In this study, I examined how the silicon content of silicon-based negative electrodes controls the thermal runaway behavior of lithium-ion power batteries for an electric vehicle battery pack. My work focused on silicon suboxide (SiOx) mass fractions of 0%, 5%, 10%, and 15% in 1 A·h soft-pack cells with an NCM811 positive electrode and a SiOx/graphite negative electrode. I used adiabatic accelerating rate calorimetry (ARC) and lateral heating to connect SEI stability, characteristic temperatures, voltage collapse, heat release, and flame jetting. The central problem is that silicon provides very high theoretical capacity, but it also undergoes large volume changes during lithiation and delithiation. These volume changes repeatedly fracture and reform the solid electrolyte interphase (SEI), consuming electrolyte, exposing fresh anode surface, and generating flammable gases. In an electric vehicle battery pack, that chain can transform a local fault into a pack-level thermal propagation event.
My experimental design was built around two complementary test paths. First, I used ARC to measure self-heating onset, runaway trigger, maximum temperature, and voltage loss under nearly adiabatic conditions. Second, I used lateral heating with a ceramic heater to create an external abuse condition similar to localized heating in an electric vehicle battery pack. I placed thermocouples on the cell surface, inside the jelly roll between the current collector and separator, and on the heater surface. I recorded temperature, voltage, and video simultaneously. I repeated each condition three to four times to check repeatability and to separate true silicon-content effects from cell-to-cell variation.
My main finding is that increasing SiOx content improves capacity potential but degrades thermal safety. As SiOx increases from 0% to 15%, the SEI thermal stability threshold T1 decreases by approximately 40.8%. The runaway trigger temperature T2 increases, meaning that once the cell reaches the critical condition, the subsequent temperature rise is more violent. The maximum runaway temperature T3 increases by approximately 30.3% when the SiOx mass fraction is 15%, reaching 836.2 °C under ARC. Under lateral heating, the 15% SiOx cell reaches 1 003.5 °C. The flame jetting becomes more intense and shorter, with a measured duration near 10 s. Together, these results show that the electric vehicle battery pack designer cannot treat silicon content as only an energy-density variable; it is also a thermal-runaway variable.
Experimental Matrix and Cell Design
I selected a 1 A·h soft-pack format because it allows controlled ARC and lateral heating measurements while retaining the main material chemistry of a power cell used in an electric vehicle battery pack. The positive electrode was NCM811, with a Ni:Co:Mn molar ratio of 8:1:1 and a D50 particle size between 10 μm and 20 μm. The negative electrode was a SiOx/graphite composite. The SiOx mass fraction was varied while the graphite balance changed accordingly. The nominal voltage was 3.55 V for silicon-containing cells and 3.7 V for the graphite-only reference. The charge cutoff voltage was 4.3 V, and the discharge cutoff voltage was 2.75 V. All cells were filled and formed under the same conditions to reduce differences in SEI formation and electrolyte wetting.
| Parameter | Graphite Reference | Silicon-Containing Cells |
|---|---|---|
| Positive electrode | NCM811 | NCM811 |
| Negative electrode | Graphite | SiOx/graphite composite |
| SiOx mass fraction | 0% | 5%, 10%, 15% |
| Nominal voltage | 3.7 V | 3.55 V |
| Nominal capacity | 1 A·h | 1 A·h |
| Charge cutoff | 4.3 V | 4.3 V |
| Discharge cutoff | 2.75 V | 2.75 V |
| Cell format | Soft pack | Soft pack |
For ARC testing, I used an EV-ARC instrument with a temperature accuracy of ±0.1 °C and K-type thermocouples with a response time below 1 s. The ARC mode was heating-waiting-searching, with a temperature step equivalent to a heating rate of 0.1 °C/min, a sensitivity of 0.02 °C/min, and a nitrogen atmosphere at 50 mL/min. The waiting time was 15 min per step. This mode is slow enough to detect exothermic reactions before they become self-sustaining, which is essential for observing SEI decomposition in an electric vehicle battery pack material system. For lateral heating, I used a ceramic heater with dimensions of 3.5 cm × 3.5 cm and an adjustable power of 0–48 W. The power ramp was 5 W/min. The heater was attached to the cell surface to create a localized hot spot, and I recorded the time from heating onset to runaway, the maximum temperature, and the flame duration.
| Measurement | Instrument or Sensor | Accuracy or Condition |
|---|---|---|
| Adiabatic calorimetry | EV-ARC | ±0.1 °C, sensitivity 0.02 °C/min |
| Temperature | K-type thermocouple | Class I, ±1.5 °C, response < 1 s |
| Atmosphere | N2 purge | 50 mL/min |
| Heating mode | Heating-waiting-searching | 0.1 °C/min step, 15 min wait |
| Lateral heater | Ceramic plate | 3.5 cm × 3.5 cm, 0–48 W |
| Heater ramp | Power controller | 5 W/min |
| Voltage | Data acquisition | Continuous recording |
| Flame observation | Camera | High-speed video |
I defined three characteristic temperatures to compare the cells. T1 is the self-heating onset temperature, where the cell first produces measurable heat. T2 is the runaway trigger temperature, where the temperature rise becomes exponential. T3 is the maximum temperature reached during runaway. These definitions are useful for an electric vehicle battery pack because they separate the early SEI-dominated stage from the later cell-level and combustion-dominated stage. I used the following criteria:
$$T_1 = \min\{T : \dot{T} \ge 0.02\,^\circ\mathrm{C\,min^{-1}}\}$$
$$T_2 = \min\{T : \dot{T} \ge 10\,^\circ\mathrm{C\,min^{-1}}\}$$
$$T_3 = \max_t T(t)$$
I estimated heat release from the temperature rise and the effective heat capacity of the cell. For a first-order energy balance, I used:
$$Q = \int_{T_1}^{T_3} M C_p(T)\,dT \approx M C_p (T_3 – T_1)$$
$$q_m = \frac{Q}{M} = C_p (T_3 – T_1)$$
Here Q is the total heat release, M is cell mass, Cp is effective specific heat, and qm is specific heat release per unit mass. The cell mass was approximately 16.2 g for silicon-containing cells and 16.4 g for the graphite reference. I also used the self-heating rate as a diagnostic:
$$SR(T) = \frac{dT}{dt}$$
For voltage, I used a simple internal-resistance description to relate voltage collapse to internal short circuits and polarization:
$$V(t) = V_{OCV}(t) – I(t) R_{int}(t)$$
When the cell enters runaway, Rint(t) falls and local short circuits appear, so V(t) drops rapidly. I recorded the time of first micro-short, the time of full short, and the final voltage. These voltage signatures are important for fault detection in an electric vehicle battery pack because they can precede visible smoke or flame.
SEI Stability and Silicon Content
The SEI is a thin passivation film that forms on the negative electrode and controls the first thermal decomposition step. In a graphite anode, the SEI is relatively stable, although it still decomposes at elevated temperature and can react with the electrolyte. In a silicon-based anode, the situation is more severe. Silicon particles expand greatly when they alloy with lithium. The volume change can be approximated as:
$$\epsilon_v = \frac{V_{Li_xSi} – V_{Si}}{V_{Si}} \times 100\% \approx 300\%$$
This expansion imposes mechanical stress on the SEI. A simple stress relation is:
$$\sigma_{SEI} \propto E_{SEI} \frac{\Delta V_{Si}}{V_{Si}}$$
Repeated expansion and contraction cause the SEI to crack, spall, and reform. Each reformation consumes lithium and electrolyte, and the new SEI contains more organic and less stable components. As a result, the effective SEI stability decreases as SiOx content increases. I represent this qualitatively as:
$$R_{SEI}(w) = R_{SEI,0} – \beta w$$
Here w is the SiOx mass fraction, R_SEI is an SEI stability metric, and β is a positive coefficient. This equation expresses the experimental trend: more silicon means lower SEI stability and a lower T1. The early decomposition reactions can be summarized as:
$$\mathrm{SEI} \rightarrow \mathrm{Li_2CO_3 + C_2H_4 + CO_2 + \cdots}$$
$$Li_xC_6 + EC \rightarrow \mathrm{SEI} + C_2H_4 + \cdots$$
For silicon, the exposed surface also reacts with electrolyte after SEI fracture:
$$Li_xSi + EC \rightarrow \mathrm{SEI} + C_2H_4 + \cdots$$
The positive electrode contributes oxygen at high temperature. For NCM811, oxygen release can be written schematically as:
$$Li_{1-x}Ni_{0.8}Co_{0.1}Mn_{0.1}O_2 \rightarrow Li_{1-x-\delta}Ni_{0.8}Co_{0.1}Mn_{0.1}O_2 + \frac{\delta}{2}O_2$$
The released oxygen can react with hydrogen, methane, ethylene, and other gases generated at the negative electrode. These cross-talk reactions add heat and accelerate the runaway:
$$H_2 + \frac{1}{2}O_2 \rightarrow H_2O + \Delta H$$
$$CH_4 + 2O_2 \rightarrow CO_2 + 2H_2O + \Delta H$$
$$C_2H_4 + 3O_2 \rightarrow 2CO_2 + 2H_2O + \Delta H$$
The total heat released during runaway is the sum of several contributions:
$$H_{total} = Q_{SEI} + Q_{anode} + Q_{cathode} + Q_{electrolyte} + Q_{combustion}$$
The temperature rise is then:
$$\frac{dT}{dt} = \frac{1}{M C_p} \sum_i Q_i$$
This energy balance explains why a small increase in silicon content can produce a disproportionately large increase in T3. The silicon expands, the SEI breaks, the electrolyte reacts, gases form, oxygen is released from the positive electrode, and combustion adds a final heat pulse. In an electric vehicle battery pack, this sequence can heat neighboring cells and trigger propagation.
ARC Results: Characteristic Temperatures
In my ARC tests, I observed that T1 decreased as SiOx content increased. The graphite reference had the highest T1, near 158.9 °C. The 5% SiOx cell had a lower T1, near 127.6 °C. The 10% and 15% cells showed even lower SEI stability thresholds, with T1 near 94 °C in the low-temperature limit of the observed trend. This means that the SEI begins to decompose earlier when more silicon is present. The practical implication for an electric vehicle battery pack is that the early warning window shrinks as silicon content rises.
| SiOx Mass Fraction | T1 Trend | T2 Trend | T3 Trend | Key Observation |
|---|---|---|---|---|
| 0% | Highest T1, about 158.9 °C | Reference trigger | Reference T3 | Graphite SEI is most stable |
| 5% | Lower T1, about 127.6 °C | Trigger at about 173 °C | Higher T3 | Early SEI decomposition appears |
| 10% | Lower T1, about 94 °C | Trigger near 201–203 °C | Higher T3 | Runaway becomes more violent |
| 15% | Lowest T1 in the series | Trigger near 246 °C | 836.2 °C under ARC | T3 is 30.3% higher than reference |
As SiOx increased, T2 increased. This may appear counterintuitive at first, but it is consistent with the SEI-breakdown sequence. A less stable SEI decomposes earlier and produces a first exothermic event at lower temperature. However, the full runaway trigger requires the accumulation of enough heat, gas, and cross-talk reactions. In the high-silicon cells, the early reactions are more distributed, and the critical exponential rise is reached at a higher temperature. Once it is reached, however, the stored chemical energy and the gas combustion are greater. Therefore, T2 rises while T3 rises even more strongly.
The maximum temperature T3 increased with SiOx content. For the 15% SiOx cell, T3 reached 836.2 °C under ARC, which is approximately 30.3% higher than the graphite reference. This temperature is far above the melting point of common separator materials and above the thermal stability limit of many cell components. In an electric vehicle battery pack, such a temperature can easily heat adjacent cells, open safety vents, and ignite vent gas. The combination of lower T1 and higher T3 means that silicon content narrows the safety margin at both ends: the cell starts to self-heat earlier, and it ends at a more destructive temperature.
Heat Release and Unit Energy
I calculated heat release from the temperature rise using the energy balance described above. The reported heat release values increased with SiOx content. For the 15% SiOx cell, the unit heat release was 42.3% higher than that of the graphite reference. This increase is significant because it means that each unit of cell mass releases more heat during runaway. In an electric vehicle battery pack, higher specific heat release increases the probability of cell-to-cell propagation and makes thermal barriers more difficult to design.
| SiOx Mass Fraction | Relative Unit Heat Release | Increase Versus Reference | Thermal Hazard Trend |
|---|---|---|---|
| 0% | 1.000 | Reference | Baseline |
| 5% | 1.102 | +10.2% | Moderate increase |
| 10% | 1.253 | +25.3% | Significant increase |
| 15% | 1.423 | +42.3% | Highest measured hazard |
The heat release trend can be represented as:
$$q_m(w) = q_{m,0}(1 + \alpha w)$$
where w is the SiOx mass fraction and α is an empirical coefficient. For the measured range, α is approximately 0.028 per percent SiOx when the 15% cell is compared with the graphite reference. This linear approximation is useful for first-order estimation, but the true behavior may be nonlinear because SEI fracture, gas generation, and oxygen release interact. For a high-confidence electric vehicle battery pack design, nonlinear multiphysics models should be used after the basic silicon-content trend is known.
Voltage Behavior in ARC
Voltage is a sensitive indicator of internal short circuits and polarization. In my ARC tests, I recorded the time of first micro-short and the time of full short. The graphite reference showed a late first micro-short at approximately 400 167 s, followed by full short at approximately 774 130 s, with the voltage falling to about 0.86 V. The 5% SiOx cell showed a much earlier first micro-short at approximately 39 874 s, followed by full short at approximately 394 577 s, with the voltage falling to about 0.34 V. The 10% SiOx cell showed a first short near 295 883 s, a voltage drop to about 1.81 V near 377 975 s, a temporary recovery, and then full short near 553 381 s. The 15% SiOx cell showed intermittent short events beginning near 561 s and another event near 652 549 s. These results show that higher silicon content tends to shift short-circuit events earlier and makes the voltage trajectory more fragmented.
| SiOx Mass Fraction | First Micro-Short Time | Full Short Time | Voltage at Full Short | Voltage Signature |
|---|---|---|---|---|
| 0% | 400 167 s | 774 130 s | 0.86 V | Late, relatively smooth collapse |
| 5% | 39 874 s | 394 577 s | 0.34 V | Earlier collapse |
| 10% | 295 883 s | 553 381 s | 0.01–0.03 V | Drop, recovery, then full short |
| 15% | 561 s | 652 549 s | 0.08–0.34 V | Intermittent short events |
The intermittent behavior in the 15% SiOx cell is important for an electric vehicle battery pack. A voltage recovery after a short does not mean the fault is gone. It can mean that the internal short path has temporarily changed due to local melting, gas formation, or electrode movement. The cell may appear to recover, then fail again with greater severity. Therefore, a battery management system in an electric vehicle battery pack should treat repeated voltage anomalies as a high-risk precursor, not as a transient noise event.
Lateral Heating Results
Lateral heating creates a localized hot spot that can trigger runaway in a manner similar to an external fire, a nearby failing cell, or a mechanical damage site in an electric vehicle battery pack. I placed a ceramic heater on the cell surface and raised the power gradually. I then measured the maximum temperature, the time from heating onset to runaway, and the flame duration. The graphite reference reached a maximum temperature near 832.4 °C, with runaway starting near 533.2 s and a flame duration near 24 s. The 5% SiOx cell reached a maximum temperature near 827.1 °C, with runaway starting near 421.4 s and a flame duration near 20 s. The 10% SiOx cell reached a maximum temperature near 956.8 °C, with runaway starting near 301.9 s and a flame duration near 15 s. The 15% SiOx cell reached the highest maximum temperature, 1 003.5 °C, with a flame duration near 10 s. The flame was more intense and ejected more forcefully, but it lasted for a shorter time.
| SiOx Mass Fraction | Maximum Temperature T3 | Time to Runaway | Flame Duration | Flame Behavior |
|---|---|---|---|---|
| 0% | 832.4 °C | 533.2 s | 24 s | White smoke, then sustained flame |
| 5% | 827.1 °C | 421.4 s | 20 s | Smoke from tabs, then jet flame |
| 10% | 956.8 °C | 301.9 s | 15 s | More violent jet, shorter duration |
| 15% | 1 003.5 °C | 805.4 s to peak | 10 s | Intense jetting, black and silver particles |
The lateral heating results confirm the ARC trend. Higher SiOx content produces a higher maximum temperature and a more violent flame jet. The shorter flame duration does not mean lower hazard. On the contrary, a shorter and more intense jet means that the energy is released faster. In an electric vehicle battery pack, a fast energy release can exceed the venting capacity of the enclosure, cause pressure spikes, and project hot particles onto neighboring cells. The black particles are burned active material, and the silver particles are fragments of the positive current collector. These particles can bridge electrical gaps and create new short circuits.
I observed the following sequence in the high-silicon cells. First, the voltage dropped as the heater created a local hot spot. Second, white smoke appeared from the tabs or the seal region. Third, the smoke transitioned into a jet flame. Fourth, the flame intensity peaked and then decayed rapidly. For the 15% SiOx cell, the runaway flame duration was approximately 10 s, but the peak temperature was above 1 000 °C. This temperature is high enough to melt aluminum and to damage ceramic thermal barriers. Therefore, an electric vehicle battery pack with high-silicon cells needs stronger thermal propagation protection than a graphite-only pack.
Quantitative Relation: Silicon Content, SEI Stability, and Thermal Runaway
I propose a quantitative relation that links silicon content, SEI stability, and thermal runaway. The relation is not a single universal equation, but a set of coupled trends that can be fitted to experimental data. The first trend is that T1 decreases with SiOx content:
$$T_1(w) = T_{1,0} – k_1 w$$
The second trend is that T2 increases with SiOx content:
$$T_2(w) = T_{2,0} + k_2 w$$
The third trend is that T3 increases with SiOx content:
$$T_3(w) = T_{3,0} + k_3 w$$
The fourth trend is that specific heat release increases with SiOx content:
$$q_m(w) = q_{m,0}(1 + \alpha w)$$
The fifth trend is that the time from trigger to flame termination decreases with SiOx content:
$$t_{TR}(w) = t_{TR,0} – \gamma w$$
I can combine these into a thermal runaway index for an electric vehicle battery pack:
$$TRI(w) = \frac{T_3(w) – T_1(w)}{T_2(w) – T_1(w)} \cdot q_m(w) \cdot \frac{1}{t_{TR}(w)}$$
This index increases with SiOx content because the numerator rises, the heat release rises, and the duration falls. It is a simple way to compare chemistries during early design. A higher TRI means a more severe runaway for a given cell mass and a higher risk of propagation in an electric vehicle battery pack.
| Relation | Expression | Physical Meaning | Observed Trend |
|---|---|---|---|
| SEI stability | $$R_{SEI}(w) = R_{SEI,0} – \beta w$$ | More SiOx lowers SEI stability | T1 decreases |
| Trigger temperature | $$T_2(w) = T_{2,0} + k_2 w$$ | Runaway trigger shifts upward | T2 increases |
| Maximum temperature | $$T_3(w) = T_{3,0} + k_3 w$$ | Peak runaway becomes hotter | T3 increases |
| Specific heat release | $$q_m(w) = q_{m,0}(1 + \alpha w)$$ | More heat per unit mass | +42.3% at 15% SiOx |
| Flame duration | $$t_{TR}(w) = t_{TR,0} – \gamma w$$ | Energy release becomes faster | Shorter, more intense jet |
I also define a propagation factor for an electric vehicle battery pack:
$$\lambda_{prop} = \frac{d_{cell}}{t_{TR}}$$
where d_cell is the characteristic cell spacing and t_TR is the runaway duration. If t_TR becomes very short, λ_prop increases, meaning that heat can cross the gap to the next cell before the first cell has finished venting. High-silicon cells therefore require larger spacing, better thermal insulation, or active cooling to keep λ_prop below a safe limit.
Gas Generation and Flame Jetting
The flame jet is driven by gas generation and venting. During SEI decomposition and electrolyte reduction, gases such as H2, CH4, C2H4, CO, and CO2 are produced. The pressure inside the cell rises, the safety vent opens, and the gas mixture can ignite if it mixes with oxygen and reaches an ignition source. The jet power can be approximated as:
$$P_{vent} = \frac{dm}{dt} \Delta H_{comb}$$
The flame length is related to jet velocity and duration:
$$L_{flame} = u_{jet} t_{flame}$$
For the 15% SiOx cell, the jet was more intense and the duration was shorter. This suggests that dm/dt was higher and t_flame was lower. The total heat release was still higher, so the energy was concentrated into a shorter time. In an electric vehicle battery pack, this concentration can damage the vent channel, rupture the enclosure, and ignite nearby materials. The black and silver particles observed in my tests are solid evidence that the jet contains both burned electrode material and metallic collector fragments.
Implications for Electric Vehicle Battery Pack Design
My results have direct implications for electric vehicle battery pack design. First, silicon content should be limited when thermal propagation is a primary safety requirement. In my experiments, 15% SiOx produced the highest T3, the highest unit heat release, and the most violent jet. I therefore recommend a SiOx mass fraction of 10% or below for cells that must meet strict thermal propagation requirements in an electric vehicle battery pack. This threshold balances the capacity benefit of silicon against the measured increase in thermal hazard.
Second, the battery management system should monitor voltage anomalies at high frequency. The intermittent short events in the 15% SiOx cell show that internal short circuits can appear, disappear, and reappear. A simple threshold on voltage may miss these events. A better approach is to monitor voltage derivative, cell-to-cell voltage difference, and temperature rise rate together. This multi-signal approach can detect early SEI decomposition and internal short formation before visible smoke appears.
Third, thermal barriers must be designed for higher peak temperatures. The 15% SiOx cell reached 1 003.5 °C under lateral heating. Many common thermal barrier materials cannot survive this temperature for long. I recommend ceramic fiber, aerogel composites, or intumescent materials that can withstand short-duration heat fluxes above 1 000 °C. The barrier should also resist particle impact from the jet. A simple gap or low-density foam may be penetrated by hot particles.
Fourth, venting must be designed for faster gas release. The high-silicon cells produced a shorter and more intense jet. The vent path should have low flow resistance, a large cross-sectional area, and a direction that does not impinge on neighboring cells. In an electric vehicle battery pack, the vent gas should be routed away from the passenger cabin and away from other modules.
| Design Area | Risk with High SiOx | Recommended Action |
|---|---|---|
| Cell chemistry | Lower T1, higher T3 | Limit SiOx to 10% or below |
| Battery management | Intermittent shorts | Monitor voltage derivative and temperature rate |
| Thermal barrier | Peak temperature above 1 000 °C | Use ceramic or aerogel composite barriers |
| Venting | Fast, particle-rich jet | Use low-resistance, directed vent channels |
| Module spacing | Short runaway duration | Increase spacing or add active cooling |
| Materials | SEI fracture and gas generation | Use nano-Si/C, prelithiation, or solid electrolytes |
Mitigation Strategies at the Material Level
At the material level, several strategies can improve the thermal stability of silicon-based negative electrodes in an electric vehicle battery pack. Nano-silicon/carbon composites can reduce mechanical stress because the carbon matrix buffers volume changes. Prelithiation can reduce the first-cycle lithium loss and stabilize the SEI. Electrolyte additives can promote a more inorganic and thermally stable SEI, for example by increasing LiF content. Solid electrolytes can eliminate flammable liquid electrolyte and reduce gas generation. However, each strategy has trade-offs. Solid electrolytes can have high interfacial resistance, and nano-silicon can have low tap density. The design must balance energy density, cycle life, and thermal safety.
I can represent the effect of a mitigation strategy as a shift in the characteristic temperatures:
$$T_1^{mit} = T_1 + \Delta T_{SEI}$$
$$T_3^{mit} = T_3 – \Delta T_{comb}$$
If a surface coating reduces SEI fracture, ΔT_SEI is positive and T1 increases. If a flame retardant or solid electrolyte reduces combustion, ΔT_comb is positive and T3 decreases. The goal is to increase T1 and decrease T3 simultaneously. In my tests, increasing SiOx did the opposite: it decreased T1 and increased T3. Therefore, any mitigation must reverse both trends.
Limitations and Future Work
My study has several limitations. I used fresh cells and did not include long-term cycling, so the effect of aging on SEI stability is not captured. In a real electric vehicle battery pack, cells age differently depending on temperature, state of charge, and current profile. Aged silicon-based cells may have a thicker and more fractured SEI, which could lower T1 further and increase gas generation. I also used single cells rather than full modules. A full electric vehicle battery pack has busbars, cooling plates, enclosures, and neighboring cells that change heat transfer and propagation. Future work should include aged cells, module-level tests, and full-pack propagation tests.
Future work should also include quantitative gas analysis. I observed flame jetting, but I did not measure the gas composition in real time. Gas chromatography or mass spectrometry could quantify H2, CH4, C2H4, CO, and CO2. This would allow a better link between SEI stability and combustion heat. In addition, multiphysics modeling should couple electrochemical reactions, SEI growth, mechanical stress, gas generation, venting, and combustion. Such a model would help optimize silicon content and thermal protection for an electric vehicle battery pack.
Conclusions
I draw the following conclusions from my experimental study of silicon content and thermal runaway in lithium-ion power batteries for an electric vehicle battery pack.
First, increasing SiOx content reduces SEI thermal stability. The self-heating onset temperature T1 decreased by approximately 40.8% as SiOx increased from 0% to 15%. This means that the SEI begins to decompose earlier, narrowing the early warning window in an electric vehicle battery pack.
Second, increasing SiOx content raises the runaway trigger temperature T2 and the maximum runaway temperature T3. The 15% SiOx cell reached 836.2 °C under ARC, which is 30.3% higher than the graphite reference. Under lateral heating, the 15% SiOx cell reached 1 003.5 °C. This temperature is high enough to damage neighboring cells and challenge common thermal barriers.
Third, increasing SiOx content increases unit heat release. The 15% SiOx cell released 42.3% more heat per unit mass than the graphite reference. This increases the risk of cell-to-cell propagation in an electric vehicle battery pack.
Fourth, flame jetting becomes more intense and shorter as SiOx increases. The 15% SiOx cell showed a flame duration near 10 s, with black and silver particles in the jet. A shorter duration does not mean lower hazard; it means faster energy release and higher peak power.
Fifth, I recommend limiting SiOx mass fraction to 10% or below for cells intended for an electric vehicle battery pack with strict thermal propagation requirements. Higher silicon content may be acceptable only with advanced mitigation, such as nano-Si/C composites, stable SEI additives, solid electrolytes, ceramic thermal barriers, and directed venting.
Sixth, I propose a quantitative relation linking silicon content, SEI stability, and thermal runaway. The relation uses T1, T2, T3, specific heat release, and flame duration as measurable outputs. It can be used as a first-order design tool for comparing silicon-based chemistries in an electric vehicle battery pack.
| Conclusion | Measured or Inferred Result | Design Consequence |
|---|---|---|
| SEI stability falls | T1 decreases by about 40.8% | Earlier self-heating, shorter warning time |
| Runaway becomes hotter | T3 reaches 836.2 °C under ARC | Stronger thermal barriers needed |
| Lateral heating is severe | T3 reaches 1 003.5 °C | Particle-resistant barriers needed |
| Heat release rises | +42.3% at 15% SiOx | Higher propagation risk |
| Flame jet is faster | 10 s duration at 15% SiOx | Low-resistance directed venting needed |
| Safety threshold | SiOx ≤ 10% recommended | Balance energy density and safety |
Overall, my results show that silicon content is a first-order safety variable for an electric vehicle battery pack. It cannot be optimized only for capacity. The SEI must remain stable, the runaway trigger must stay high enough to allow detection, the maximum temperature must stay below the failure limit of the thermal barrier, and the vent gas must be managed without igniting neighboring cells. The quantitative relations and experimental trends I have presented provide a basis for designing safer silicon-based lithium-ion batteries for an electric vehicle battery pack.