H2Leak units work by first charging the sealed system with a 5% hydrogen / 95% nitrogen mixture. At that concentration the blend stays below hydrogen’s flammability threshold, so it’s handled and shipped as a non-flammable industrial gas rather than a fuel gas.

Hydrogen’s small molecular diameter — around 2.9 Å — lets it diffuse through micro-leaks that a larger refrigerant molecule would take longer to find its way through, and it dissipates from the surrounding area quickly once a joint is opened for repair. That’s a large part of why it has been gaining ground on helium as a tracer: it’s also far less exposed to helium’s ongoing supply and price volatility.

H2Leak’s DXR-240 is a practical example of the category: it pairs a dedicated H₂/N₂ tracer detection block with a separate refrigerant detection block in the same handheld, so a technician can test with the tracer mix or sniff the system’s own HFC/HFO/HC refrigerant without switching instruments.

Hydrogen sensors are also typically contact (impact) sensors: the sensing element sits right at the probe tip and reacts when tracer gas physically touches it, rather than drawing a sample through a capillary and pump into the instrument body the way a thermal conductivity detector does. Because there’s no internal gas path to clog, a hydrogen sniffer isn’t vulnerable to the capillary blockages — from dust, oil mist, or moisture — that can put a thermal conductivity unit out of service and require it to be sent in for repair.

A thermal conductivity detector — sometimes called a katharometer — holds an electrically heated filament in a temperature-controlled cell. Under normal conditions, heat flows away from that filament to the cell body at a steady, known rate. Gases like helium, hydrogen, and carbon dioxide carry heat away from a surface at a very different rate than air does, so when one of them is present the filament’s temperature — and therefore its electrical resistance — shifts in a measurable way.

This is the same basic detection principle used inside gas chromatographs, scaled down into portable sniffer wands. Most handheld helium tracer-gas sniffers use thermal conductivity as their core sensing method, which makes TCD the de facto standard for helium leak location in the field. It also shows up in CO₂ and hydrogen sniffing — a well-known example is the sensor Honeywell rates to catch hydrogen concentrations as small as 50 ppm with minimal maintenance over a decade of service.

Because it has no ferrous parts that would pose a hazard near a strong magnet, TCD is also the basis for several handheld helium sniffers built specifically for use inside MRI scanner rooms.

This family actually covers two related but distinct sensors, and it’s worth telling them apart.

Heated diode detectors — the standard sniffer on most HVAC/R service trucks — draw sample air across a ceramic element heated to roughly 600–800°C. At that temperature, halogenated refrigerant molecules (HFC, HCFC, HFO, and the H₂/N₂ tracer mix) break apart, releasing chlorine or fluorine ions that generate a current proportional to concentration. Because it reads the refrigerant itself, there’s no need to pre-charge the system with anything — a technician can sniff a live, in-service unit directly.

Catalytic bead (pellistor) sensors work on a different target: a heated catalytic element oxidizes combustible fuel gas — methane, propane — on contact, and the resulting temperature rise is read as a change in resistance. These are built for combustible-gas safety detection, not refrigerant or tracer-gas leak location, and the two shouldn’t be used interchangeably.

Combustible-gas sensors also carry a specific handicap when pressed into service as hydrogen tracer detectors. Their sensitivity is calibrated against a full-scale reading at 100% concentration of the target gas. Hydrogen tracer mixtures are capped at 5% H₂ in nitrogen precisely because that’s the highest concentration that stays non-flammable — there’s no safe way to run the mixture up to 100% to use as a calibration reference. The practical result is that a combustible-gas sensor reading a 5% H₂/N₂ tracer effectively loses about a factor of 20 in usable sensitivity compared to reading the same sensor against a full-scale, 100%-concentration gas.