Short answer: Choose aerogel (usually encapsulated) when the main problem is heat conduction between closely packed prismatic or pouch cells and you need the thinnest barrier. Choose mica where the barrier faces flame, particle jets or high voltage, for example pack lids, busbar shields and vent paths. Choose ceramic-fibre paper for low-cost, high-temperature insulation where some extra thickness is acceptable. For cell-to-cell positions in high-energy packs, composites that combine these materials are now the norm.
Side-by-side
| Property | Silica aerogel pad | Mica sheet | Ceramic-fibre paper |
|---|---|---|---|
| Thermal conductivity, ~25 °C | ≈0.015–0.025 W/(m·K) | ≈0.2–0.6 W/(m·K) | ≈0.035–0.05 W/(m·K) |
| Thermal conductivity, ~400–600 °C | ≈0.035–0.09 | ≈0.2–0.6 | ≈0.06–0.16 |
| Temperature limit | Sinters above ≈650–800 °C unless shielded | Very high; resists burn-through | Very high (fibre rating ≈1,100 °C+) |
| Compression / swelling | Compressible, partial rebound | Rigid | Limited rebound |
| Dielectric strength | Moderate (depends on encapsulation) | Excellent | Moderate when dry |
| Particle and flame erosion | Weak if unfaced | Excellent | Moderate |
| Relative cost per m² | High | Low | Low |
Values are typical ranges. See the materials data page for details and sources.
Five questions that decide the choice
- What is the cell format? Prismatic and pouch cells have large flat faces in direct contact, so conduction dominates and thin, low-conductivity layers such as aerogel pay off. Cylindrical cells touch along lines, so potting compounds, foams and module-level shields often matter more than flat barriers.
- How much does the stack swell over life? If the barrier also has to take up cell swelling, it needs a compressible core (aerogel or foam) and a known compression-set behaviour. A rigid mica sheet will not do this alone.
- Where does the vent jet go? If the hot jet from a venting cell can hit the barrier, it needs a tough facing, such as mica or ceramic paper, or the core will be eroded within seconds.
- What voltage does the barrier have to withstand? In high-voltage packs, check the dielectric withstand of the whole barrier, ideally after heat exposure, not just the room-temperature value.
- Which test do you have to pass? Under China’s GB 38031-2025, the pack must not catch fire or explode after a cell goes into runaway. For stationary storage, UL 9540A data is used to decide installation requirements. The trigger method and observation time in your target test set how much margin you need.
Common mistakes
- Comparing room-temperature conductivity only. What matters is performance at several hundred °C, and the materials rank differently there.
- Ignoring compression. A barrier that is 3 mm thick when fitted may be 2 mm after years of swelling, with a matching loss of insulation.
- Treating the barrier as the whole solution. Busbars, cooling plates and shared vent channels can carry heat and flame around the best barrier. Design the heat and gas paths as a system.
- Skipping ageing before the propagation test. Test the barrier in the state it will be in after ageing, not only when new.
Related
Frequently asked questions
Is aerogel better than mica for thermal runaway protection?
For blocking heat conduction between cells, yes: aerogel's thermal conductivity is roughly ten times lower. For resisting flame, particle jets and high voltage, mica is better. That is why many production barriers laminate an aerogel core with mica facings.
How thick should a cell-to-cell barrier be?
Typical production barriers are 1–3 mm for aerogel or composite pads between prismatic or pouch cells, and 0.2–1 mm for mica sheets. The required thickness depends on cell chemistry, capacity, the trigger method in your target standard and how much compression the barrier sees over life, so it has to be confirmed by a propagation test.
Can I use a phase-change material as a thermal runaway barrier?
Not on its own. PCMs absorb heat only while melting, at a few tens of °C, and organic PCMs are flammable. They help with normal temperature management, not with stopping propagation.