Hydrogen Sensor for Batteries

Four sensing principles dominate. Catalytic pellistors burn the analyte on a heated bead and are robust but power-hungry and poisoning-prone. Electrochemical cells are sensitive but have limited life and temperature range. Thermal conductivity sensors work at high concentration only. Semiconducting metal-oxide sensors are sensitive but typically require heating to several hundred degrees and drift with humidity.

The Battery Sense core is a chemiresistive nanomaterial device operating without a heater. Resistance changes as hydrogen adsorbs on the engineered sensing layer, and on-die electronics convert that into a calibrated concentration. Removing the heater removes the dominant power draw, the dominant failure mechanism and the ignition-source objection to placing an active element inside a battery enclosure.

For battery early warning, three numbers determine whether a device is fit for purpose: limit of detection, resolution and range. The BD-100 is specified at 10 ppm LOD for hydrogen with 5 ppm resolution over 0–50,000 ppm, better than 3% of reading accuracy.

A 10 ppm LOD is four thousand times below the hydrogen lower flammability limit. That is the margin that converts a detector into a diagnostic instrument: it can see the trace evolution that precedes a vent, not just the accumulation that precedes an explosion.

A sensitive sensor that alarms on cleaning solvent is worse than no sensor, because operators learn to ignore it. Battery environments contain methane, carbon monoxide, hydrocarbon vapours, cleaning agents and, in mixed rooms, lead-acid charging gas. High selectivity against these interferents is a hard requirement.

Battery Sense addresses this in two ways. First, the sensing layer is engineered for selectivity against common interferents including CH₄ and CO. Second, the platform measures multiple channels simultaneously, so the analytics can require a hydrogen-plus-electrolyte-VOC signature before escalating — a pattern that a solvent event or a lead-acid charger does not produce.

For hostile environments the membrane itself can be tailored. In the ENGIE co-development, a custom anti-poison membrane was formulated for that facility's specific gas environment, blocking interferents while preserving full sensitivity to the target analyte.

Detectors with poor low-end sensitivity frequently compensate with rate-of-rise logic: alarm when concentration climbs faster than a threshold. This works for a fast, energetic leak and fails for a slow one.

That failure mode is not hypothetical in battery applications. LFP packs can generate gas gradually over long periods; a rate-triggered device may never fire, and a percent-LEL device will not register the concentration at all. Measuring true concentration in real time removes the dependence on how fast the fault develops.

Hydrogen is buoyant and diffuses quickly, which is often used to argue for ceiling-mounted detection. In an actively ventilated battery room, that same property means the gas is transported and diluted away from the source before it reaches the ceiling — the ventilation designed to keep the room safe is also destroying your early-warning signal.

Detecting inside the enclosure sidesteps the problem. Ventilation continues to do its job for life safety while the sensor sees undiluted concentration at the source and reports which unit is venting.