1. a wide range of materials absorb infrared radiation (due to intramolecular vibrations)
2. for any one material, the strength of absorption (absorbance) varies with wavelength (its absorption spectrum)
3. different materials have different absorption spectra.

The basic principles of operation of infrared gas sensors rely on the exploitation of these facts. Typical infra-red spectra for carbon monoxide, propane, n-hexane and carbon dioxide are shown in figure 1.

Figure 1: Absorption Spectra

There are certain basic components common to all infrared gas sensors: an infrared source (e.g. incandescent lamp), a detector (e.g. thermopiles, pyroelectric detectors), a means to select appropriate wavelengths (e.g. band pass interference filter) and a sample cell. Radiation from the source passes through the sample cell and wavelength selector. The choice of wavelength has a large bearing on the relative selectivity of the sensor. The radiation NOT absorbed by the sample is then detected and the ratio of this to the incident provides a measure of the concentration of target gas in the sample. A second detector (or channel) tuned to a different wavelength that is not attenuated by any species likely to be present in the sample is normally used to provide this reference measurement.

A further component that enhances the performance of IR gas sensors is a temperature sensor. All these components have temperature dependencies which must be compensated to provide an accurate measure of gas concentration. This temperature sensor (typically a thermistor) should be sited within, or in very close proximity to, the detector(s).

Infrared sensors effectively give a measure of the number of target gas molecules in the light path between source and detector. Consequently, the output signal not only varies with concentration but also barometric pressure i.e. they are partial pressure devices. For very high measurement accuracy, compensation for barometric pressure is, therefore, required. This dependency also infers that sensors with longer optical path lengths (i.e. distance traveled by radiation between source and detector(s)) will have increased sensitivity and tend to have a lower dynamic range but increased resolution.

In a single target gas, fixed optical light path device under constant barometric pressure, the signal output (and signal/noise ratio) approximately exponentially decays with increasing concentration i.e. infrared gas sensors are inherently non-linear. The measurement accuracy decreases with increasing concentration.

The components described above form a typical infrared gas sensor. However, some supporting electronics is required in any practical system. The more common detector technologies provide very small analogue signal outputs that require amplification. Basic analogue filtering of the amplified output signal can then enhance measurement accuracy.

The source also requires a driver circuit. It is usual practice to modulate the source output by pulsing (although some older design used fixed illumination and mechanical choppers). This creates periodic variations in the emitted intensity and so allows the use of synchronous detection techniques.

To carry out the temperature and barometric compensations, it is common practice to use computational algorithms inside a microprocessor. This first requires the analogue signals to be converted into digital signals. The compensated data is then transmitted to the user in some form.

An overall block schematic of a typical 2 channel IR gas sensor with separate supporting electronics is shown in Fig 2.

Figure 2: Block schematic of high accuracy 2 channel infrared gas sensor