DetectionBatteryChemistry

LFP vs NMC: Tuning Detection Thresholds

By Engineering — Sensing · July 11, 2026 · 7 min read

LFP vents about a third the gas of NMC and burns far cooler, but proportionally richer in hydrogen. A detector tuned only on NMC under-warns on LFP.

LFP and NMC fail differently enough that a detector calibrated on one under-warns on the other. A lithium-iron-phosphate cell in thermal runaway vents roughly a third the gas of a nickel-manganese-cobalt cell and peaks far cooler, but its stream is proportionally richer in hydrogen. Tune the thresholds for LFP's weaker, hydrogen-led signal and you cover NMC — not the reverse.

What the vent-gas numbers actually say

LFP produces less gas, cooler, with a different balance. In one 2025 comparative bench study, an LFP cell released about 0.02 mol of gas per Ah against roughly 0.07 mol/Ah for NMC — some 3.5 times less. The composition inverts as well: the LFP stream ran about 41% hydrogen and 8% carbon monoxide, while NMC ran about 19% hydrogen and 36% carbon monoxide. Peak vent-gas temperature followed the same split, near 446°C for LFP versus roughly 1050°C for NMC. The direction is what matters: less total gas, a hydrogen-dominated fraction, and a far weaker thermal punch.

~0.02 vs 0.07

Vent gas, mol/Ah — LFP vs NMC (~3.5× less for LFP)

41% vs 19%

Hydrogen fraction of vent stream — LFP vs NMC

~446 vs 1050°C

Peak vent-gas temperature — LFP vs NMC

Why LFP starves the thermal channel

A thermal detector keys on the heat an event puts into the air and the surrounding steel, and LFP gives it less to work with. With a peak vent-gas temperature under half NMC's and a lower total heat release, LFP's thermal signature climbs slower and lower above the deck's moving baseline — the sea-state, engine and solar swings a marine detector already has to reject. A fusion rule that leans on the thermal channel, calibrated against NMC's violent thermal rise, quietly de-rates its own sensitivity for exactly the chemistry that produces the weakest heat signal. The event is still detectable; the margin is thinner, and it is thinnest for the cell that signals most quietly.

Why the gas channel shifts to hydrogen

Because LFP's vent is hydrogen-led and carbon-monoxide-poor, CO is a weak primary for it. CO is a strong early co-indicator for NMC — over a third of its vent stream — but collapses to under a tenth for LFP, so a detector that weights CO heavily is reading the wrong tracer on an LFP deck. Hydrogen is the common thread: small, fast-diffusing, effectively absent from a clean deck's background, and the dominant fraction of LFP's stream. But LFP delivers that hydrogen inside a much smaller total volume, so the absolute concentration reaching a sensor is lower. The channel to trust is the same across both chemistries; the trip level for LFP has to sit lower.

Chemistry-agnostic hardware, chemistry-aware thresholds. The same gas-and-thermal layer catches both — but only if its trip levels are set from the chemistry that signals weakest, not the one that signals loudest.

Set thresholds from both chemistries, not one

Derive the trip levels from cell-level abuse data for both chemistries — the same UL 9540A-style vent-gas and temperature curves used to characterise a cell, read across LFP and NMC rather than one. Calibrate only on NMC and two things go wrong: the hydrogen trip sits too high for LFP's lower absolute concentration, and the fusion weighting over-trusts the thermal and CO channels that LFP under-feeds. The fix is not different sensors per deck — it is a single threshold set anchored to the harder-to-detect LFP case, then validated against NMC to confirm it does not raise false alarms. Weakest-signal-first is the design rule, because the alternative fails silently.

The deck is mixed — design for the weakest signal

You cannot tune per vehicle because you rarely know the chemistry per vehicle. When the car carrier Morning Midas was abandoned and later sank in the North Pacific in June 2025, it was carrying 3,048 vehicles — including 70 full EVs and 681 hybrids — out of China, where LFP now dominates domestic production. The loading interface almost never records cell chemistry per unit, so any deck is a blend of LFP, NMC, hybrid packs and ICE vehicles. A detector tuned for the weakest, coolest, hydrogen-led LFP vent covers the louder NMC event by construction; tuned the other way, it leaves an LFP cell venting under a threshold set for a fire that burns twice as hot.

Sources

1. Batteries (MDPI), 2025, 11(9):323 — comparative thermal-runaway vent-gas study across NMC, LFP and sodium-ion cells: per-Ah gas volume (~0.02 vs ~0.07 mol/Ah), composition (LFP ~41% H₂ / 8% CO; NMC ~19% H₂ / 36% CO) and peak vent-gas temperature (~446°C LFP vs ~1050°C NMC).
2. Batteries (MDPI), 2025, 11(10):352 — ignitability of real LFP vent gas below the theoretical lower explosive limit: LFP's cooler vent still forms a flammable-gas hazard in an enclosed space.
3. Frontiers in Physics, 2025 — review of gas-detection technology for lithium-ion thermal runaway: gas signals give earlier warning than surface-temperature sensing in the initial stage.
4. Peer-reviewed thermal-runaway reviews on LFP vs NMC/NCM cathodes — onset-temperature ranges (~220–300°C LFP vs ~150–210°C NMC) and the olivine-vs-layered-oxide oxygen-release difference.
5. The Maritime Executive / WorkBoat, June 2025 — Morning Midas (Zodiac Maritime) car-carrier fire and abandonment: 3,048 vehicles incl. 70 EV and 681 hybrid, en route from China.
[VERIFY: exact per-Ah gas volumes, H₂/CO fractions and peak vent-gas temperatures — the MDPI Batteries 11(9):323 PDF was access-blocked to the bot; confirm the precise figures against the open version before publication.]

Frequently asked

Questions, answered

Do LFP batteries need different fire detection than NMC?+

Not different hardware — different thresholds. A gas-and-thermal detection layer catches both chemistries, but LFP vents less gas, far cooler, and proportionally richer in hydrogen. A detector calibrated only on NMC's hotter, CO-heavy vent sets its trip levels too high for LFP's weaker signal. The right approach is chemistry-agnostic sensors with thresholds anchored to the harder-to-detect LFP case.

Which gas should a detector watch for on an LFP deck?+

Hydrogen. LFP's vent stream is roughly 41% hydrogen and under 10% carbon monoxide, the inverse of NMC, so CO — a strong early tracer for NMC — is a poor primary for LFP. Hydrogen is common to both, fast-diffusing and near-absent from a clean deck's background. Because LFP produces less total gas, the hydrogen trip level has to sit lower to catch it early.

Why is an LFP cell harder to detect thermally?+

It runs cooler and releases less heat. Peak vent-gas temperature is near 446°C for LFP against roughly 1050°C for NMC, and total heat release is lower, so LFP's thermal signature climbs slower and less far above a marine detector's moving baseline of sea-state, engine and solar swings. A fusion rule that leans on the thermal channel loses margin on exactly the chemistry that heats the least.

Can one detector cover a mixed LFP and NMC deck?+

Yes, if it is tuned to the weakest signal. Car carriers rarely record cell chemistry per vehicle, so any deck is a blend of LFP, NMC, hybrid and ICE units. Thresholds anchored to the coolest, lowest-volume, hydrogen-led LFP vent cover the louder NMC event by construction. Tuned the other way, the detector can leave an LFP cell venting under a threshold set for a hotter fire.