23.6. NON-DISPERSIVE LUFT DETECTOR SPECTROSCOPY

23.6Non-dispersive Luft detector spectroscopy

Non-dispersive analysis, while newer in discovery than dispersive analysis (Isaac Newton’s 17thcentury prism), has actually seen far earlier application as continuous process analyzers. The basic design was developed during the years 1937-1938 by Dr. Luft and Dr. Lehrer in the laboratories of the German chemical company I.G. Farbenindustrie. By the end of World War II, over four hundred of these innovative instruments were in service in German chemical plants. Unlike most industrial analyzer technologies which are nothing more than adaptations of laboratory tests previously used by chemists to take manual measurements of substances, the invention of the first non-dispersive process gas analyzer embodied a wholly new analytical technique.

Industrial non-dispersive analyzers typically use either infrared or ultraviolet light sources, because most substances of interest absorb wavelengths in those regions rather than in the visible light spectrum. Non-dispersive spectroscopy using infrared light is usually abbreviated NDIR, while non-dispersive spectroscopy using ultraviolet light is abbreviated NDUV and non-dispersive spectroscopy using visible light is abbreviated NDVIS. Historically, NDIR is the more prevalent of the three technologies. Also, gas analysis is the more common application of non-dispersive spectroscopy in industry, as opposed to liquid analysis, which is why all the examples in this portion of the book assume the analysis of a process gas.

In the case of absorption spectrometers such as non-dispersive analyzers, the challenge is to selectively measure the concentration of certain light-absorbing substances amidst the presence of other substances also absorbing certain wavelengths of light. If the substance of interest is the only substance present in the mixture capable of absorbing light, selectivity is guaranteed. However, most applications in industry are not this easy, with the mixture containing other light-absorbing substances besides the one of interest. Some of these substances may absorb completely di erent wavelengths of light, while others may have absorption bands overlapping the absorption bands of the substance of interest (i.e. the interfering substances absorb some of the same wavelengths of light absorbed by the substance of interest, in addition to absorbing some unique wavelengths of their own).

Dispersive spectrographs achieve selectivity by “disassembling” the spectrum into individual wavelengths and measuring them one by one, but a non-dispersive analyzer must somehow distinguish di erent spectral responses without this “disassembly” of wavelengths. The bulk of this section is devoted to a discussion of exactly how selectivity is accomplished using the NDIR technique.

23.6.1Single-beam analyzer

Non-dispersive analyzers employ the principle of spectrographic absorption to measure how much of a particular substance exists within a sample. NDIR gas analyzers shine light through a windowed sample chamber (typically called a cell ), through which a fresh flow of process gas continually moves. Certain “species” (compounds) of gas within the sample stream absorb part of the incident light, and therefore the light exiting the cell becomes partially depleted of those wavelengths. A heatsensitive detector placed behind the cell measures how much infrared light did not get absorbed by the sample gas. If we imagine the concentration of light-absorbing gas increasing over time, more of the infrared light entering the cell will being absorbed by the gas and converted into heat within the cell, leaving less light exiting the cell to generate heat at the detector. The simplest style of non-dispersive analyzer uses a single light source, shining continuously through a single gas cell, and eventually falling on a small thermopile (converting the received infrared light into heat, and then into a voltage signal):

Single-beam non-dispersive analyzer

Window

This crude analyzer su ers from multiple problems. First, it is non-selective: any light-absorbing gas entering the sample cell reduces heat at the detector (i.e. generates less thermopile voltage), regardless of the species. It might work well enough in an application where the only light-absorbing gas in the process mixture happens to be the one gas we are interested in measuring, but most industrial analyzer applications are not like this. In most cases, our process sample contains multiple species of gases capable of absorbing light within a similar range of wavelengths, but we are only interested in measuring one of them. An example would be the measurement of carbon dioxide (CO2) concentration in the exhaust gas of a combustion furnace: most of the gases exiting the furnace do not absorb infrared light (nitrogen, oxygen), but CO2 gas does. However, carbon monoxide (CO), water vapor (H2O), and sulfur dioxide (SO2) also absorb infrared light, and are all normally present in the exhaust gas of a furnace to varying degrees. Since our crude NDIR analyzer is non-selective, it cannot di erentiate between carbon dioxide and any of the other infrared-absorbing gases present in the exhaust gas.

Another significant problem with this analyzer design is that any variations in the light source’s output cause both a zero shift and a span shift in the instrument’s calibration. Since light sources tend to weaken with age, this flaw necessitates frequent re-calibration of the analyzer.

Finally, since the detector is a thermopile, its output will be a ected not just by the light falling on it, but also by ambient temperature, causing the analyzer’s output to vary in ways completely unrelated to sample gas composition.

Another detector problem still remains, in that an imbalance will develop if one detector happens to “drift” in voltage apart from the other, so they are no longer in perfect counter-balance even with the same received light intensities. This might happen if one of the thermopiles experiences greater ambient temperature than the other, perhaps due to convective heat transfer from hot process sample gas in the nearby sample cell and not the reference cell. An ingenious solution to this problem is to insert a spinning metal “chopper” wheel in the path of both light beams, causing the light beams to pulse through the sample and reference cells at a low frequency (typically a few pulses per second):

Dual-beam non-dispersive analyzer

Parabolic

mirror

Light source

"Chopper"

wheel Window

The e ect of the “chopper” is to make the detector assembly output a pulsating (“AC”) voltage signal rather than a steady voltage signal. The peak-to-peak amplitude of this pulsating signal represents the di erence in light intensity between the two detectors, but any “drift” will manifest itself as a constant or very slowly-changing (“DC”) bias voltage. The following table illustrates the detector assembly signal for three di erent gas concentrations (none, little, and much) both with and without a mis-match in detector signals due to thermal drift:

Absorbing gas concentration