Every gas detector is a sensing principle wearing a housing. The principle decides which gases the instrument can see, what can blind it, how long it lives and what its readings mean — yet buyers often compare housings and displays while the principle goes unexamined. This article compares the four principles that carry industrial gas detection: catalytic bead, electrochemical, non-dispersive infrared (NDIR) and photoionisation (PID), with the failure modes vendors mention last.
Catalytic bead: the combustible workhorse
A catalytic sensor burns flammable gas on a heated, catalyst-coated bead; the heat changes the bead's resistance, and the bridge circuit reads the change as %LEL. It responds to nearly all combustibles, it is cheap and proven over decades, and it anchors the EX channel of portable platforms.
Its blind spots are structural: it needs oxygen to burn the gas, so it under-reads in inerted atmospheres; and it can be poisoned — silicones, lead, sulfur compounds and halogenated vapours coat or kill the catalyst, sometimes permanently and silently. A catalytic fleet in a silicone-using plant is a fleet on a shorter calibration leash.
Electrochemical: the toxic specialist
Electrochemical cells generate a current by oxidising or reducing the target gas at an electrode; the current is proportional to concentration at ppm levels, with low power draw that suits personal monitors. CO, H2S, SO2, NO, O2 and their family are its home ground — the 0–10 ppm spans with 0.001 ppm resolution in the portable tables are electrochemical work.
Its costs are finitude and cross-sensitivity: cells are consumed by exposure and time (oxygen cells literally eat themselves measuring air), and cousins interfere — hydrogen on a CO channel is the textbook example. Configuration, filters and honest datasheets manage both.
NDIR: the infrared analyst
NDIR passes infrared light through the sample and measures absorption at the gas's characteristic wavelength. Nothing is consumed, nothing burns, poisons do not apply, and life is long — which is why NDIR anchors online analyzers like the MIC300-F08 (CO, CO2, CH4, NH3, H2S, SO2) and fixed points like the MIC100's infrared channels.
Its boundary is physics: only gases that absorb infrared can be measured — hydrogen, oxygen and chlorine are invisible to it — and wet, dirty samples demand conditioning. Within its gas list it is the stability champion; outside it, it is blind.
PID: the volatile scout
A PID floods the sample with ultraviolet light and counts the ions from compounds the light can ionise. It reaches ppm and below for a wide range of volatile organics, starts instantly and carries no fuel — the scout that maps vapour geography in screening work and the second detector in dual instruments like the PTM100.
Its readings are compound-relative (response factors differ per substance), methane is beyond its lamp, and dirty streams foul the lamp. As a quantifier it defers to FID; as a finder, nothing else covers ground as fast.
The professional conclusion is that principles are complements wearing different blind spots. A platform that mixes them per channel — as the portable and fixed ranges here do — lets each gas be measured by the physics that suits it, which is the closest thing gas detection has to a universal rule.
Electrochemical cells: the toxic-gas workhorse
An electrochemical cell is a small fuel cell run in reverse accounting: the target gas reacts at an electrode and the current produced is proportional to the gas concentration. The principle delivers ppm sensitivity for carbon monoxide, hydrogen sulfide, ammonia, chlorine, sulfur and nitrogen oxides and oxygen itself, at milliwatt power, in a package the size of a coin. Its costs are finitude and temperament: the electrolyte and electrodes are consumed by the reaction, so cells carry working lives measured in months to a few years; response drifts with temperature and humidity extremes; and cross-sensitivity is real — a hydrogen sulfide cell that stirs for sulfur dioxide, an ammonia cell that notices amines. None of this disqualifies the principle; it defines its maintenance: bump tests that prove response, calibrations that restore span, replacement dates planned rather than discovered.
Catalytic beads: simple, hungry, poisonable
The catalytic bead burns the gas it measures. A heated catalyst-coated element oxidises combustibles at its surface, and the heat released changes its resistance in proportion to concentration — below the lower explosive limit, where the mixture itself cannot burn. The virtues are maturity, simplicity and a response to virtually every combustible gas. The vices are structural: it needs oxygen to burn (readings fall in inerted atmospheres), it draws real power, and its catalyst can be poisoned — permanently by silicones and lead, temporarily by sulfur and halogens — or inhibited for hours after exposure. Every catalytic instrument's manual lists its poisons; every site using the principle should know which of them walk its own processes, because a poisoned bead does not read low dramatically — it reads low plausibly, which is far more dangerous.
Infrared: light as the reagent
Infrared sensors measure absorption spectra rather than chemistry: hydrocarbons and carbon dioxide absorb characteristic wavelengths, and the attenuation reports concentration without consuming anything. No oxygen is needed, no catalyst can be poisoned, spans run to 100% volume, and service lives stretch because nothing is spent. The principle's blindness is equally clean: it cannot see gases that do not absorb — hydrogen above all — and it costs more, draws more power than electrochemistry and demands optical cleanliness. The mature judgement is combinational: infrared for combustibles in poison-bearing or oxygen-poor service, catalytic for general duty on a budget, and explicit provision for hydrogen wherever it is part of the hazard set.
Photoionisation: the vapour census
A PID's ultraviolet lamp ionises molecules whose ionisation energy falls below the lamp's — 10.6 electron-volts in the common form — and the resulting current counts a whole family of organic vapours at ppm and sub-ppm levels. That breadth is its glory (a single instrument screens solvents, fuels, monomers and refrigerants) and its limit (it reports a total, in isobutylene equivalents, not an identity). Humidity quenches response, lamps foul and need cleaning, and correction factors translate the total toward specific compounds when the target is known. In skilled hands — leak surveys, spill response, tank entry screening — it answers the question 'is something here, and roughly how much' faster than any rival.
Thermal conductivity, paramagnetism and the bulk gases
Percent-level measurement belongs to different physics: thermal-conductivity cells compare a gas's heat conduction against a reference and thrive on binary mixtures — hydrogen in air, methane in nitrogen, carbon dioxide in biogas; paramagnetic and zirconia analysers exploit oxygen's magnetic and ionic peculiarities for precise bulk oxygen. These principles reward stable sampling systems and punish casual ones, because at percent resolution the sampling line is part of the measurement. Choosing them is choosing to engineer the sample path — filtration, pressure, flow, drying — as deliberately as the analyser itself.
Principles in combination: the platform idea
Real instruments are treaties between principles: the four-gas portable weds electrochemistry to catalysis; the wide-range analyser adds infrared, PID and thermal conductivity to cover what the first two cannot. The buyer's skill is reading a proposed configuration as a set of principle choices, each with its physics, interferences and maintenance rhythm — and insisting the seller explain, gas by gas, which principle reports it and what that principle cannot see. A configuration that survives that conversation is worth its price; one that dissolves under it never was.
Semiconductor sensors: the honest account
Semiconductor (metal-oxide) sensors change resistance as gases adsorb onto a heated oxide surface. They are cheap, rugged, long-lived and sensitive — and broadly, cheerfully non-specific, answering a crowd of reducing gases, humidity and temperature with one resistance. That profile earns them a real living in fixed leak alarming for defined gases in defined places — refrigerant leaks in plant rooms, methane sentinels in boiler houses — where the question is binary and the atmosphere is known. It disqualifies them from quantitative multi-gas duty, where their answer cannot be assigned to a compound. The principle's reputation suffers when it is sold past its competence; inside its competence it is excellent value, and specifiers should hold both thoughts at once, because that is what principle literacy means.
Optical and NDIR variants beyond the bench
Optical sensing has been escaping the bench: open-path infrared lines that integrate gas along a beam across a tank farm or a plant boundary, point NDIR sensors shrunk into fixed detectors and better portables, laser (TDLAS) instruments that interrogate one gas's absorption line with a specificity no broadband sensor can match. Each variant trades the same currency differently — path integration buys area coverage and speed at the price of localisation; point optics buy precision at one address; lasers buy selectivity and remote reach at the price of a single-gas view. The principle map is no longer a list but a landscape, and the modern configuration conversation moves across it: which questions need paths, which need points, which need a laser's single-mindedness.
Ultrasonic and the sensing of leaks by ear
One more principle deserves its paragraph because it measures the event rather than the gas: ultrasonic leak detection hears the turbulence of pressurised gas escaping, regardless of the gas's chemistry, the wind, or the direction of the plume. It cannot say what or how much, and in noisy plants it must be tuned to its environment, but as a fixed early-warning layer over pressurised inventories it catches the release at the moment of birth, before any concentration-based sensor can smell the diluted result. Layered schemes that pair ultrasonic ears with concentration noses illustrate the deepest version of this article's thesis: principles are answers to different questions — is it leaking, what is it, how much is here — and mature detection asks all three questions on purpose.
The selection conversation, scripted
A competent selection conversation for any channel runs five questions, in order. What gas, at what concentrations, in whose company? (The companions decide interferences.) What must the reading survive — oxygen scarcity, poisons, humidity, pressure, temperature? (The environment dismisses principles.) What decision does the number drive, and how fast? (The decision sets range, response and alarming.) What maintenance will this owner actually perform? (The honest answer dismisses more principles.) And what record must the reading leave? Five questions, and the principle usually selects itself — the conversation's product is not an instrument but the reasoning, documented, that the next decade's modifications can argue with. Instruments bought without the conversation are replaced with it, eventually, at the usual surcharge.
Principles over a decade: the ownership view
Principles age differently, and ownership reveals it. Electrochemical fleets live on a rhythm of cell replacements that becomes a predictable annuity; catalytic fleets on a slower rhythm punctuated by poisonings that arrive as events; infrared fleets on long quiet plateaus ended by optics and electronics rather than chemistry; PID fleets on lamps and cleaning, the most hands-on annuity of all. A decade's cost model per principle — consumables, bench hours, downtime, spares depth — is a legitimate procurement document, and asking suppliers to help build it separates those who sell instruments from those who support fleets. The cheapest principle to buy is regularly the dearest to own in the wrong atmosphere, and the atmosphere, as this article has insisted from its first page, is the specification's true starting point.
Combinations the field has settled on
Field practice has converged on combinations worth naming as defaults. The personal safety set: electrochemistry for the toxics and oxygen, catalysis or infrared for combustibles, the choice between the last two made on poison census and oxygen behaviour. The process set: NDIR for the infrared-active majors, thermal conductivity for the binary bulk pairs, paramagnetic or zirconia for oxygen precision, electrochemistry demoted to the trace duties it does best. The survey set: PID for breadth, FID for verdicts, TDLAS where distance and selectivity justify a single-gas laser. Defaults are not decisions — the site's atmosphere can overrule any of them — but they are the right place to start the argument, and an evaluation that ends elsewhere should be able to say why in the atmosphere's own terms.
The principle nobody sells: sampling
Every principle in this survey shares one dependency that no sensor catalogue lists: the sample must arrive. Diffusion faces wait for the atmosphere to visit; pumped systems fetch it through lines that filter, condense, adsorb and delay it; extractive benches receive it conditioned or receive a fiction. More field readings are falsified by sampling than by sensing, by a wide and unglamorous margin — the blocked inlet, the flooded trap, the cracked fitting breathing room air, the probe in the wrong stratum. A principles education that ends at the sensor is half an education; the other half is plumbing, placement and patience, and it belongs in every training that this article's readers design.
Principles in the standards ecosystem
Principles do not float free of standards: performance standards certify instruments per principle and per gas, area classifications assume them, and method prescriptions (the LDAR article's FID is the type case) entrench them in law. This ecosystem explains the market's conservatism — a novel principle must win certification, method acceptance and insurer confidence before its physics earns it a living — and it gives buyers a useful heuristic: a principle's installed base in their own industry is evidence about more than fashion. Novelty is not disqualified; it is priced. The buyer of a genuinely new sensing approach is buying a share in its evidence base, and should do so knowingly, with pilot deployments and parallel measurements against the incumbent principle.
The hybrid future arriving quietly
The frontier is combinational rather than revolutionary: instruments that fuse principles and vote between them, sensors that compensate their own interferences from a companion channel's data, networks in which cheap broad sensors cue expensive specific ones, calibration that travels from a reference instrument down through a fleet by comparison. Each development relaxes one of this article's trade-offs without abolishing the underlying physics — the poisons still poison, the lamps still foul, the samples still must arrive. Principle literacy therefore remains the buyer's permanent equipment, because it is the faculty that evaluates each new combination: which question does this answer, what is it blind to, and what will its maintenance ask of us in year four.
A principle audit for an existing fleet
Owners of mature fleets can run this article backwards as an audit. List every channel in service with its principle; beside each, write the atmosphere's poison census, oxygen behaviour and humidity truth; mark the mismatches the years have created — the catalytic point downstream of the new silicone process, the electrochemical cell asked to live in a stream that eats it quarterly, the PID surveying a methane duty it cannot see. Fleets accumulate mismatches the way harbours accumulate silt, through changes nobody connected to the instruments. The audit reconnects them, and its findings are usually a short, affordable list — a point re-principled here, a scrubber there — standing between the fleet and its next instructive failure.
The principles at the extremes
Extremes expose principles honestly. At the cold extreme, electrochemical cells slow and thicken, catalytic beads spend their heat budget on the weather, batteries of every chemistry shrink. At the hot extreme, electrolytes dry, plastics creep, optical benches drift with their own expansion. In the pressure extremes — hyperbaric works, altitude, process pressure — partial pressure rewrites every percentage. Instruments earn their environmental ratings in chambers, but fleets earn their reputations at the extremes their sites actually visit, and the selection conversation should visit them by name: the coldest morning, the hottest roof, the wettest pit the fleet will ever work. Principles do not fail at the average; specify for the tails.
The comparison, pocket edition
Carry the article in one pocket paragraph. Electrochemistry counts toxics at ppm and dies by inches doing it. Catalysis burns what it measures and can be poisoned by the neighbourhood. Infrared sees with light, needs no oxygen, and is blind to hydrogen. The PID censuses vapours it cannot name. Thermal conductivity weighs bulk mixtures; paramagnetism singles out oxygen; semiconductors alarm cheaply and specifically nowhere. And beneath them all, the sample must arrive. A specifier who can recite that paragraph, and knows which clause their atmosphere argues with, is most of the way to a configuration that will still make sense at its first incident review.
Principles and the apprentice's bench
There is a bench in every good training room where the principles sit side by side in the flesh — a cell, a bead, an infrared bench, a lamp — and apprentices pass gas over each in turn and watch the personalities this article has described become needle movements. The exercise teaches what no table can: the bead's hearty surge, the cell's patient climb, the lamp's instant census, the bench's unhurried authority. Programmes that keep such a bench produce technicians who diagnose instruments rather than merely suspect them, because they have seen healthy behaviour with their own eyes and carry the memory into every fault-finding that follows. Principles, in the end, are learned the way atmospheres are measured: by exposure.