test Case 4 proved that even during fully developed, violent TR with active flaming, a sufficiently large dose (7.2 kg) could suppress the event. The flame was extinguished almost instantly due to oxygen dilution and thermal shock, and $$T_{avg}$$ was pulled down from 313.1°C to 75.6°C, with no subsequent re-ignition or temperature runaway.
To quantitatively analyze the effect of dose, a separate series held the injection timing constant (early-stage TR at 135°C) while varying $$m_{LN2}$$. The results are consolidated in the table below, which includes key derived metrics for cooling performance.
| Test Case | LN2 Dose, $$m_{LN2}$$ (kg) | Max Cell Temp After Injection, $$T_{max}$$ (°C) | Peak Rebound Temp, $$T_{re,peak}$$ (°C) | Temperature Drop, $$\Delta T_{m-r} = T_{max} – T_{re,peak}$$ (°C) | Suppression Outcome |
|---|---|---|---|---|---|
| 5 | 6.2 | 117.7 | 100.1 | 17.6 | Failed (TR continued) |
| 3/6 | 6.7 | 132.6 | 66.7 | 65.9 | Success |
| 7 | 8.0 | 132.4 | 28.7 | 103.7 | Success |
The metric $$\Delta T_{m-r}$$, the net temperature drop from the point of maximum temperature after LN2 impact to the subsequent peak rebound temperature, is a direct indicator of the cooling effectiveness on the energy storage cell. It clearly increases with dose, showing that more LN2 removes more residual heat, preventing a stronger rebound.
A more rigorous thermodynamic analysis can be performed by modeling the energy storage cell as a lumped thermal mass. The total theoretical heat accumulation in the cell from the start of LN2 injection to its maximum temperature is:
$$Q_{total} = c_b m_b (T_{max} – T_i)$$
where $$c_b$$ is the specific heat capacity of the cell (≈1.1 kJ/kg·K), $$m_b$$ is its mass (1.73 kg), and $$T_i$$ is the cell temperature at the start of injection (~135°C).
The theoretical maximum heat that the injected LN2 can absorb through vaporization is:
$$Q_{LN2, theory} = m_{LN2} \cdot r_{LN2}$$
where $$r_{LN2}$$ ≈ 199 kJ/kg.
The actual heat absorbed by the LN2 from the energy storage cell itself, responsible for its cooling, is:
$$Q_{b, LN2} = c_b m_b (T_{max} – T_{re,peak})$$
From these, we can define two key performance ratios. The Cooling Efficiency, $$\eta_c$$, is the fraction of the cell’s accumulated heat that was removed by the LN2:
$$\eta_c = \frac{Q_{b, LN2}}{Q_{total}}$$
The LN2 Utilization Efficiency, $$\eta_e$$, is the fraction of the LN2’s theoretical cooling capacity that was actually used to cool the cell:
$$\eta_e = \frac{Q_{b, LN2}}{Q_{LN2, theory}}$$
Applying these formulas to the data yields profound insights. As the dose increased from 6.2 kg to 8.0 kg, the heat extracted from the cell, $$Q_{b, LN2}$$, increased dramatically from 33.4 kJ to 197.3 kJ. Consequently, the Cooling Efficiency, $$\eta_c$$, rose sharply from 18% to 89%. This means that with 8 kg of LN2, nearly 90% of the heat that had built up in the energy storage cell post-injection was stripped away, explaining the successful suppression. However, the Utilization Efficiency, $$\eta_e$$, peaked at 16.2% for the 6.7 kg dose and decreased slightly for the 8 kg dose. This indicates diminishing returns; a significant portion of the LN2’s cooling capacity is always “wasted” on cooling the enclosure, the atmosphere, and possibly undergoing free expansion without contacting the cell surface. There is an optimal trade-off between achieving reliable suppression and agent economy.
The mechanism of suppression is multifaceted. Upon injection, LN2 first acts as a powerful physical coolant. The sudden local cooling on the casing of the energy storage cell can quench surface reactions and contract materials, potentially impeding internal processes. As it vaporizes, it generates a large volume of cold nitrogen gas. This gas inertes the headspace, starving any flames of oxygen and preventing ignition of ejected volatiles. Simultaneously, this cold gas envelops the cell, enhancing convective heat transfer. The most critical effect, however, is the conductive heat extraction through the cell casing. By rapidly lowering the external temperature, LN2 creates a steep thermal gradient that draws heat from the core of the energy storage cell. If this heat extraction rate exceeds the internal chemical heat generation rate, the temperature declines, and the self-accelerating reactions are halted. The experiments show that this threshold cooling power is not reached with a small dose during active TR but is achievable with a larger, sustained flow.
In conclusion, this investigation confirms that liquid nitrogen is a highly effective agent for suppressing thermal runaway in a large-format LFP energy storage cell. Its efficacy is supremely sensitive to both the timing of application and the delivered dose. Intervention before the safety vent opens (pre-venting) can prevent TR with a minimal dose, highlighting the critical role of early thermal fault detection systems in battery management. Once TR is initiated, successful suppression requires a sufficient mass of LN2 to overcome the cell’s intense internal heat generation rate. While higher doses provide greater cooling power and reliability, the utilization efficiency of the agent does not scale linearly, presenting an optimization challenge for system design. The derived thermodynamic model provides a quantitative framework for estimating required LN2 mass based on cell parameters and TR state. For practical implementation in protecting energy storage systems, these findings suggest that a targeted LN2 suppression system, triggered by early warning signals (e.g., off-gas detection or rapid temperature rise) and delivering a calculated dose directly to the initiating energy storage cell, could effectively contain a failure and prevent propagation within a module, thereby enhancing the inherent safety and viability of large-scale lithium-ion battery energy storage.