As an Amazon Associate, Akermin earns from qualifying purchases. Product links on this page are affiliate links.
Infrared sensing passes light through a sample and measures how much the gas absorbs at wavelengths characteristic of its molecular bonds, so nothing is consumed and the element can live a long time. A heated diode dissociates halogenated molecules at a hot ceramic element and measures the ion current that results, which makes it blind to anything without chlorine or fluorine and gradually erodes the element. A metal oxide film sits hot and changes resistance when almost any reducing gas adsorbs onto it, which makes it cheap, robust and indiscriminate. Selectivity, service life and what fools each one follow directly from those three mechanisms.
Three sensing principles, three different things being counted
It is worth being precise about what each element physically counts, because every behavioural difference between the three follows from it. The infrared element counts photons that failed to arrive. The heated diode counts charge carriers produced by a chemical dissociation. The metal oxide element counts nothing directly at all; it reports a bulk resistance that happens to vary with surface chemistry.
Counting photons is a physical measurement of the molecule itself, so it is inherently selective and inherently non destructive. Counting ions from a dissociation is a chemical measurement, so it is selective for whatever can dissociate that way and inherently consumes a little of the element each time. Reporting a surface resistance is an indirect measurement with no selectivity mechanism at all, which is why the third principle appears in alarms and inexpensive sniffers and almost never in instruments expected to quantify anything.
Those three sentences predict most of the specification differences in the category. The head to head between the first two is laid out in the infrared and heated diode comparison, and the same mechanism argument appears in air quality work in the comparison of infrared and metal oxide carbon dioxide sensors, where the selectivity gap has the same cause.
Infrared absorption responds to the molecule, not to a reaction
Molecules absorb infrared light at wavelengths that correspond to the vibrational modes of their bonds. A sensing bench shines a broadband or filtered source across a measured path, pulls sample air through that path, and compares the energy arriving at the detector with the energy arriving in clean air. The fraction missing is a measure of how many absorbing molecules are in the path. Because the absorption bands differ between molecules, the choice of optical filter determines which gases the bench responds to and how strongly.
Three practical consequences follow. Response is fast once the sample has reached the bench, but the sample has to be pumped there, so plumbing and flow rate matter as much as the optics. Nothing is consumed, so ageing is slow and dominated by the source and by contamination of the optical windows. And zeroing in clean air is both easy and necessary, because the measurement is a ratio against a reference condition rather than an absolute count.
The Fieldpiece DR82 infrared refrigerant leak detector states detection of leaks below 0.03 ounces per year, which is under a gram per year, compatibility with CFC, HCFC, HFC and HFO refrigerants and blends, an IP54 rating with stated resistance to moisture and oil, and a ten year sensor life. That last figure is the one the principle earns: a long life claim is credible for an element that is not consumed by the measurement, and the infrared detector roundup shows how consistently the class makes it.
A heated diode counts halogen ions released at a hot element
A heated diode element holds a ceramic surface at high temperature with a bias across it. When a halogenated molecule reaches that surface it breaks apart, and the chlorine or fluorine ions released increase the current flowing between the electrodes. The current is the signal. Nothing else in ordinary service air produces it in the same way, which gives the principle genuine selectivity for halogenated refrigerants without needing optics.
The trade is that the element is in the reaction. It erodes slowly in normal use, it is poisoned by oil mist and it saturates when exposed to a large leak, after which it needs time in clean air before it can discriminate again. Manufacturers answer the first problem by making the probe or the element replaceable, the second with filters and handling guidance, and the third with an auto zero function and selectable sensitivity levels.
The RALBIL heated diode leak detector states a maximum sensitivity of 0.05 ounces per year, given as 1.5 grams per year in a stationary state, a replaceable probe, a warm up of no more than thirty five seconds with automatic zeroing on power up and manual zeroing on demand, nine leak levels on the display and three sensitivity settings. The phrase in a stationary state is doing real work in that specification: it is the still air condition the figure depends on. The rest of the class is covered in the heated diode roundup.
Metal oxide films change resistance when almost anything adsorbs
A metal oxide element, usually tin dioxide, is held at several hundred degrees by an integral heater. At that temperature oxygen from the air chemisorbs onto the surface and traps electrons, raising the resistance of the film to a stable baseline. When a reducing gas arrives it reacts with that adsorbed oxygen, releases the trapped electrons and the resistance falls. The fall is the signal.
Any reducing gas will do it. Alcohols, solvents, cooking vapours, exhaust fumes, hydrogen and the target gas all push the resistance the same way, and water vapour competes for the same surface sites, which makes the baseline move with humidity and temperature. The principle is therefore excellent at answering is there something here and poor at answering what and how much. It is cheap, mechanically robust, needs no pump and tolerates being left powered for years, which is exactly the profile a domestic alarm wants.
The plug in natural gas and carbon monoxide alarm is a typical product of that reasoning. It states detection of dangerous levels of natural gas, audible and visual alerts, outlet installation with no setup and a ten year sealed battery, with a sensor described only as highly responsive. Its sensing technology is not stated in the listing bullets, and neither is an alarm concentration, which is normal for the category and is also why a device like this belongs on a wall rather than in a hand.
Sensor life, warm up and drift compared
| Behaviour | Infrared absorption | Heated diode | Metal oxide film |
|---|---|---|---|
| What produces the signal | Light absorbed by molecular bonds | Ion current from halogen dissociation | Resistance fall as reducing gas adsorbs |
| Selectivity | Set by the optical filter, good | Halogenated molecules only | Almost none |
| Blind to | Gases outside the filter band | Hydrocarbons, ammonia, carbon dioxide | Little, which is the problem |
| Element consumed by use | No | Yes, gradually | No, but the surface ages |
| Dominant drift mechanism | Optical contamination and source ageing | Erosion and oil poisoning | Humidity and temperature |
| Behaviour after a large exposure | Recovers as the sample clears | Saturates, needs clean air time | Recovers slowly, baseline shifted |
Read down the blind to row before anything else, because it is the only row that can make an instrument useless rather than merely imperfect. An element that cannot respond to the molecule in the system is not a weak detector; it is the wrong detector, and no amount of sensitivity on the specification sheet changes that.
Matching the principle to the gas and the job
Start from the charge rather than the instrument. Halogenated refrigerants suit both infrared and heated diode elements, and the choice between them is a trade between long sensor life with stable behaviour on one side and a replaceable element with fast recovery on the other. Hydrocarbon charges and ammonia need a combustible or electrochemical instrument, since neither halogen principle can see them at all. Mildly flammable blends sit between the two worlds and bring their own requirements, which is the subject of leak detection on R32 and R454B systems.
Then match the threshold to the task. Maximum sensitivity belongs to a final hunt on a system that has been ventilated and left to settle, not to a first pass through a plant room, where it will respond to the background and tell you nothing about position. Most instruments offer selectable levels precisely so that the same element can do both jobs, and using the lowest level that still distinguishes the joint from the air is the habit that finds leaks fastest.
Frequently Asked Questions
Why does an infrared sensing element last longer?
Because nothing is consumed to make the measurement. Light crosses a sample path, the gas absorbs part of it at characteristic wavelengths and a detector reports what arrived. No reagent is used up, no catalyst is poisoned and no element is eroded, so the limits on life are the source lamp, the optics staying clean and the electronics. Elements that work by reaction, in contrast, change slowly every time they are exposed.
Can a heated diode find hydrocarbon refrigerants?
No, and this is the single most important limitation of the principle. The element works by releasing halogen ions from molecules that contain chlorine or fluorine, so propane, isobutane, ammonia and carbon dioxide produce no response at all because there is no halogen to release. Work involving those refrigerants needs a different principle entirely, which is why instruments aimed at flammable hydrocarbon charges use combustible gas sensing instead.
Why do metal oxide detectors need warming up?
The film only works hot. A heater brings the oxide layer to a few hundred degrees so that oxygen chemisorbs onto the surface and sets a stable baseline resistance; reducing gases then react with that adsorbed oxygen and the resistance falls. Until the element reaches temperature and the baseline settles there is nothing meaningful to compare a reading against, which is why these instruments insist on a warm up period and an auto zero in clean air.
What makes a sensor drift?
Different things per principle. An infrared bench drifts mainly through optical contamination and source ageing, both slow and both correctable by zeroing. A heated diode drifts as the element erodes and as oil contaminates it, which shows as a falling response rather than a moving zero. A metal oxide film drifts with humidity and temperature almost in real time, because water vapour competes for the same adsorption sites the target gas needs.
Does a higher sensitivity figure make a sensing principle better?
Only within the same type of work. A very low threshold is an advantage when hunting a slow loss on a sealed system and a liability in a space with background vapour, where it produces a response everywhere. Selectivity often matters more than threshold: an element that responds only to the family you are chasing will find the leak faster than a more sensitive one that also responds to the solvent somebody used nearby.
Which principle should a general service technician carry?
It depends on the charges in the systems being serviced rather than on any ranking of the principles. Halogenated refrigerants are well served by both heated diode and infrared elements, with infrared favoured where sensor life and stability matter and heated diode where a replaceable element and a fast recovery from saturation are preferred. Hydrocarbon and ammonia charges need a combustible or electrochemical instrument, because neither of the two halogen principles can see them.