Non-dispersive infrared sensors read a gas by the wavelength it absorbs, and the filter is what defines that wavelength. In breath analysis and medical gas monitoring the margin between a correct reading and a silent error is thin, so the filter's failure modes matter. This guide covers the ways an IR filter drifts, leaks or degrades, and the practical fixes.
IR Filters in Breath Analysis and Medical Gas Sensing
An NDIR gas sensor shines an infrared source through a sample cell and measures how much light survives at the gas's absorption line. A narrowband filter selects that line while a reference channel watches a nearby wavelength that the gas ignores. The filter's centre therefore sets which gas and which line the instrument reports: carbon dioxide near 4.26 micrometres, methane near 3.3, water vapour near 2.7, nitrous oxide near 3.95, and so on. Because the whole reading rests on the filter passing exactly the right band and nothing else, its stability is the instrument's stability.
Coating and deposition considerations
The bandpass is a multilayer thin-film stack tuned to a mid-infrared line, and in this band the coating's temperature coefficient is the first thing to respect. A stable design keeps the centre shift small across the instrument's operating range, and the substrate is chosen for transmission at the target wavelength rather than for visible clarity. A hard, sealed topcoat helps where the filter may see a humid, chemical-laden sample stream. Durability here is not about surviving a drop but about not changing its spectrum when warm, damp or briefly exposed to the gas it measures.
Design rules that reduce cost
Centre wavelengths should follow the standard absorption lines so the coating is a known, repeated design rather than a one-off. Relaxing the bandwidth slightly, where the gas line is broad enough, cuts layer count and yield loss. Combining the measurement and reference functions into a shared substrate where the optics allow avoids two separate filters. Standard substrate sizes and a hard-coat option chosen only where needed keep cost down; over-specifying bandwidth or temperature stability for a bench instrument that never leaves a climate-controlled room is the common waste.
Common failure modes and how they show up
The classic failure is temperature drift: as the instrument warms, the thin-film centre shifts and the reading creeps, often mistaken for a calibration slip. Moisture adsorption on or into the coating shifts and broadens the band, showing up as a slow positive drift in humid samples. Delamination or a cracked layer opens out-of-band leakage that lets a reference channel pick up signal it should reject. Ghost transmission from an un-wedged substrate adds a fixed offset. Each of these presents as a calibration or drift complaint rather than an obvious broken part, which is why they are easy to misdiagnose.
Inspection and measurement
The only reliable check is the actual transmission spectrum on an FTIR or equivalent across the band and a few micrometres either side, taken at the instrument's operating temperature, not just at room temperature. An environmental chamber soak verifies the temperature shift the field will see, and a humidity soak reveals adsorption drift before it reaches a patient. Out-of-band transmission should be measured to the required blocking depth, because a leak there is invisible in a simple peak check. Logging the as-received spectrum against the certificate catches a drift at incoming inspection rather than in the field.
Handling, cleaning and packaging
Hydrocarbon fingerprints absorb strongly in the IR and can sit on the surface as a permanent shift, so gloves and a clean environment are not optional. Solvents must be IR-compatible and used sparingly; aggressive wiping scratches the coating and opens leakage. Desiccant packaging fights moisture adsorption during storage and transit, and the filter should be kept sealed until fitted. Because the failure modes are subtle, the handling chain is part of the specification: a filter handled carelessly can fail the instrument without ever looking damaged.
Requirements specific to Breath Analysis and Medical Gas Sensing
A medical gas sensor is specified by the target gas absorption line it must isolate and the optical density it must hold at every other line in the source's spectrum, because any leak becomes a reading error. The filter is qualified across the instrument's full ambient temperature range, since a breath or ward environment is not climate-controlled, and it is assessed for moisture stability because exhaled and humidified samples are exactly the conditions that shift a marginal coating. Where the sensor is part of a disposable or patient-near device, the filter and its package must be compatible with the cleaning or disinfection regime, and the design keeps a safe margin between the specified drift and the clinical alarm threshold rather than relying on frequent recalibration.
- Centre on the gas line; block all others at required OD
- Qualify the band shift across the full ambient range
- Test moisture stability; exhaled samples are humid
- FTIR spectrum at operating temperature, not just room
- IR-compatible handling; fingerprints shift the band
Framework references: performance is specified per target gas absorption line and required out-of-band optical density; where the device is a medical product, applicable electrical-safety and performance standards apply to the instrument as a whole. Coating stability (temperature, humidity) is assessed against the instrument operating range; confirm the current official text of each applicable standard rather than a single historical value.
Selection data at a glance
| Parameter | Typical value | Notes |
|---|---|---|
| Centre line | CO2 4.26 / CH4 3.3 / H2O 2.7 um | per gas |
| Out-of-band | block to required OD | no leak |
| Temp shift | characterised over range | field drift |
| Moisture | stable when humid | breath sample |
| Handling | gloves, IR-safe solvent | no fingerprint |
Frequently asked questions
Why does the reading drift even though the filter looks fine?
Temperature and moisture shift a thin-film centre without any visible damage. The instrument warms and the band moves a few nanometres, which is enough to change the gas reading. It reads as calibration drift but the cause is the filter's environment.
Can I clean an IR filter with any solvent?
No. Hydrocarbon residues absorb in the IR and can sit on the surface as a permanent shift, and aggressive wiping scratches the coating. Use an IR-compatible solvent sparingly, with gloves, and only when necessary.
Why measure the spectrum at operating temperature?
A room-temperature peak check misses the temperature shift the field will see. If the instrument runs warm, the centre at use is what matters, and only a temperature-swept spectrum reveals it.
Need a quotation for ir filters?
Send your drawing, target specification or coating requirement and an optical engineer will reply within 24 hours with a price and a lead time.
Request a Quote Email an engineer